<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Manish Maheshwari on Medium]]></title>
        <description><![CDATA[Stories by Manish Maheshwari on Medium]]></description>
        <link>https://medium.com/@manishlens?source=rss-6ddd323af000------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*reyHk3gIVTolQvMKposcoQ.jpeg</url>
            <title>Stories by Manish Maheshwari on Medium</title>
            <link>https://medium.com/@manishlens?source=rss-6ddd323af000------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 01:23:14 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@manishlens/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[The New AI Moat Is Gross Margin]]></title>
            <link>https://medium.com/@manishlens/the-new-ai-moat-is-gross-margin-775d7294a765?source=rss-6ddd323af000------2</link>
            <guid isPermaLink="false">https://medium.com/p/775d7294a765</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[agentic-ai]]></category>
            <dc:creator><![CDATA[Manish Maheshwari]]></dc:creator>
            <pubDate>Thu, 03 Sep 2026 17:49:18 GMT</pubDate>
            <atom:updated>2026-09-03T18:09:32.292Z</atom:updated>
            <content:encoded><![CDATA[<h4>SaaS made marginal cost disappear. AI brought it back. Every customer interaction now has a meter running. And gross-margin architecture is becoming product architecture.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qcV1ks4yyo4gOgmyNK8hWA.png" /><figcaption>AI Generated Image on AI Economics</figcaption></figure><p>For the first phase of the AI boom, everyone asked the same question: <strong>Who has the smartest model?</strong></p><p>That may soon be the wrong question.</p><p>The more important question for the next phase is: <strong>Who can deliver intelligence most economically?</strong></p><p>This is not a semantic distinction. It is a fundamental change in the business model of software.</p><p>For decades, software enjoyed one of the best economic structures ever invented. You built the product once, and the marginal cost of serving the next customer was close to zero. That created the extraordinary gross margins of SaaS.</p><p>AI changes that bargain.</p><p>Every prompt has a cost. Every agent loop has a cost. Every retry, verification step, tool call, long context window and hidden reasoning cycle consumes compute. As AI becomes more capable, it is also becoming more agentic … and therefore more computationally hungry.</p><p>The paradox is that intelligence is getting cheaper <strong>per token</strong> while becoming more expensive <strong>in aggregate</strong>.</p><p>That is the economic problem every AI founder, CFO and CIO now needs to understand.</p><h3>From training economics to inference economics</h3><p>The first era of generative AI was defined by training.</p><p>Billions of dollars went into GPUs, data centres and massive training runs. The strategic question was whether a company could marshal enough compute to build the next frontier model.</p><p>But training is largely a fixed investment. Inference is a variable cost.</p><p>A model might be trained periodically. Once deployed, however, it may be called billions of times. McKinsey projects that by 2030 inference will overtake training as the dominant AI workload, accounting for more than half of AI compute.</p><p>For most companies building <em>on top</em> of foundation models, that future has effectively arrived already. They are not paying to train GPT, Claude, Gemini or another frontier model. Their cost of goods sold is what happens every time their customers use those models.</p><p>And agentic AI magnifies this dramatically.</p><p>A recent study of eight frontier models performing software-engineering tasks found that identical tasks could vary by as much as <strong>30 times in token consumption</strong> across runs. More tokens did not reliably produce better outcomes; accuracy frequently plateaued at intermediate levels of spend.</p><p>Another 2026 study of multi-agent software development found that <a href="https://arxiv.org/abs/2601.14470"><strong>59.4% of tokens were consumed during iterative code review</strong></a>, not initial code generation.</p><p>That is a profound insight.</p><p>The expensive part of AI is increasingly not producing the first answer. It is <strong>reasoning, checking, refining, coordinating and trying again</strong>.</p><blockquote><strong>Inference is becoming the cost of goods sold for intelligence.</strong></blockquote><h3>The SaaS pricing model will not survive unchanged</h3><p>This creates an uncomfortable problem on the revenue side.</p><p>Traditional SaaS largely charges for access: per user, per month.</p><p>But suppose an AI agent does work that previously required five employees. Should the company providing the agent earn less because only one human now logs into the software?</p><p>Clearly not.</p><p>At the other extreme, charging customers purely by tokens is economically neat but strategically dangerous. It aligns revenue with cost, yet reduces the product to a markup on somebody else’s compute.</p><p>Customers do not actually want tokens.</p><p>A bank wants fraud prevented. A retailer wants customer queries resolved. A developer wants functioning code. A logistics company wants shipments routed. A finance team wants invoices reconciled.</p><p><strong>The closer the unit of pricing gets to the unit of customer value, the stronger the business model becomes.</strong></p><p>There are three broad models emerging.</p><p>The first is <strong>seat pricing</strong>, which remains simple but increasingly misaligns price with value.</p><p>The second is <strong>consumption pricing</strong>, which aligns revenue with usage but can create volatile bills and encourages customers to compare vendors on raw compute cost.</p><p>The third is <strong>outcome pricing</strong>: charging per resolved ticket, completed workflow, qualified lead, generated application, reconciled invoice or some other measurable result.</p><p>Outcome pricing is intellectually attractive, but it is not universally practical. Attribution can be messy. Outcomes may occur months later. Customers and vendors may disagree about causality. And the AI company takes on performance risk.</p><p>So I suspect the dominant architecture will be <strong>hybrid</strong>: a recurring platform fee for access, integrations and workflow value; included consumption with transparent overages; and an outcome-linked component where attribution is sufficiently clear.</p><p>This does something important: it separates what the customer buys from what the AI company must optimize internally.</p><p>Customers buy <strong>outcomes</strong>.</p><p>The company manages <strong>tokens</strong>.</p><h3>The real unit economics: cost per successful outcome</h3><p>This leads to the metric I believe will matter far more than cost per million tokens:</p><h3>Revenue per accepted outcome − fully loaded cost per accepted outcome = AI gross margin</h3><p>The denominator is where many AI businesses are currently fooling themselves.</p><p>It must include not simply visible output tokens, but input context, cached and uncached tokens, reasoning, retries, failed runs, tool calls, retrieval, orchestration, human verification and infrastructure overhead.</p><p>McKinsey’s May 2026 enterprise AI FinOps survey gives a sense of how quickly this problem is emerging. As companies move from experimentation toward broader deployment, AI spending rises almost fourfold; <strong>93% of respondents said they were exceeding their AI budgets</strong>. McKinsey estimates that disciplined AI consumption management can save roughly <strong>20–30%</strong> of AI costs.</p><p>The implication is simple: token optimization can no longer remain an engineering task buried inside the infrastructure team.</p><p>It needs to become a management discipline.</p><p>Think of it as <strong>AI FinOps</strong>.</p><h3>The cheapest token is the one you never consume</h3><p>The first generation of AI applications often followed a simple architecture: send the problem to the best available model and hope the economics improve as model prices decline.</p><p>That is unlikely to be enough.</p><p>The winning architecture will increasingly resemble a portfolio manager allocating scarce capital.</p><p>A simple classification problem should not automatically go to the most capable frontier model. A deterministic calculation may not need an LLM at all. Stable prompt context should be cached rather than repeatedly processed. Long conversation histories should be compressed. Agents should receive only the relevant pieces of tool output. Expensive reasoning should be reserved for problems where it materially changes the probability of success.</p><p>For operators, I would think about AI cost management across five layers:</p><ul><li><strong>Model routing:</strong> use the smallest or cheapest model capable of achieving the required quality.</li><li><strong>Context engineering:</strong> reduce unnecessary prompt, history and retrieval payloads.</li><li><strong>Workflow design:</strong> stop agents from entering expensive retry or refinement loops without clear incremental value.</li><li><strong>Infrastructure optimization:</strong> use caching, batching, appropriate latency tiers, open-weight models and specialized inference infrastructure where appropriate.</li><li><strong>Economic observability:</strong> know the cost, quality, latency and business outcome of every important class of request.</li></ul><p>This is where a new moat begins to appear.</p><p>If two companies use the same frontier models, the company with superior routing data, workflow design, evaluation systems and cost observability may enjoy structurally better gross margins.</p><p><strong>The model may be rented. The economic intelligence around the model can still be proprietary.</strong></p><h3>There is also a problem with the meter</h3><p>There is a more uncomfortable issue hiding underneath the token economy.</p><p>In many AI systems today, the organization selling you tokens also defines the model architecture that determines how many tokens are consumed.</p><p>That does not mean providers are behaving improperly. In fact, transparency is improving. OpenAI, for example, reports reasoning-token usage, while noting that reasoning tokens are not visible as answer text and are billed as output tokens. Google similarly includes thinking tokens in output pricing for Gemini models.</p><p>But the structural problem remains.</p><p>The customer can see <strong>how much computation was billed</strong>. It is much harder to know whether that much computation was <em>necessary</em> to produce the desired outcome.</p><p>This is a classic information asymmetry.</p><p>Imagine buying electricity from a utility that not only sets the tariff but also designs your appliances and dynamically determines how long each one runs. You might trust the utility. But you would still want an independent meter.</p><p>AI needs its equivalent.</p><p>Over time, enterprises should demand provider-independent observability: request-level cost attribution, clear separation of input, cached, reasoning and output consumption, configurable computation budgets, portable telemetry and benchmarks based on <strong>cost per successful outcome</strong> rather than simply cost per million tokens.</p><p>The market should also separate three functions that are still too frequently bundled together:</p><p><strong>creating intelligence, routing intelligence and measuring intelligence.</strong></p><p>That separation will be healthy for the ecosystem.</p><h3>Why the Stripe–OpenRouter combination is strategically interesting</h3><p>This is also the lens through which I view Stripe’s recently announced agreement to acquire OpenRouter.</p><p>On August 19, Stripe announced that it had agreed to acquire OpenRouter, subject to customary closing conditions. OpenRouter provides access and routing across more than 400 models from over 80 providers. It says it now processes more than <strong>10 trillion tokens a day</strong> and has seen at least tenfold annual growth in inference volume.</p><p>The obvious interpretation is that Stripe is expanding deeper into AI infrastructure.</p><p>I think the more interesting interpretation is economic.</p><p>Stripe has spent years helping internet businesses optimize the <strong>revenue side</strong> of their P&amp;L: payments, authorization, fraud, billing, pricing and monetization.</p><p>OpenRouter operates much closer to the <strong>cost side</strong> of an AI company’s P&amp;L: which model handles a request, at what price, with what latency, reliability and capability.</p><p>Put those together and an intriguing possibility emerges.</p><p>The infrastructure layer for AI companies may eventually help answer two questions simultaneously:</p><p><strong>How do I maximize the revenue generated by intelligence?</strong></p><p>and</p><p><strong>How do I minimize the cost of producing that intelligence?</strong></p><p>For founders, that could be powerful.</p><p>But there is also an important counterargument, and it should not be dismissed.</p><p>A routing layer becomes valuable precisely because it is model-agnostic. The more central such a layer becomes, the more important neutrality, transparency and customer control become. OpenRouter has publicly said that it will continue to place models on equal footing and that routing decisions will remain driven by what is best for the user.</p><p>That principle will matter.</p><p>If an intermediary becomes another black box, the industry has simply moved the opacity one layer higher.</p><p>If, instead, it creates greater competition between models, exposes comparative economics and gives customers more choice, it could become an important counterweight to concentration at the foundation-model layer.</p><blockquote><strong>In AI infrastructure, neutrality is not the absence of a moat. Trust can itself become the moat.</strong></blockquote><h3>India may have an unexpected advantage</h3><p>These economics should matter especially to Indian founders.</p><p>India has spent decades building businesses under constraints that Silicon Valley often did not face: lower average revenue per user, extreme scale, price-sensitive consumers and demanding unit economics.</p><p>That discipline created innovation in telecom, digital payments, ecommerce and financial services.</p><p>AI could be the next chapter.</p><p>An Indian company earning in rupees while buying dollar-denominated frontier intelligence cannot afford careless inference economics. But that constraint may become an advantage.</p><p>Indian founders can become world-class at <strong>frugal intelligence</strong>: dynamically choosing models, aggressively managing context, combining open and proprietary models, designing agents that stop when marginal reasoning no longer adds value, and monetizing the resulting outcome rather than the underlying computation.</p><p>For founders building across the India-US corridor, this could be a genuine arbitrage opportunity.</p><p>AI may flatten geography.</p><p>But economics still rewards efficiency.</p><h3>The three disciplines that will separate winners</h3><p>For founders, investors and enterprise leaders, I would focus on three things.</p><p><strong>First, price the outcome and meter the inputs.</strong><br> Do not confuse the unit customers value with the unit suppliers charge you for.</p><p><strong>Second, build AI FinOps before the bill forces you to.</strong><br> Every production AI workflow should have visibility into quality, cost, latency, retries and business outcomes. Gross-margin architecture is now product architecture.</p><p><strong>Third, preserve optionality.</strong><br> Models will leapfrog one another. Prices will move. New open models will emerge. Routing, observability and portable workflows will become strategic insurance against both technological obsolescence and pricing power.</p><p>The first era of AI rewarded access to intelligence.</p><p>The next era will reward the efficient allocation of intelligence.</p><p>The companies that win will not necessarily own the smartest model. They will know <strong>which problem deserves which model, how much reasoning the problem deserves, when another token stops creating value, and how to charge the customer for the outcome rather than the computation.</strong></p><p>That is a very different competitive advantage.</p><p>And it may prove much more durable.</p><blockquote><strong>The next great AI moat may not be the model. It may be the economics around it.</strong></blockquote><p><em>Disclosure: I lead Stripe’s business in India. Stripe announced on August 19, 2026 that it had agreed to acquire OpenRouter. The transaction was announced as subject to customary closing conditions. The views expressed here are my own.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=775d7294a765" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[When Machines Become Customers]]></title>
            <link>https://medium.com/@manishlens/when-machines-become-customers-73c7c3019d41?source=rss-6ddd323af000------2</link>
            <guid isPermaLink="false">https://medium.com/p/73c7c3019d41</guid>
            <category><![CDATA[agentic-commer]]></category>
            <category><![CDATA[agentic-ai]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[ecommerce]]></category>
            <dc:creator><![CDATA[Manish Maheshwari]]></dc:creator>
            <pubDate>Sun, 19 Jul 2026 20:07:10 GMT</pubDate>
            <atom:updated>2026-07-19T20:07:10.232Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*WKXofnm5XC5shzs_k2I8iA.jpeg" /><figcaption>Representational image (Source: Pixabay)</figcaption></figure><p>What happens when the customer at the checkout is no longer a person?</p><p>Imagine asking an AI assistant to plan a family holiday. It compares flights and hotels, checks cancellation policies, stays within budget and completes the bookings. Or consider a manufacturer whose software agent monitors inventory and orders components before production is disrupted.</p><p>This is not distant science fiction. AI is moving from generating answers to executing actions. Commerce will shift from people navigating websites to software acting on their behalf.</p><p>That transition is often presented as a better shopping experience. It is far more consequential. It requires us to redesign trust in the digital economy.</p><h3>From conversation to economic agency</h3><p>The first generation of generative AI was largely advisory. It could recommend a product or explain an insurance policy. The emerging generation is operational: it can plan, compare alternatives and complete multi-step tasks with limited supervision.</p><p>An AI system that recommends a laptop is a source of information. One authorised to buy it is an economic actor, even if it is not a legal person.</p><p>The momentum is visible. Stripe’s analysis of the top 100 AI companies on its platform found that they reached $1 million in annualised revenue in a median of 11.5 months, which is around four months faster than the fastest-growing software-as-a-service companies during the subscription boom. This is platform-specific evidence, but it shows how quickly AI-native businesses can scale.</p><p>Yet intelligence is advancing faster than the infrastructure through which it must transact.</p><h3>The delegation stack</h3><p>A useful way to think about agentic commerce is through a three-layer delegation stack: intent, identity and accountability.</p><p><strong>Intent</strong> asks what the agent was allowed to do. Did I authorise it to spend up to ₹50,000 on any flight, or only on a refundable flight within a defined window? Could it accept a subscription or substitute one brand for another?</p><p>Human instructions are often ambiguous. Financial transactions cannot be. Natural-language intent must become machine-enforceable mandates: limits on value, merchant, category and time. Higher-risk actions should trigger human approval.</p><p><strong>Identity</strong> asks who or what initiated the transaction. Merchants and banks must distinguish an authorised agent from a malicious bot or compromised system.</p><p>Payment networks are beginning to address this. Mastercard’s Agent Pay and Visa Intelligent Commerce are developing agent-specific credentials, controls and trust signals. The broader principle is crucial: an agent should not receive unrestricted access to a consumer’s primary payment credential. It should receive constrained, revocable authority appropriate to the task.</p><blockquote>The payment credential of the agentic era should be a programmable permission, not a digital copy of the consumer’s wallet.</blockquote><p><strong>Accountability</strong> asks what happens when the outcome is wrong. AI systems are probabilistic. They can misunderstand instructions or choose an option that technically satisfies a prompt while violating the user’s real preference.</p><p>Who is responsible when an agent books the wrong flight, buys a counterfeit product or enters an unintended recurring commitment?</p><p>Agentic commerce will require traceable records of the mandate, the agent’s identity, its decisions and the approvals obtained. Auditability cannot be an afterthought; it must become part of the transaction.</p><h3>Open standards, not premature winners</h3><p>Without shared standards, every AI platform would need separate integrations with merchants, banks and payment providers. Large businesses might absorb that complexity. Small merchants would not.</p><p>Initiatives such as the Universal Commerce Protocol, co-developed by Google and Shopify, alongside the Agentic Commerce Protocol, Google’s Agent Payments Protocol and the Machine Payments Protocol from Stripe and Tempo, are exploring different layers of this emerging stack.</p><p>No single protocol can yet be declared the universal winner. The objective should be interoperability without forced uniformity: open specifications, portable identity, consumer choice and competitive implementation.</p><h3>India’s advantage</h3><p>India enters this transition with an unusual advantage. UPI showed how interoperable infrastructure can expand participation while allowing banks and technology companies to compete at the application layer. ONDC is attempting a related model for digital commerce.</p><p>Agentic commerce could become the next frontier.</p><p>An Indian merchant should be discoverable by multiple AI assistants without building a bespoke integration for each one. Consumers should authorise agents with precise limits rather than surrender broad access to accounts and personal data. Banks should identify agent-initiated transactions and apply proportionate controls.</p><p>But scale also magnifies harm. An assistant that understands our finances and vulnerabilities could serve the consumer or quietly optimise for the platform paying it. Consumers must know when recommendations are sponsored, what data shaped a decision and whose commercial interests the agent represents.</p><p>Three priorities follow: build machine-readable, revocable consent; design for interoperability without locking the ecosystem into one private standard; and make consequential agent actions reconstructable through audit trails and clear liability rules.</p><p>India helped make digital payments instantaneous. It can now help make machine-mediated commerce trustworthy.</p><blockquote>The winners of the agentic economy will not be those that give machines the most power, but those that give human beings the most reliable control over that power.</blockquote><p><em>Manish Maheshwari is Country Business Head, India, at Stripe. The views expressed are personal.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=73c7c3019d41" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Deepfakes and Democracy: India’s Test of Digital Maturity]]></title>
            <link>https://medium.com/@manishlens/deepfakes-and-democracy-indias-test-of-digital-maturity-e30630356d96?source=rss-6ddd323af000------2</link>
            <guid isPermaLink="false">https://medium.com/p/e30630356d96</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[india]]></category>
            <category><![CDATA[regulation]]></category>
            <category><![CDATA[deepfakes]]></category>
            <category><![CDATA[democracy]]></category>
            <dc:creator><![CDATA[Manish Maheshwari]]></dc:creator>
            <pubDate>Sun, 26 Oct 2025 07:08:30 GMT</pubDate>
            <atom:updated>2025-10-26T07:08:30.663Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Regulation done right is not a brake. It’s a steering wheel.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*j0nmPVr8nD874DWrKIobGg.png" /><figcaption>Pixellized Image Next to Map of India — AI Generated</figcaption></figure><p>India has reached a turning point in its digital evolution. The Ministry of Electronics and Information Technology (MeitY) has proposed draft amendments to the IT Rules, 2021, to regulate synthetic content, including AI-generated videos, images, and voices. The draft, open for public consultation until November 6, 2025, aims to make the creators and platforms behind such content accountable and transparent.</p><p>If adopted, India would become one of the first democracies to formally address the dangers of AI-driven misinformation. In a year when deepfakes have infiltrated politics, entertainment, and social discourse, the timing could not be more urgent.</p><p>At stake is something deeper than data or privacy. At stake is the integrity of truth itself.</p><p><strong>What the Draft Proposes</strong></p><p>The draft amendments would:</p><ul><li>Define “synthetically generated information” as content created or altered by algorithms to resemble authentic media.</li><li>Require platforms that create or host such content to label it clearly, for example, dedicating at least 10% of visual space or the first 10% of audio to disclaimers.</li><li>Mandate automated detection systems and user declarations for synthetic media uploads.</li><li>Preserve safe-harbour protection for intermediaries that remove harmful synthetic content, while penalising those that don’t.</li></ul><p>The government’s goal is to curb the spread of impersonation, fake news, and deepfake-based fraud without stifling innovation. But translating that intent into effective enforcement will be India’s toughest governance challenge yet.</p><p><strong>Balancing Innovation and Integrity</strong></p><p>Every technological revolution tests society’s immune system. The internet tested privacy; social media tested civility; AI is now testing reality itself.</p><p>Deepfake tools can empower creators, educators, and entrepreneurs. A small business can use AI to make multilingual ads; a filmmaker can restore lost footage. But the same technology can also destroy reputations, manipulate elections, or incite violence.</p><p>The paradox is: we need AI for growth, but we need governance for trust. MeitY’s draft rules are a recognition that truth has become an infrastructure problem, not just a moral one. The question is how to build that infrastructure without choking creativity.</p><p><strong>A Smarter Path Forward</strong></p><p>India must resist the temptation to legislate faster than it can enforce. The draft’s proposed “10% visual disclaimer” is symbolically strong but technically weak. A more durable approach would rest on three pillars:</p><ol><li><strong>Verification Infrastructure:</strong> Build a digital provenance framework, akin to Aadhaar for authenticity, where each piece of content carries an invisible but verifiable signature.</li><li><strong>Tiered Accountability:</strong> Differentiate between platforms that host, generate, or monetize synthetic media. Responsibility should rise with influence.</li><li><strong>AI Literacy:</strong> Equip citizens to detect manipulation. Technology alone can’t defend democracy; informed citizens can.</li></ol><p>Such a system would make India not just compliant, but competitive. A model for balancing innovation with integrity.</p><p><strong>Learning from the World</strong></p><p>Globally, regulators are grappling with the same dilemma. The EU’s AI Act mandates watermarking of synthetic content. The US relies on voluntary corporate commitments. China requires government pre-approval for “deep synthesis” media.</p><p>India’s challenge is unique. Its digital population is vast, multilingual, and heavily reliant on social media for news. The risk of viral misinformation is therefore exponentially higher. That’s why India must pioneer a third way. Neither laissez-faire Silicon Valley nor state-controlled Beijing. A model that empowers creators, educates users, and enforces accountability.</p><p><strong>The Democratic Stakes</strong></p><p>Democracy runs on trust. And trust is fragile when truth becomes fluid. A manipulated video of a candidate, a cloned voice of a journalist, or a forged government order can undermine public confidence faster than any fact-check can repair it.</p><p>The solution isn’t censorship. It’s clarity. Regulate authenticity, not opinion.</p><p>If India can institutionalize transparency in AI-generated media, it won’t just protect its elections. It will export a model of digital responsibility for the world.</p><p><strong>The Road Ahead</strong></p><p>Public feedback on the draft amendments will close on November 6. The months that follow will decide whether India builds a framework of trust or a bureaucracy of fear. The government must consult widely, with startups, academics, technologists, and civil-society groups, to ensure regulation guides innovation instead of stifling it. This is not a fight against deepfakes alone. It’s a fight for the authenticity of the public sphere.</p><p><strong>If We Get This Right</strong></p><p>India has a once-in-a-generation opportunity to lead the world in ethical AI governance.<br>The same nation that built Aadhaar to verify identity can now build the rails for verifying truth.</p><p>Because the challenge of our age isn’t fake news. It’s fake reality. And the solution isn’t fear. It’s trust.</p><p><strong><em>Trust will be the ultimate moat in the age of AI.</em></strong></p><p>This article was first published in The Indian Express on October 24, 2025. Link <a href="https://indianexpress.com/article/opinion/columns/deepfakes-indias-ai-draft-rules-ai-it-regulations-10324555/">here</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e30630356d96" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[China Saw It Coming: The Long Game of Geo-Nationalism]]></title>
            <link>https://medium.com/@manishlens/china-saw-it-coming-the-long-game-of-geo-nationalism-efab98c93e93?source=rss-6ddd323af000------2</link>
            <guid isPermaLink="false">https://medium.com/p/efab98c93e93</guid>
            <category><![CDATA[digital]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[policy]]></category>
            <category><![CDATA[nationalism]]></category>
            <category><![CDATA[china]]></category>
            <dc:creator><![CDATA[Manish Maheshwari]]></dc:creator>
            <pubDate>Fri, 17 Oct 2025 10:01:14 GMT</pubDate>
            <atom:updated>2025-10-17T10:01:14.005Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*KVTdnaT-FDX252seBgthBA.png" /><figcaption>China and the Wall — AI Generated</figcaption></figure><p><strong>The World Laughed at the Firewall — Then Imitated It</strong></p><p>In 2009, when China blocked Facebook, Twitter, and YouTube, the world rolled its eyes. Western commentators saw paranoia; Silicon Valley saw an open goal. “Let China wall itself off,” they said. “The internet will route around it.”</p><p>Fifteen years later, the picture looks very different.</p><p>While Silicon Valley built for scale, Beijing built for sovereignty. The West treated technology as a global commons. China treated it as contested territory — the next arena of sovereignty and statecraft.</p><p>Today, as the U.S. bans TikTok, Europe regulates AI, and nations from India to Indonesia assert digital sovereignty, it’s clear: <em>China saw geo-nationalism coming long before anyone else.</em></p><p><strong>From Isolation to Insight</strong></p><p>The Great Firewall was never just about censorship. It was an act of strategic imagination.</p><p>By walling off foreign platforms, China didn’t retreat from the internet — it <em>re-architected</em> it. It created space for domestic champions to grow, and in doing so, built one of the world’s most self-contained and self-reliant digital ecosystems.</p><p>Then came <em>Made in China 2025</em>, a policy to reduce dependence on Western technology and dominate critical sectors like semiconductors, AI, and 5G. And soon after, the <em>Digital Silk Road</em>, Beijing’s plan to export its digital infrastructure across Asia, Africa, and Latin America.</p><p>These weren’t isolated initiatives. They were the scaffolding of a new model: geo-national technology — technology aligned with national interest, designed for geopolitical leverage.</p><blockquote><em>What looked like isolation was actually anticipation.</em></blockquote><p><strong>The Architecture of Sovereignty</strong></p><p>China’s playbook for digital power can be understood through three interlocking pillars: control, scale, and narrative.</p><p><strong>1. Control the Stack</strong></p><p>Beijing realized early that sovereignty begins at the hardware layer.<br>Data sovereignty is meaningless if your chips are made abroad, your cloud runs on foreign code, or your supply chain depends on another superpower.</p><p>So China poured hundreds of billions into domestic chip fabrication, cloud infrastructure, and satellite networks.<br>The result: partial self-reliance in every layer of the digital stack, from silicon to software.</p><p>When the U.S. restricted exports of advanced semiconductors in 2023 and 2024, China didn’t collapse. It improvised. Because for two decades, it had been preparing for that day.</p><blockquote><em>You can’t be sovereign on borrowed compute.</em></blockquote><p><strong>2. Scale the Ecosystem</strong></p><p>The next step was scale, not just in production but in <em>platform reach</em>.</p><p>Inside its firewall, China incubated national champions: Alibaba in commerce, Tencent in social media, Baidu in search, and ByteDance in AI-driven entertainment.<br>These companies thrived in a market of a billion users, protected from foreign competition but fiercely competitive with one another.</p><p>Outside its borders, China extended influence through the <em>Digital Silk Road, </em>a network of fiber cables, smart-city systems, and 5G networks across the developing world. By financing and building digital infrastructure abroad, Beijing exported its technological standards and governance models.</p><p>This was statecraft by other means: a way to project power through platforms rather than politics.</p><p><strong>3. Shape the Narrative</strong></p><p>Perhaps China’s most under-appreciated move was narrative discipline.<br>While Western democracies treated digital regulation as a constraint, Beijing presented it as <em>nation-building</em>.</p><p>Control was framed not as censorship, but as <em>sovereignty and stability.</em><br>Chinese citizens came to see platforms like WeChat, Alipay, and Douyin not as state instruments, but as national achievements — symbols of self-reliance in a global system stacked against them.</p><p>That sense of collective digital destiny gave China something many open societies lack: alignment between the state, private enterprise, and the citizen.</p><p><strong>The Result: From Follower to Framer</strong></p><p>Two decades ago, China was the student of Western technology.<br>Today, it is the one writing the rules — whether in AI governance, 5G infrastructure, or export controls.</p><p>When the West debated whether platforms were too powerful, China had already fused its digital ecosystem with state priorities. When Washington began restricting chip exports, Beijing had already built fallback capacity. When Europe drafted the AI Act, China had already implemented algorithmic filing requirements.</p><p>In many ways, the geo-national age we live in is China’s design — not its dilemma.</p><p><strong>The Inevitable Catch-Up</strong></p><p>The irony is that the same governments that once ridiculed China’s “digital sovereignty” are now adopting their own versions of it.</p><ul><li>The U.S. forces TikTok to restructure its ownership.</li><li>The EU mandates algorithmic transparency.</li><li>Democracies everywhere debate data localization and AI safety.</li></ul><p>Each move echoes a principle China adopted long ago: <em>control the technology you depend on, or someone else will control you.</em></p><blockquote><em>While Silicon Valley built for users, China built for nations.</em></blockquote><p><strong>What the World Misunderstood</strong></p><p>For years, Western analysts saw China’s model as illiberal and unsustainable. They missed the deeper logic: China was building <em>resilience</em> in an age of interdependence.</p><p>It understood that the 21st century wouldn’t be defined by territory but by <em>technological spheres of influence.</em><br>And it acted accordingly — creating digital borders before digital warfare became mainstream, investing in chips before supply chains fractured, and blending public and private sector capacity before AI became dual-use.</p><p>The result is that China enters the 2030s not as a tech colony or consumer market, but as a geopolitical force with its own stack, standards, and story.</p><p><strong>The Lesson for the World</strong></p><p>You don’t have to admire China’s methods to acknowledge its foresight.<br>It saw that technology would be the new language of power — and built fluency early.</p><p>Geo-nationalism isn’t an accident of politics. It’s the natural evolution of technology once it becomes too strategic to be neutral.</p><p>China didn’t stumble into that reality. It <em>architected</em> it. <em>China saw the age of geo-national tech — and built the playbook everyone else is now following.</em></p><p><strong>The Closing Thought</strong></p><p>History often rewards those who see the next paradigm before it becomes visible to everyone else.<br>China did that in technology. It turned sovereignty into software, regulation into leverage, and data into diplomacy.</p><p>The next question — and the subject of my next column — is whether <em>India</em> can write the sequel: Can a democracy design its own model of digital sovereignty, one built not on control but on trust?</p><p>For now, one thing is clear.<br><em>China saw it coming. And the world is only just catching up.</em></p><p>***</p><p>This article was published in India Today on October 15, 2025. Link <a href="https://www.indiatoday.in/opinion/story/china-saw-it-coming-the-long-game-of-geo-nationalism-2803582-2025-10-15">here</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=efab98c93e93" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[TikTok and the Death of Borderless Tech]]></title>
            <link>https://medium.com/@manishlens/tiktok-and-the-death-of-borderless-tech-b96b2e138073?source=rss-6ddd323af000------2</link>
            <guid isPermaLink="false">https://medium.com/p/b96b2e138073</guid>
            <category><![CDATA[borderlesstech]]></category>
            <category><![CDATA[digital-sovereignty]]></category>
            <category><![CDATA[geopolitics-tech]]></category>
            <category><![CDATA[geonationaltech]]></category>
            <category><![CDATA[future-of-tech]]></category>
            <dc:creator><![CDATA[Manish Maheshwari]]></dc:creator>
            <pubDate>Mon, 29 Sep 2025 05:48:44 GMT</pubDate>
            <atom:updated>2025-09-29T05:48:44.425Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="Flaming TikTok logo over a divided world — AI Generated" src="https://cdn-images-1.medium.com/max/947/1*JSps2dt0Xf3jIZc7_6GVig.jpeg" /><figcaption>Flaming TikTok logo over a divided world — AI Generated</figcaption></figure><p>For much of the past two decades, technology giants behaved as if they were sovereign entities, floating above national boundaries. Google promised to “organize the world’s information.” Facebook sought to “connect the world.” Microsoft ran cloud infrastructure across continents with little regard for political borders. Even Twitter, where I once worked, saw itself as a global town square, not an American company.</p><p>That illusion has now cracked. The TikTok saga — culminating in Donald Trump approving its continued presence in the U.S. but only after forcing a structural split between its U.S. and non-U.S. arms — marks a turning point. For the first time, a major social media giant had to reconfigure itself to align with the national priorities of a host country. TikTok may be the first domino, but it will not be the last.</p><p><strong>From Supra-National to Geo-National</strong></p><p>The rise of Big Tech can be divided into three phases.</p><p>The first was the supra-national era of the 2000s and 2010s. Tech companies saw themselves as global utilities. Their allegiance was to users, shareholders, and ecosystems — not governments. Borders mattered only insofar as local regulations demanded minor adjustments.</p><p>The second phase began around 2020, the national security era. TikTok was banned in India. Huawei was excluded from Western 5G networks. Semiconductors became a battlefield in the U.S.–China rivalry. Suddenly, platforms were not just businesses — they were treated as strategic assets, potential threats, and bargaining chips in diplomacy.</p><p>Now we are entering the geo-national era. Companies must align their structures, supply chains, and governance models with national priorities. TikTok’s split is only the most visible case. Nvidia cannot ship its most advanced GPUs to China. Rare earth minerals are being stockpiled and friend-shored. And in artificial intelligence, nations are moving quickly to assert control: the European Union with its AI Act, the United States with executive orders on AI safety and export restrictions, and China with its algorithmic transparency rules.</p><p>The message is clear: technology is no longer borderless infrastructure. It is national infrastructure, and therefore inherently political.</p><p><strong>The Counterargument</strong></p><p>Some argue this shift is overstated. TikTok’s algorithm is largely the same everywhere. Google, Meta, and Microsoft continue to operate across more than a hundred countries. AI researchers from Boston to Beijing still co-author papers.</p><p>But this view misses the deeper reality. The user experience may remain global, but the ownership, governance, and compliance structures increasingly reflect national interests. A platform may look supra-national on the surface, yet beneath it is being sliced into national variants, each subject to its host country’s geopolitical gravity.</p><p><strong>India at the Crossroads</strong></p><p>India has already played a decisive role in this story by banning TikTok in 2020, well before the U.S. moved against it. That decision reflected not only national security concerns but also a recognition of the strategic value of controlling digital rails. Since then, India has doubled down on building its own digital public infrastructure — from UPI in payments to ONDC in commerce.</p><p>For Washington, TikTok represents a Trojan horse of Chinese influence. For Beijing, it is one of the few examples of a Chinese consumer technology platform succeeding globally. For the Global South, this is both a risk and an opening: the risk of being forced to pick sides in a tech cold war, and the opportunity to build sovereign alternatives that align with national priorities.</p><p><strong>Lessons for AI</strong></p><p>The TikTok story is not just about social media. It foreshadows the future of artificial intelligence.</p><p>AI may appear borderless today — models trained in California are accessible from Bengaluru, and researchers in Shenzhen and Berlin publish on the same arXiv preprints. But the same forces that fractured TikTok will shape AI. Who controls the data? Who controls the chips? Who regulates how models are deployed? These questions will increasingly be answered through the lens of national interest.</p><p>The European Union’s AI Act imposes obligations on high-risk systems. The U.S. has already restricted exports of advanced GPUs and demanded safety commitments from leading AI labs. China requires algorithm filings with regulators. These moves, while different in intent, all reflect the same reality: nations will not allow AI to remain a supra-national domain.</p><p>The world of AI, like the world of social media, will be one of “modular globalism”: interoperable at the user level, but fragmented at the governance level.</p><p><strong>The Road Ahead</strong></p><p>What does this mean for business and society? For founders, it means political strategy is now part of go-to-market. Scaling globally requires not just product-market fit but regulator-market fit. For investors, national risk is the new market risk. Regulatory choke points can alter valuations as much as adoption curves. For policymakers, the challenge is to exercise sovereignty without suffocating innovation. Fragmentation is a risk, but so is blind dependence on supra-national platforms.</p><p>If India’s digital public infrastructure offers one lesson, it is this: sovereignty and openness can coexist. The task is to design networks of networks, where nations retain control but users and businesses still benefit from global interoperability.</p><p><strong>Closing Thought</strong></p><p>The TikTok saga is not an anomaly. It is a preview of the next decade. Social media platforms, AI models, semiconductor supply chains, even rare earths — all are being pulled into the gravitational field of geopolitics.</p><p>Tech is no longer supra-national. It is geo-national. And the winners will be those who learn to operate at the intersection of innovation and sovereignty. History won’t wait. Neither should we.</p><p>***</p><p>This article was published in India Today on September 28, 2025. Link <a href="https://www.indiatoday.in/opinion/story/tiktok-saga-highlights-shift-to-geo-national-tech-governance-in-india-2794633-2025-09-28">here</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b96b2e138073" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[PhonePe Races Ahead: Why Flipkart Hesitates on IPO]]></title>
            <link>https://medium.com/@manishlens/phonepe-races-ahead-why-flipkart-hesitates-on-ipo-904385a08698?source=rss-6ddd323af000------2</link>
            <guid isPermaLink="false">https://medium.com/p/904385a08698</guid>
            <category><![CDATA[ecommerce]]></category>
            <category><![CDATA[upi]]></category>
            <category><![CDATA[payments]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[ipo]]></category>
            <dc:creator><![CDATA[Manish Maheshwari]]></dc:creator>
            <pubDate>Fri, 26 Sep 2025 02:51:47 GMT</pubDate>
            <atom:updated>2025-09-26T04:08:17.007Z</atom:updated>
            <content:encoded><![CDATA[<p><em>PhonePe is racing to IPO while Flipkart hesitates. Same parent, same market, different outcomes. What does this reveal about India’s maturing tech ecosystem — and what should founders learn?</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fAT73--g4rFTAYV_ZUBVpA.png" /><figcaption>PhonePe vs. Flipkart</figcaption></figure><p>What if two companies born under the same roof, nurtured by the same parent, and scaling in the same Indian market end up taking radically different IPO paths? That’s the paradox of PhonePe and Flipkart.</p><p>PhonePe is sprinting toward a public listing, while Flipkart, once the poster child of India’s startup boom, is still figuring out its IPO playbook. This divergence isn’t just about two companies — it reveals deeper lessons about India’s capital markets, strategic positioning, and the maturing tech ecosystem.</p><h3>The Setup: India’s IPO Moment</h3><p>India is at an inflection point.<br>Its consumer internet giants — Paytm, Zomato, Nykaa, Policybazaar — have already tested public markets with mixed results. Now comes the second wave: fintechs, e-commerce platforms, and SaaS companies that are better capitalized, more disciplined, and operating in a post-correction environment.</p><p>For founders and investors, the IPO question is no longer “if” but “how” and “when.” The stakes are massive: public markets not only provide liquidity but also signal legitimacy, setting benchmarks for the entire ecosystem.</p><h3>Why PhonePe Is Ready</h3><p>PhonePe’s IPO readiness rests on three pillars:</p><ol><li><strong>Clear Market Leadership</strong><br>PhonePe dominates India’s UPI payments with over 45% share. Unlike Paytm, which juggles multiple business lines, PhonePe has built a razor-sharp identity around payments and financial services.</li><li><strong>Simpler Story for Public Investors</strong><br>Public markets reward clarity. PhonePe can tell a straightforward story: growing transaction volumes, financial inclusion, and monetization through insurance, mutual funds, and lending. Investors like clean narratives.</li><li><strong>Regulatory Tailwinds</strong><br>The Indian government has blessed UPI as the backbone of digital payments. PhonePe is not fighting regulators; it’s aligned with the system. That makes it a safer bet compared to fintechs seen as too disruptive.</li></ol><p>In short: PhonePe looks like a utility with upside. Markets love that.</p><h3>Why Flipkart Hesitates</h3><p>Flipkart, in contrast, faces headwinds:</p><ul><li><strong>Profitability Puzzle: </strong>E-commerce remains a capital-intensive business. Margins are thin, and competition with Amazon and Reliance JioMart is unrelenting. Investors know this is a long game — but public markets are impatient.</li><li><strong>Complex Structure: </strong>Flipkart’s sprawl includes fashion (Myntra), logistics (Ekart), and now even healthcare. Great for private growth, messy for public narratives.</li><li><strong>Global Parent Dynamics: </strong>Walmart owns a majority stake in Flipkart. That creates tension: should Flipkart IPO in India, the US, or both? Walmart wants global visibility; Flipkart wants local resonance. This delay in alignment slows decision-making.</li></ul><p>Flipkart is still a juggernaut, but unlike PhonePe, its IPO case is harder to compress into a three-minute elevator pitch.</p><h3>Lessons: What Founders and Investors Should Learn</h3><p>This divergence offers a powerful framework for thinking about IPO readiness.</p><h3>1. IPOs Are About Narrative, Not Just Numbers</h3><p>PhonePe’s path shows that clarity of narrative — not just GMV or user counts — drives market confidence. Founders should ask: <em>Can I explain my business in one sentence to a retail investor?</em></p><h3>2. Spin-Offs Create Sharper Value</h3><p>PhonePe’s separation from Flipkart unlocked its ability to chart an independent IPO. Conglomerates may create synergies privately, but in public markets, investors reward focus.</p><h3>3. Timing Is a Strategic Weapon</h3><p>PhonePe is capitalizing on momentum in digital payments before saturation sets in. Flipkart, by waiting, risks being overtaken by changing market moods or regulatory shifts.</p><h3>4. The US–India Corridor Matters</h3><p>Flipkart’s global entanglements highlight a bigger theme: Indian companies must decide whether their “center of gravity” lies in India’s capital markets or on Wall Street. That decision shapes everything from valuation multiples to governance expectations.</p><h3>The Bigger Picture: India’s Tech Maturity</h3><p>PhonePe and Flipkart’s diverging IPO strategies mirror India’s broader startup ecosystem. The first generation of unicorns is no longer just chasing growth at all costs. They’re wrestling with profitability, governance, and market positioning.</p><p>This is a healthy sign. India is moving from adolescence to adulthood in tech. The capital markets will demand discipline — but they’ll also reward it.</p><h3>Actionable Takeaways</h3><p>For founders, investors, and policymakers, three lessons stand out:</p><ul><li><strong>Simplify the story</strong> — Public markets punish ambiguity. Be clear about your identity and path to profitability.</li><li><strong>Design for separation</strong> — Build companies so business lines can stand independently if needed. It maximizes optionality.</li><li><strong>Think globally, act locally</strong> — The IPO venue debate (India vs. US) will become a recurring theme. Founders must weigh valuation premiums against domestic relevance.</li></ul><h3>Closing with Power</h3><p>PhonePe and Flipkart teach us that IPOs are not just financial events — they are narrative events. They crystallize a company’s story in the minds of millions of investors.</p><p>If India’s tech ecosystem gets this right, we won’t just create a few successful IPOs. We’ll build the muscle memory for turning unicorns into enduring public institutions.</p><blockquote><strong>Tweetable line:</strong><br>“<em>The IPO isn’t the finish line. It’s the moment your story meets the market. PhonePe knows its story. Flipkart is still editing the draft.</em>”</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=904385a08698" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[When AI Becomes a VC: The Future of Investing]]></title>
            <link>https://medium.com/@manishlens/when-ai-becomes-a-vc-the-future-of-investing-877d3bc467d7?source=rss-6ddd323af000------2</link>
            <guid isPermaLink="false">https://medium.com/p/877d3bc467d7</guid>
            <category><![CDATA[venture-capital]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Manish Maheshwari]]></dc:creator>
            <pubDate>Mon, 22 Sep 2025 11:56:10 GMT</pubDate>
            <atom:updated>2025-09-24T06:26:28.784Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fOvOv08y7tzzPfi_LWN1Xg.png" /><figcaption>AI in VC Investing</figcaption></figure><h3>What if the next unicorn isn’t picked by a human at all?</h3><p>For decades, venture capital has been defined by human judgment — the gut feel of a partner, the pattern recognition honed over years, the serendipitous coffee chat that sparks conviction. But what happens when those instincts are no longer uniquely human?</p><p>We are standing at the most important inflection point since the invention of the spreadsheet. Just as Excel industrialized financial analysis in the 1980s, AI is about to industrialize <em>judgment</em>. The very essence of what makes venture capital special — selection, support, and signaling — is being redefined by algorithms that can see patterns we cannot, test scenarios at scale, and even negotiate terms.</p><p>This isn’t man vs. machine. This is man + machine reshaping the business of funding the future.</p><h3>Why AI Matters to Venture Capital — Right Now</h3><p>AI is not just another tool in the VC toolkit; it changes the physics of the industry itself. Three forces are converging:</p><ol><li><strong>Data Explosion</strong> — Startups leave digital breadcrumbs everywhere: product usage, hiring patterns, GitHub commits, customer reviews. AI can synthesize these signals into a live pulse of a company’s trajectory.</li><li><strong>Capital Commoditization</strong> — With trillions of dollars sloshing in private markets, money itself is no longer scarce. What’s scarce is differentiated insight. AI flattens access to data but also raises the bar for edge.</li><li><strong>Speed of Innovation</strong> — Founders can now spin up MVPs with AI copilots in weeks, not months. This compresses cycles and forces VCs to adapt from quarterly check-ins to real-time monitoring.</li></ol><p>The stakes? If VCs don’t embrace AI, they risk becoming the <em>disrupted</em> rather than the disruptors.</p><h3>A New Framework: The AI-Driven VC Flywheel</h3><p>Think of AI as an <strong>invisible co-investor</strong> that rewires the venture value chain. The flywheel looks like this:</p><ol><li><strong>Deal Sourcing</strong> → AI scrapes signals across LinkedIn, GitHub, academic preprints, and product analytics to surface startups before they hit TechCrunch.</li><li><strong>Due Diligence</strong> → Natural language models synthesize customer interviews, legal docs, and competitive landscapes in hours, not weeks.</li><li><strong>Portfolio Support</strong> → AI copilots provide portfolio companies with instant access to best practices in hiring, go-to-market, and fundraising.</li><li><strong>LP Relations</strong> → Automated reporting, performance simulation, and predictive liquidity models strengthen trust with limited partners.</li></ol><p>Each loop makes the fund smarter and faster — a compounding edge in a business where timing is everything.</p><h3>The Counterarguments — and Why They Fall Short</h3><p>Skeptics raise three main objections:</p><ul><li><strong>“VC is about relationships, not data.”</strong> True, trust still matters. But relationships now begin with algorithms. Founders are more likely to take a VC meeting if the outreach is hyper-personalized by AI, referencing their actual traction, not a cold pitch.</li><li><strong>“AI can’t predict black swans.”</strong> Also true. But neither can humans. What AI does offer is probabilistic foresight — surfacing anomalies and tail risks faster than gut feel ever could.</li><li><strong>“This will make all VCs the same.”</strong> Paradoxically, the opposite may happen. Just as every hedge fund has access to Bloomberg but only a few generate alpha, the differentiator will be <em>how</em> each VC trains, fine-tunes, and integrates AI into its philosophy.</li></ul><p>In short: AI won’t erase human edge; it will amplify the difference between average and exceptional.</p><h3>Global Lens: The US–India Corridor</h3><p>Nowhere will this transformation be more visible than in the US–India corridor.</p><ul><li><strong>For US investors</strong>, AI reduces friction in scouting Indian startups, bridging distance with real-time diligence. Suddenly, a seed-stage founder in Bangalore is as visible to a San Francisco fund as one in Palo Alto.</li><li><strong>For Indian founders</strong>, AI levels the playing field. They can build capital-efficient companies with AI copilots, making them attractive to global investors despite smaller domestic funding pools.</li><li><strong>For policy makers</strong>, AI raises questions of regulation and trust. If capital allocation decisions are increasingly machine-driven, whose values and biases are embedded in those models?</li></ul><p>AI flattens geography but geopolitics raises new walls. Smart founders will build bridges.</p><h3>Second-Order Effects: What Happens Next</h3><p>If we look around the corner, three seismic shifts emerge:</p><ol><li><strong>AI-First Funds</strong> — Expect to see VC firms where AI agents run sourcing, diligence, and portfolio dashboards, with humans as final editors of conviction. The “AI partner” may soon sit on the investment committee.</li><li><strong>Founder Empowerment</strong> — Startups can bypass traditional VC altogether, raising from AI-matched pools of micro-investors via tokenized deals. The monopoly on judgment is broken.</li><li><strong>Regulatory Scrutiny</strong> — As AI begins to influence capital flows, regulators will demand explainability. Transparency in models will become a prerequisite, just as financial disclosures are today.</li></ol><p>History won’t wait. Neither should we.</p><h3>Actionable Takeaways for Stakeholders</h3><ul><li><strong>For Founders:</strong> Treat AI as your invisible chief of staff. Use it not only to build product but to craft your fundraising narrative and target the right investors.</li><li><strong>For VCs:</strong> Don’t just adopt AI tools — integrate them into your culture. The firm that learns fastest compounds advantage.</li><li><strong>For LPs:</strong> Ask funds not “Do you use AI?” but “How does AI change your decision-making?” The difference reveals true edge.</li><li><strong>For Policymakers:</strong> Design regulation as a steering wheel, not a brake. Encourage innovation while demanding transparency in capital-allocation algorithms.</li></ul><p>The winners in this new era will be those who pair <strong>machine precision with human conviction</strong>. AI won’t replace VCs — but VCs who don’t leverage AI will be replaced by those who do.</p><p>If we get this right, we won’t just fund unicorns. We’ll fund <em>civilizations worth living in</em>.</p><p>(Note: AI tools have been generously used to write this piece 😃)</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=877d3bc467d7" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Race for Dominance: Meta’s Threads vs. Musk’s Twitter — Who Will Emerge Victorious?]]></title>
            <link>https://medium.com/@manishlens/the-race-for-dominance-metas-threads-vs-musk-s-twitter-who-will-emerge-victorious-60660ec44562?source=rss-6ddd323af000------2</link>
            <guid isPermaLink="false">https://medium.com/p/60660ec44562</guid>
            <category><![CDATA[mark-zuckerberg]]></category>
            <category><![CDATA[social-media]]></category>
            <category><![CDATA[twitter]]></category>
            <category><![CDATA[meta]]></category>
            <category><![CDATA[elon-musk]]></category>
            <dc:creator><![CDATA[Manish Maheshwari]]></dc:creator>
            <pubDate>Sun, 09 Jul 2023 12:10:25 GMT</pubDate>
            <atom:updated>2023-07-09T12:25:41.897Z</atom:updated>
            <content:encoded><![CDATA[<h3>The Race for Dominance: Meta’s Threads vs. Musk’s Twitter — Who Will Emerge Victorious?</h3><figure><img alt="Twitter’s Larry Bird Fading" src="https://cdn-images-1.medium.com/max/480/1*U2inacK7_Jjt7kY5pXPRQw.png" /><figcaption>Twitter’s Larry Bird Fading</figcaption></figure><p>In the realm of social media, the landscape is ever-changing, giving rise to opportunities for newcomers to challenge the status quo. This dynamic is evident in the recent launch of Meta’s new social media platform, Threads. With Twitter still dealing with the tumultuous takeover by Elon Musk, the question on everyone’s mind is: does Threads stand a chance of dethroning the microblogging giant?</p><h3>Threads: The New Kid on The Block</h3><p>Thread’s primary appeal is its combination of long-form content with the immediacy and reach of social media, a fusion that has garnered significant attention from users craving a richer content experience. Meta has clearly been watching the social media landscape and has made a strategic move to fill in the gap left open by Twitter’s brevity-focused approach.</p><p>Moreover, with a backing like Meta, the company that owns Facebook, Instagram, and WhatsApp, Threads stands a good chance of attracting a sizeable user base right out of the gate. The platform can leverage the established user base of these sister applications, capitalizing on cross-platform integrations and the vast resources available at Meta’s disposal.</p><h3>The Musk Effect: A Double-Edged Sword for Twitter</h3><p>On the other side of the ring, we have Twitter, under the new management of the maverick tech entrepreneur, Elon Musk. Musk’s takeover of Twitter has seen some strategic mistakes which could provide Threads with the leg-up it needs to pose a significant challenge.</p><p>One of the major missteps has been Musk’s effort to push for monetization at the cost of user reach and creator’s control over their digital identity. The introduction of a subscription model, Twitter Blue, was met with mixed reactions. The public has grown accustomed to free social media, and asking them to pay, albeit for advanced features, has proven a hard pill to swallow. And taking back the verified blue ticks from established creators and celebrities and asking them to pay shook their trust in Twitter.</p><h3>The Battlefront: What Twitter Can Do</h3><p>Despite the odds, it’s far too early to write off Twitter. It still holds significant brand value and boasts an influential user base, which it can leverage. However, it needs to rethink some of its strategies to maintain its stronghold.</p><p>Firstly, Twitter should focus on refining and improving user and creator experience rather than radical changes. The introduction of audio tweets and other novel features was a good start, but more needs to be done. This could include better content curation and recommendation systems, improved search and discovery functionalities, and creating a more interactive and engaging interface for both users and creators.</p><p>It should reconsider its monetization strategy. While a subscription model is not inherently wrong, Twitter needs to ensure that its value proposition is clear and compelling. Twitter could look to providing unique, value-adding features that would incentivize users to upgrade to the paid version.</p><p>Lastly, Twitter can work on greater transparency, accountability, and user control. This could mean clearer policies around content moderation, stricter handling of misinformation and hate speech, and providing users and creators with more control over their data and privacy.</p><h3>The Bottom Line</h3><p>In the face of growing competition from Threads, Twitter needs to go back to its roots and focus on what made it popular in the first place — a platform for quick, real-time updates and conversations. It needs to recognize that while change is necessary, not all change is beneficial, and sometimes, subtler, user-centric improvements can make all the difference.</p><p>As for Threads, the platform is promising, and it’s off to a good start. However, the real test lies in its ability to sustain momentum, provide a robust and reliable platform, and continually evolve to cater to users’ and creators’ changing preferences and needs. The day news starts breaking on Threads first is the day when the countdown to Twitter’s demise will begin.</p><p>In the world of social media, it’s not just about who arrives first, but who adapts the fastest and the smartest. As we watch this intriguing battle unfold, one thing is clear — the true winners will be the users and creators who stand to benefit from the innovations that this competition will undoubtedly bring.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=60660ec44562" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Monetize Your Craft with Fanory and Strengthen Your Relationship with Superfans]]></title>
            <link>https://medium.com/@manishlens/monetize-your-craft-with-fanory-and-strengthen-your-relationship-with-superfans-4a6a35663922?source=rss-6ddd323af000------2</link>
            <guid isPermaLink="false">https://medium.com/p/4a6a35663922</guid>
            <category><![CDATA[artist]]></category>
            <category><![CDATA[creator-economy]]></category>
            <category><![CDATA[creators]]></category>
            <category><![CDATA[monetization]]></category>
            <category><![CDATA[social-media]]></category>
            <dc:creator><![CDATA[Manish Maheshwari]]></dc:creator>
            <pubDate>Sat, 15 Apr 2023 09:35:20 GMT</pubDate>
            <atom:updated>2023-04-15T09:35:20.386Z</atom:updated>
            <content:encoded><![CDATA[<p>Gone are the days when artists, writers, photographers, and other creators depended solely on publishers or galleries to monetize their craft. Today, with the help of social media platforms, creators can easily connect with their audience and showcase their talent to a global audience. However, there is a downside to this. Platforms like Instagram, YouTube, and Facebook take a significant share of the creator’s earnings and expose them to platform risks. Therefore, it’s time for creators to think strategically and identify their superfans to build a direct relationship with them. This blog post will introduce you to Fanory, a direct monetization tool for creators, and show you how you can use it to monetize your craft and build a stronger relationship with your superfans.</p><p>Identify Your Superfans:<br>Your superfans are those who love your work and are both willing and able to pay for it. They may already be following you on social media, commenting on your posts, sharing your content, and engaging in your online community. To identify your superfans, you can use social media analytics tools or simply ask your audience about their interest and willingness to pay for your work.</p><p>Connect Directly with Your Superfans:<br>Once you’ve identified your superfans, it’s time to build a direct relationship with them. By connecting with them one-on-one and providing them exclusive access to your work, you can strengthen their loyalty, and encourage them to support you financially. Fanory, a direct monetization tool, makes it easy for creators to connect with superfans and provide them access to exclusive content or services.</p><p>Leverage Fanory’s Monetization Tools:<br>Fanory provides creators with multiple monetization options that allow them to earn money directly from their superfans. Creators can set up paywalls or subscriptions for their exclusive content or services, receive donations from their superfans, or sell merchandise directly on Fanory’s platform. Fanory handles all the payment processing, ensuring secure and hassle-free transactions.</p><p>Diversify Your Income Stream:<br>Working with direct monetization tools like Fanory allows creators to have an additional income stream that is independent of social media platforms. Diversifying your income stream is crucial for creators, as it provides them with financial stability and independence. Creators should not rely solely on one platform or source of income, as it exposes them to risk and limits their earning potential.</p><p>Nurture Your Relationship with Superfans:<br>Lastly, creators should understand the importance of nurturing their relationships with their superfans. By consistently providing high-quality content, engaging with their audience, and showing appreciation for their support, creators can build a strong and loyal fanbase. With Fanory, creators can easily track their superfans’ engagement and support, and reward them accordingly, which strengthens their relationship and encourages them to continue supporting their craft.</p><p>To conclude, creators need to be strategic and proactive in monetizing their craft and building a direct relationship with their superfans. Fanory provides creators with a direct monetization tool that allows them to diversify their income stream, reduce their platform risks, and strengthen their relationships with superfans. By leveraging Fanory’s monetization tools, creators can earn money directly from their superfans and provide them with exclusive access to their work. Furthermore, Fanory’s analytics tools allow creators to track their performance and engagement, enabling them to fine-tune their strategy and grow their fanbase. If you’re a creator who wants to monetize your craft and build a loyal fanbase, Fanory is the way to go!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4a6a35663922" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[India’s National Education Policy 2020: Acknowledging problems and showing good intent]]></title>
            <link>https://medium.com/@manishlens/indias-national-education-policy-2020-acknowledging-problems-and-showing-good-intent-5dd47eac7659?source=rss-6ddd323af000------2</link>
            <guid isPermaLink="false">https://medium.com/p/5dd47eac7659</guid>
            <category><![CDATA[education]]></category>
            <category><![CDATA[national-education-policy]]></category>
            <category><![CDATA[higher-education]]></category>
            <category><![CDATA[indian-education]]></category>
            <category><![CDATA[india]]></category>
            <dc:creator><![CDATA[Manish Maheshwari]]></dc:creator>
            <pubDate>Tue, 08 Sep 2020 18:23:54 GMT</pubDate>
            <atom:updated>2020-09-08T18:23:54.034Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_D2CwR9wm28BrDLmBZhDBg.jpeg" /><figcaption>Shifting focus (Image: SRCC)</figcaption></figure><p>Last month, the Indian government published the National Education Policy 2020 outlining its new education system’s vision. The new policy is a revision after 34 years! The last time it published an education policy was in 1986.</p><p>This development is momentous. India has the world’s largest population of young people (age &lt;35 years). Furthermore, this population will continue to grow for several decades! So, this policy is not just crucial for India but also for the world.</p><p>Education is very close to my heart. I attribute my achievements to it. Without educational opportunities, it would not have been possible for me to achieve socio-economic mobility to move from an ordinary family in India’s hinterland to India’s capital for my undergraduate studies and, after that, to the US for graduate school.</p><p>Education can also address deep-rooted institutional inequality and prejudices to create a level playing field. My educational background has provided me unhindered access to opportunities globally that my parents or community members could not even dream of. Therefore, I believe that education is the best tool to create a fairer and more inclusive world where each one can achieve their full potential.</p><p>I, being a product of the Indian education system, have strong views on this subject. Furthermore, since I had the good fortune to study at a multi-disciplinary Ivy League university for my graduate program and get exposed to a global student body across a multitude of discipline, I can compare and contrast things to tease out what works and what doesn’t in the administration of education in India.</p><p>First of all, I must commend the policymakers for acknowledging some of the major issues upfront. These issues range from low enrollment ratio, particularly among socio-economically disadvantaged groups, coaching culture to pass the exams, rigid academic compartmentalization, factory production of engineers and doctors, overlooked vocational education, unwieldy regulation, and lack of focus on teachers. More importantly, there is an emphasis on changing education’s intent to make it experiential, holistic, integrated, inclusive, inquiry-driven, discovery-oriented, learner-focused, discussion-based, flexible, and enjoyable. There is also a genuine effort to embrace technology-driven changes to make education future-ready. Focus on teachers is noteworthy, especially how to empower and motivate them, tap into local and women talent, and ensure professional development, career management, and progression. I summarized the key points of National Education Policy 2020 in the <a href="https://medium.com/@manishm345/indias-national-education-policy-2020-key-points-a78957747718?source=friends_link&amp;sk=54e18fd5577c9434b2c147e0f2b18a4f">other piece</a> I wrote earlier, which can be accessed <a href="https://medium.com/@manishm345/indias-national-education-policy-2020-key-points-a78957747718?source=friends_link&amp;sk=54e18fd5577c9434b2c147e0f2b18a4f">here</a>.</p><p>However, there are three specific areas where the policy falls short:</p><p><strong>1. Framework for higher education:</strong> India is large and diverse. Furthermore, there is a vast disparity in educational need ranging from excellence to expansion to inclusion. A single broad-brush solution is unlikely to work. What is needed is a tiered framework to deliver the right balance of teaching, research, and training at a large scale and facilitate consistent administration of tier-appropriate governance structure and success metrics. The conceptual tiers would comprise:</p><p>(a) research and teaching-focused institutions,</p><p>(b) teaching-focused institutions,</p><p>(c) community-focused institutions, and</p><p>(d) employment skills-focused institutions.</p><p>The central government should focus on the first tier of institutions (e.g., by converting IITs, IIMs, Central Universities into world-class, full-fledged, multi-disciplinary universities focused on research and training), while continuing to ensure inclusion in the other tiers. There is a need to establish linkages across the different tiers to maximize system-wide leverage.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/717/1*nVxd4Hzi2tAWV55BXNes8A.png" /><figcaption>A tiered framework of institutions can effectively fulfill nation’s diverse needs</figcaption></figure><p><strong>2. Governance:</strong> While there is a discussion on this in the National Education Policy, what is needed is a far more significant governance improvement. This improvement can be achieved by:</p><p>(a) ensuring that an independent and fully empowered board runs educational institutions with critical constituents: government, faculty, alumni, donors/industry,</p><p>(b) moving to a system of greater financial self-sufficiency by raising fees to cover at least 20% of the operating budget and by generating more funds from research and endowments while establishing scholarships and education loans supported by the government for economically weaker sections,</p><p>(c) creating a single, independent regulator at arm’s length from players,</p><p>(d) allowing multiple independent accreditation agencies for quality assurance and encouraging ranking of institutions by the market (e.g., media, employer associations, student bodies) to enhance competition and transparency, and</p><p>(e) granting greater academic and administrative autonomy to educational institutions and matching it with higher accountability for results via global benchmarking.</p><p><strong>3. Role of the private sector: </strong>There is a cursory discussion on this in the National Education Policy. My view is that a country like India requires the amount of work that is both large and difficult. It is neither possible nor appropriate for the government alone to do it. Therefore, the role of the private sector is necessary and complementary. One must look at how telecommunication and aviation sectors were transformed in India with significant improvement in standards and customer satisfaction once the private sector was allowed. The government’s role should be to define the rules, ensure compliance, and step in when there is a market failure. It should unleash the private sector’s entrepreneurial spirit to rapidly address stakeholders’ changing demands to ensure relevance and raise the standards through competition. There is a need to build a partnership model for encouraging large private investments in education through the Government of India or state governments providing (a) land and infrastructure, (b) matching funds, (c)governance control by a private board of trustees (but subject to continuing review of standards and accreditation through an accreditation council or regulator). Also, provide value-added partnership to private players by making targeted interventions to prevent market failures in situations where courses are essential for nation-building but where market demand is low (e.g., Public Health).</p><p>Attracting private participation in education will require a conducive regulatory environment for both entry and operations. There should be transparent education legislation at the center, which applies uniformly across the country to domestic and foreign players. Such legislation should clearly articulate the scope and rules of the game. For example, while education provision should continue to be not-for-profit, the legislation should clearly state that private establishments are allowed to generate operating surplus for expansion and quality improvement and subcontract support services to for-profit private parties. The legislation should also lay down consistent policy around granting land for educational purposes. For example, the Morrill Act in the US created the foundation for the successful launch of 70 land-grant universities.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/709/1*0kRf2NWxqbjPY_Et-XAeqQ.png" /><figcaption>Private participation requires a conducive regulatory environment</figcaption></figure><p>In closing, what is encouraging about the new National Education Policy is the explicit call-out on the change in focus from “what to learn” to “how to learn” and from “passing the test” to “enjoyable learning.” Furthermore, it is commendable to commit to a culture of inclusivity and upholding respect, dignity, and privacy in administering education. As a nation, I am glad that we are making this shift in our education policy for a better tomorrow for India and the world.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5dd47eac7659" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>