The Story Begins…

The AT Protocol journey began in 2019, when Twitter (led by Jack Dorsey) funded Bluesky Social as an independent research project to focus on decentralization.

The objective was straightforward: to create an open standard for social media that would grant users portable identities and control over the algorithm shaping their feeds, by preventing platform lock-in.

The team, led by Jay Graber (CEO of Bluesky), developed the protocol and released it publicly as the AT Protocol in 2022. It now powers the Bluesky app, and serves as the foundation for a new, user-controlled social web.

Centralized vs Decentralized

• In centralized apps such as Instagram, TikTok, etc., one company controls everything: your posts, your account, and the rules. If the platform shuts down, your profile disappears with it.
• In decentralized apps such as Bluesky, your data is stored across multiple servers, giving you greater control and making it more difficult for any single entity to censor or delete your presence.

What’s AT protocol?

The Authenticated Transfer Protocol, a.k.a @ptorocol or (atproto) is a comprehensive framework designed to support various types of social applications, not just the microblogging platforms like Bluesky.

Think of it as a shared infrastructure that enables developers to build web apps such as:

• SkyFeed : a real-time custom feed builder
• Streamplace: an open-source livestreaming built on the atproto
• Leaflut.pub: a tool for creating and sharing long-form documents and blogs

The ultimate goal of this protocol is to bring everything together in one place, with a unified login, allowing users to share content without having to create a new account for every app.

Why it’s different from centralized protocols?

1. Ownership and Portability: users have the ability to move from one application’s data service to another without losing their their digital assets including their unique username, content or social metrics, etc.
2. Algorithmic freedom: instead of relying on one algorithm to dictate what appears on your feed, the AT protocol gives users the power to create their own custom feed, unlike centralized platforms like TikTok, Instagram..)
3. Custom feeds: you can design your feed display around any area of interest. If you are a fan of science, basketball, politics, psychology — you can tailor your feed accordingly.
4. Ad free: unlike centralized social media, atproto platforms are ad-free, allowing users to scroll and share content without seeing a single paid advertisement.

What are the challenges?

Decentralized platforms, like any other platform, face several challenges, especially when it comes to user adoption, security and content moderation:

1. User adoption: new users find the platform as less valuable than centralized apps due to a lower volume and reduced consistency of content related to their niche interests.
2. Security: operating an app comes with a considerable security cost. Managing your own server requires significant effort to ensure great security and users data protection.
3. App moderation: one of the most difficult challenges to Bluesky is keeping pace with moderation requests. Their moderation team is staffed 24/7 to review reports as quickly as they can, raising the need for significant resources to handle complex abuse cases.

Want to start your journey on Bluesky?

I’ve used Bluesky for a short time now. What immediately stands out is that it’s ad-free, as I mentioned earlier, and it delivers a wide range of topics that are aligned with what you care about.

Here are some tips on where to start:

Observe what’s going on

create an account, prepare your profile, and explore the app’s features.

Experiment Custom feeds

Start by looking for content categories you’re interested in. Once you have done this, categorize them and share your posts on those feeds. You will likely get more engagement when sharing content on themed feeds than on random ones.

Build your community

Focus on making meaningful connections (if you like to paint, connect with people interested in painting) and be curious about many topics.

Start sharing

The more your share, the more credibility and connections you can get. Your shared content could be your morning coffee, your daily quote, or even a simple post about the weather in your city.Use these post to engage with others about the that topic.

A personal take on Bluesky

Bluesky is constantly updating its platform with new functionalities, such as video uploading and constructive conversation prompts. These additions are making the app as intuitive as possible.

The company revealed: “we are seeing more people setting up small businesses and creating new apps and custom feeds directly on the AT protocol. “

For now, I see the platform as the clear leader of a new era of social media, actively challenging centralized giants like Instagram, TikTok, and Facebook.

While the goal is ambitious and user adoption remains a major challenge, the momentum is undeniable.

If you enjoyed reading this article, drop your thoughts in the comment section!

About this article

This article has been written by a student on the Grenoble Ecole de Management’s Advanced Masters in Digital Strategy Management. As part of a content creation assignment, students are given the task of writing articles based on their digital interests and disseminating the articles online. Articles are marked but we make minimal changes to the content. Thanks for reading! James Barisic, Programme Director, MS DSM.

Decentralization, AT Protocol and Bluesky 101 was originally published in Digital GEMs on Medium, where people are continuing the conversation by highlighting and responding to this story.

Powerful AI tools are having a major impact on search behavior and businesses need to realign their digital marketing strategy in response

With over 1 billion monthly active users in October 2025, AI-powered assistants are reshaping how users search for and process information about products and services. This shift raises many questions about how decisions are made, how AI operates when searching for information, and whether traditional search engines are keeping pace with these fast-moving changes in user behavior.

AI assistants are the new decision-makers

The Nuremberg Institute for Market Decisions study revealed that ChatGPT outperformed Google search in selecting the optimal theme park based on the given criteria, in both speed and accuracy.

• Higher Accuracy: ChatGPT users were significantly more likely to select the correct product or service. This advantage was particularly strong for complex searches, such as those involving theme parks, where relevant information was harder to find.

While 59% of ChatGPT users identified the best theme park based on the predefined criteria, only 39% of Google users did.

• Faster Decision-Marking: ChatGPT users turned out to be quicker in making a decision than Google Search users

→ It’s interesting because these AI systems rely on existing online datasets from big search engines like Google, Bing, Yahoo!, etc. Yet the search experience feels very different.

Tell me about the way you search since you started using generative AI tools (in the comment section)!

The way we search is shifting…

Until recently, the typical answer to any question was simple: “Just Google it”. This, led us most of the time, to opening many of those blue links, multiple tabs and getting lost in an ocean of information to find the most accurate one we need.

That linear click-to-search is changing.

Users now consider AI tools like ChatGPT, Gemini, Copilot as a primary option (sometimes mixed with Google Search) to help them get answers in a conversational manner — Instead of scrolling to the bottom of the page to find relevant information.

They want an option where they can get instant and quick responses.

Efficiency and time saving are what everyone is looking for in the business industry. They want to use their time in more productive task, rather than searching, evaluating sources and synthesizing information manually.

Why ?

The conversational format makes the process more intuitive, flexible, and user-friendly than traditional web search.

From keywords to conversations

The traditional keyword era is not fading, but is definitely being disrupted. Typing two or three isolated words no longer delivers meaningful results.

With generative AI, long-tail, questions-first queries now dominate because they allow users to engage in a 1:1 conversation until they find the most relevant information they are looking for, and more, in a short time.

Here’s the difference in action:

Traditional search --> Google
User search: content brief examples
Result: 10 different pages with different content types (blog, white paper...),
mainly focus on the topic as is and some websites offer templates

--------------------------------------------------------------------------

AI search --> Gemini
Query: Give 3-5 content brief examples
Result: 5 different examples: 
1) Blog Post Brief (SEO-Focused)
2) Email Newsletter Brief (Product Update)
3) Landing Page Brief (Lead Generation)
4) Social Media Video Brief (Explainer)

Not only does it search but it also plays a role in helping users identify what they are looking for when they don’t know what to type exactly. It has a term, and it’s called:

keyword foraging: a user conducts a preliminary search (usually in a web search engine like Google) to determine the right keywords for their information need.

How do they process the search query?

AI tools interpret intent, iterate on follow-up questions, and provide synthesized answers in real time, draw from multiple data sources

Here’s a very simplified process:

1. Trained on massive datasets: Generative AI models like ChatGPT, Gemini, Copilot, etc. are trained on diverse text sources: websites, FAQs, reviews, product descriptions, forums, and brand content (blogs, brochures, white-papers…)
2. Able to fetch up-to-date data: They gather the latest information the web can provide: prices, product updates, customer reviews and rankings.
3. Capable of interpreting user intent: they don’t match keywords only; AI goes beyond a query and searches its knowledge base for what fits best the user’s need (sometimes, it’s not accurate at all, as we all know!)
4. Skilled at comparing and synthesizing multiple data types: For example, once the comparative reasoning is done, the model can explain why a product might be better suited for a buyer’s needs: “This software offers a better module in data analysis”

In short, AI doesn’t just answer questions, it helps buyers reason, get accurate data, structure needs, compare options, and gain the clarity needed to make a sound decision.

Why it matters to your business?

Since the rise of AI, Search has became intensely personalized, concise and straight to the point. Information is available in fraction of a second, decision-making is accelerating, and demanding a new approach to engagement.

What are we talking about here?

Forrester shows that:

• 89% of B2B buyers have adopted generative AI tools (ChatGPT, Copilot, perplexity and those embedded in corporate softwares) in their buying process, making it “one of the top sources of self-guided information in every phase of their buying process.”
• 87% say that online content has a “major or moderate impact” on vendor selection.

If you’re not visible during the research phase, you will not be visible when real human interactions start!

What marketers should do to be cited?

The marketing goal has fundamentally shifted. We’ve moved from optimizing for a click to optimizing for AI selection. Since your future customers rely on AI-curated experiences, your content must now be the authoritative and accurate source that generative AI chooses to cite.

Therefore, your content must be optimized to:

• deliver direct, personalized answers
• show why buyers should choose your product or service (in seconds)
• clarify where exactly you fit in their workflow and at what cost

5 Strategies marketers should consider

Before generative AI, the goal was to get a click. Now, it’s about being the authoritative, accurate source that AI chooses to cite.

Here’s what marketing teams should implement to be cited:

1. Structure your content (blogs, landing pages, FAQs)
   LLMs process content not as continuous text but as a sequence of content. That’s why your content needs to be well organized and structured to analyze and extract precise information (same work done or that need must be done for traditional search engines algorithm).
2. Consider user intent and what makes your brand different
   AI systems prioritize understanding the user’s ultimate goal (intent) and respond by delivering one synthesized, authoritative answer. That’s why your content must include summaries, FAQs, be problem-solving oriented, and give straight to the point answers and not lengthy prose.
3. Put more work into your domain authority
   LLMs are more likely to cite your brand if your content is accurate, trustworthy, shares a clear message and when technical parameters are met.
4. Maintain accurate, up-to-date information
   Outdated information lowers credibility. A rapid content production and updates— supported by AI tools — keeps your ecosystem current and increases your authority signal inside LLMs.
5. Create semantic depth across topics
   Isolated articles no longer work. LLMs understand topics through related content such as: other articles, internal linking, and consistent terminology make your brand easier for AI to map and cite.

My take: it’s not the end of Google but the start of a new way of searching, analyzing, making decisions, and purchasing…

Generative AI is booming and moving faster than ever since 2022, making a huge leap in our personal and professional lives. That said, Google is still leading, by ranking at the top of the pyramid (and will continue, I guess)

I think companies should continue to work on their SEO, SEA, content creation, etc., and keep their eyes open to what’s coming next especially with AI mode and AI Overviews from Google.

If you enjoyed reading this article, drop your thoughts in the comment section!

About this article

This article has been written by a student on the Grenoble Ecole de Management’s Advanced Masters in Digital Strategy Management. As part of a content creation assignment, students are given the task of writing articles based on their digital interests and disseminating the articles online. Articles are marked but we make minimal changes to the content. Thanks for reading! James Barisic, Programme Director, MS DSM.

How AI is reshaping the search behavior was originally published in Digital GEMs on Medium, where people are continuing the conversation by highlighting and responding to this story.