<?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 Katya Tarapovskaia on Medium]]></title>
        <description><![CDATA[Stories by Katya Tarapovskaia on Medium]]></description>
        <link>https://medium.com/@katya_tarapovskaia?source=rss-35b9ab9537f1------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*1CngTBEns7aY4GSkeM6Bhg.png</url>
            <title>Stories by Katya Tarapovskaia on Medium</title>
            <link>https://medium.com/@katya_tarapovskaia?source=rss-35b9ab9537f1------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 00:24:43 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@katya_tarapovskaia/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[12 Behavioural Science Marketing Ideas. Where to Use Claude(Part 1)]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@katya_tarapovskaia/12-behavioural-science-marketing-ideas-where-to-use-claude-part-1-0f36a5d61b76?source=rss-35b9ab9537f1------2"><img src="https://cdn-images-1.medium.com/max/1456/0*tDWMnAmGMKxp2BAf.png" width="1456"></a></p><p class="medium-feed-snippet">This playbook is Part 1 of my series on 25 behavioural principles for ABM. In this part: the 12 principles that lower perceived risk, the&#x2026;</p><p class="medium-feed-link"><a href="https://medium.com/@katya_tarapovskaia/12-behavioural-science-marketing-ideas-where-to-use-claude-part-1-0f36a5d61b76?source=rss-35b9ab9537f1------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@katya_tarapovskaia/12-behavioural-science-marketing-ideas-where-to-use-claude-part-1-0f36a5d61b76?source=rss-35b9ab9537f1------2</link>
            <guid isPermaLink="false">https://medium.com/p/0f36a5d61b76</guid>
            <category><![CDATA[marketing]]></category>
            <category><![CDATA[b2b-marketing]]></category>
            <category><![CDATA[digital-marketing]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[ai-agent]]></category>
            <dc:creator><![CDATA[Katya Tarapovskaia]]></dc:creator>
            <pubDate>Fri, 13 Nov 2026 10:44:33 GMT</pubDate>
            <atom:updated>2026-11-13T10:44:33.276Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[7 AI Agents for GTM: Which One is Actually Worth Using?]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@katya_tarapovskaia/7-ai-agents-for-gtm-which-one-is-actually-worth-using-c27fd6090352?source=rss-35b9ab9537f1------2"><img src="https://cdn-images-1.medium.com/max/1504/1*JHuHYPTRtBX8FCJdwWsiEQ.png" width="1504"></a></p><p class="medium-feed-snippet">AI agents are everywhere right now.</p><p class="medium-feed-link"><a href="https://medium.com/@katya_tarapovskaia/7-ai-agents-for-gtm-which-one-is-actually-worth-using-c27fd6090352?source=rss-35b9ab9537f1------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@katya_tarapovskaia/7-ai-agents-for-gtm-which-one-is-actually-worth-using-c27fd6090352?source=rss-35b9ab9537f1------2</link>
            <guid isPermaLink="false">https://medium.com/p/c27fd6090352</guid>
            <category><![CDATA[claude]]></category>
            <category><![CDATA[digital-marketing]]></category>
            <category><![CDATA[ai-automation]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[ai-agent]]></category>
            <dc:creator><![CDATA[Katya Tarapovskaia]]></dc:creator>
            <pubDate>Tue, 03 Nov 2026 16:07:41 GMT</pubDate>
            <atom:updated>2026-11-03T16:07:41.284Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Your Marketing Strategy Is Only as Good as Your Account Segmentation]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@katya_tarapovskaia/your-marketing-strategy-is-only-as-good-as-your-account-segmentation-674522087e43?source=rss-35b9ab9537f1------2"><img src="https://cdn-images-1.medium.com/max/1916/1*6U1OgHTC5rZOHp-X7SVYAw.png" width="1916"></a></p><p class="medium-feed-snippet">An account fits your ideal customer profile. It shows relevant activity. Marketing adds it to a campaign.</p><p class="medium-feed-link"><a href="https://medium.com/@katya_tarapovskaia/your-marketing-strategy-is-only-as-good-as-your-account-segmentation-674522087e43?source=rss-35b9ab9537f1------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@katya_tarapovskaia/your-marketing-strategy-is-only-as-good-as-your-account-segmentation-674522087e43?source=rss-35b9ab9537f1------2</link>
            <guid isPermaLink="false">https://medium.com/p/674522087e43</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[claude]]></category>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[marketing]]></category>
            <category><![CDATA[digital-marketing]]></category>
            <dc:creator><![CDATA[Katya Tarapovskaia]]></dc:creator>
            <pubDate>Mon, 26 Oct 2026 17:11:01 GMT</pubDate>
            <atom:updated>2026-10-26T17:11:01.900Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[How to Build Your AI Marketing OS With Claude and HubSpot]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@katya_tarapovskaia/how-to-build-your-ai-marketing-os-with-claude-and-hubspot-9c32d51a50db?source=rss-35b9ab9537f1------2"><img src="https://cdn-images-1.medium.com/max/1456/0*TGdmQjLjJwp9V27L.png" width="1456"></a></p><p class="medium-feed-snippet">Everyone wants AI agents.</p><p class="medium-feed-link"><a href="https://medium.com/@katya_tarapovskaia/how-to-build-your-ai-marketing-os-with-claude-and-hubspot-9c32d51a50db?source=rss-35b9ab9537f1------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@katya_tarapovskaia/how-to-build-your-ai-marketing-os-with-claude-and-hubspot-9c32d51a50db?source=rss-35b9ab9537f1------2</link>
            <guid isPermaLink="false">https://medium.com/p/9c32d51a50db</guid>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[digital-marketing]]></category>
            <category><![CDATA[marketing]]></category>
            <category><![CDATA[marketing-automation]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Katya Tarapovskaia]]></dc:creator>
            <pubDate>Fri, 23 Oct 2026 15:06:01 GMT</pubDate>
            <atom:updated>2026-10-23T15:06:01.741Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[ChatGPT vs Claude vs Gemini for Marketing: Which AI Assistant Should You Use for What?]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@katya_tarapovskaia/chatgpt-vs-claude-vs-gemini-for-marketing-which-ai-assistant-should-you-use-for-what-4d0686048db2?source=rss-35b9ab9537f1------2"><img src="https://cdn-images-1.medium.com/max/1672/1*Ui8mmIaHs88FK6845um6Tg.png" width="1672"></a></p><p class="medium-feed-snippet">Most marketing teams are asking the wrong question.</p><p class="medium-feed-link"><a href="https://medium.com/@katya_tarapovskaia/chatgpt-vs-claude-vs-gemini-for-marketing-which-ai-assistant-should-you-use-for-what-4d0686048db2?source=rss-35b9ab9537f1------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@katya_tarapovskaia/chatgpt-vs-claude-vs-gemini-for-marketing-which-ai-assistant-should-you-use-for-what-4d0686048db2?source=rss-35b9ab9537f1------2</link>
            <guid isPermaLink="false">https://medium.com/p/4d0686048db2</guid>
            <category><![CDATA[productivity]]></category>
            <category><![CDATA[marketing-strategies]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[marketing]]></category>
            <category><![CDATA[ai-agent]]></category>
            <dc:creator><![CDATA[Katya Tarapovskaia]]></dc:creator>
            <pubDate>Tue, 13 Oct 2026 14:58:50 GMT</pubDate>
            <atom:updated>2026-10-13T14:58:50.342Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[The ABM Manager Is Becoming an Impact Driver in B2B]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@katya_tarapovskaia/the-abm-manager-is-becoming-an-impact-driver-in-b2b-9a1395a99369?source=rss-35b9ab9537f1------2"><img src="https://cdn-images-1.medium.com/max/1894/1*_WFBdMCduhX6xqb7dwfI4Q.png" width="1894"></a></p><p class="medium-feed-snippet">The ABM Manager&#x2019;s job title has barely changed.</p><p class="medium-feed-link"><a href="https://medium.com/@katya_tarapovskaia/the-abm-manager-is-becoming-an-impact-driver-in-b2b-9a1395a99369?source=rss-35b9ab9537f1------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@katya_tarapovskaia/the-abm-manager-is-becoming-an-impact-driver-in-b2b-9a1395a99369?source=rss-35b9ab9537f1------2</link>
            <guid isPermaLink="false">https://medium.com/p/9a1395a99369</guid>
            <category><![CDATA[marketing]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[b2b-marketing]]></category>
            <category><![CDATA[digital-marketing]]></category>
            <category><![CDATA[ai-agent]]></category>
            <dc:creator><![CDATA[Katya Tarapovskaia]]></dc:creator>
            <pubDate>Wed, 30 Sep 2026 14:54:19 GMT</pubDate>
            <atom:updated>2026-09-30T14:54:19.329Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[The Experience-led B2B Marketing Playbook and Where AI Actually Fits]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@katya_tarapovskaia/the-experience-led-b2b-marketing-playbook-and-where-ai-actually-fits-cec821848532?source=rss-35b9ab9537f1------2"><img src="https://cdn-images-1.medium.com/max/1308/0*WCsSIRh9_gPfBKBZ.png" width="1308"></a></p><p class="medium-feed-snippet">AI has made it easier than ever to create more campaigns. The competitive advantage now is knowing when to stop automating and create an&#x2026;</p><p class="medium-feed-link"><a href="https://medium.com/@katya_tarapovskaia/the-experience-led-b2b-marketing-playbook-and-where-ai-actually-fits-cec821848532?source=rss-35b9ab9537f1------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@katya_tarapovskaia/the-experience-led-b2b-marketing-playbook-and-where-ai-actually-fits-cec821848532?source=rss-35b9ab9537f1------2</link>
            <guid isPermaLink="false">https://medium.com/p/cec821848532</guid>
            <category><![CDATA[b2b-marketing]]></category>
            <category><![CDATA[digital-marketing]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[marketing]]></category>
            <dc:creator><![CDATA[Katya Tarapovskaia]]></dc:creator>
            <pubDate>Fri, 25 Sep 2026 10:46:24 GMT</pubDate>
            <atom:updated>2026-09-25T10:46:24.452Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[How to Build a Low-Budget Micro Marketing Campaign with AI That Converts at 30%]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@katya_tarapovskaia/how-to-build-a-low-budget-micro-marketing-campaign-with-ai-that-converts-at-30-6f6182fdf7b0?source=rss-35b9ab9537f1------2"><img src="https://cdn-images-1.medium.com/max/1188/1*wYz1-UWP8skJH6wEhNHdqQ.png" width="1188"></a></p><p class="medium-feed-snippet">The most effective ABM play I&#x2019;ve tested this year wasn&#x2019;t a webinar, paid campaign or conference sponsorship.</p><p class="medium-feed-link"><a href="https://medium.com/@katya_tarapovskaia/how-to-build-a-low-budget-micro-marketing-campaign-with-ai-that-converts-at-30-6f6182fdf7b0?source=rss-35b9ab9537f1------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@katya_tarapovskaia/how-to-build-a-low-budget-micro-marketing-campaign-with-ai-that-converts-at-30-6f6182fdf7b0?source=rss-35b9ab9537f1------2</link>
            <guid isPermaLink="false">https://medium.com/p/6f6182fdf7b0</guid>
            <category><![CDATA[sales]]></category>
            <category><![CDATA[b2b-marketing]]></category>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[marketing]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Katya Tarapovskaia]]></dc:creator>
            <pubDate>Thu, 17 Sep 2026 14:31:01 GMT</pubDate>
            <atom:updated>2026-09-17T14:31:01.755Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[How to Use Claude to Turn Sales Conversations into Targeted Campaigns]]></title>
            <link>https://medium.com/@katya_tarapovskaia/how-to-use-claude-to-turn-sales-conversations-into-targeted-campaigns-a299a137fce1?source=rss-35b9ab9537f1------2</link>
            <guid isPermaLink="false">https://medium.com/p/a299a137fce1</guid>
            <category><![CDATA[claude]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[b2b-marketing]]></category>
            <category><![CDATA[marketing]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Katya Tarapovskaia]]></dc:creator>
            <pubDate>Fri, 28 Aug 2026 14:14:45 GMT</pubDate>
            <atom:updated>2026-08-28T14:14:45.456Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*K-OeAxVgg1DeDqboCB2dTw.png" /><figcaption>www.youstellar.com</figcaption></figure><p>A few months ago, I looked at a quarterly campaign plan one of my clients had built.</p><p>Twelve campaigns were planned for Q3. The topics came from keyword research, competitor analysis and a brainstorming session from February.</p><p>There was one problem.</p><p><strong>Not a single campaign addressed the objection that had killed four enterprise deals that quarter.</strong></p><p>I knew because I’d read the sales call transcripts.</p><p>The same concern appeared in every closed-lost debrief: buyers were worried that implementation would disrupt their existing workflows.</p><p>Four deals. Same blocker. Zero content addressing it.</p><p>This isn’t unusual.</p><p>Sales teams hear why deals stall every week. That intelligence sits in call recordings, CRM notes and sales conversations — disconnected from the people deciding what marketing should build next.</p><p>The problem isn’t a lack of insight.</p><p><strong>It’s the gap between where buyer intelligence lives and where campaigns get created.</strong></p><p>I’ve been testing a different approach: connecting those signals to Claude through MCP, then using Claude as the reasoning layer across the stack.</p><p>Here’s the 5-step system.</p><h3>1. Capture: Turn Sales Calls into Buyer Signals</h3><p>Most teams capture what buyers say.</p><p>I want to know what they <strong>ask</strong>.</p><p>Questions, objections, hesitations and past failures are often better campaign inputs than keyword research.</p><p>I look for four types of signals:</p><ul><li><strong>Unsolved pains</strong> → awareness content</li><li><strong>Hesitations about switching</strong> → comparison and risk-mitigation content</li><li><strong>Past failures</strong> → differentiation content</li><li><strong>Internal blockers</strong> → sales enablement content</li></ul><p>With <a href="https://www.granola.ai/">Granola AI </a>notetaker capturing call notes and syncing them to HubSpot, I can ask Claude:</p><blockquote><em>“Look at my call notes from the past 30 days. List the questions buyers asked during discovery and Q&amp;A. Group them by theme and show me which appeared more than three times.”</em></blockquote><p>Instead of one rep remembering one objection, I get a pattern across the pipeline.</p><h3>2. Understand: Add Buyer Intelligence</h3><p>Knowing <strong>what</strong> a buyer said isn’t enough.</p><p>You also need to understand <strong>who</strong> said it.</p><p>The same objection can require very different messaging depending on the person raising it.</p><p>For example, a CFO concerned about pricing may need an ROI model. A procurement leader may need a TCO comparison and risk-mitigation proof.</p><p>That’s where buyer intelligence becomes useful.</p><p>I can combine call insights with data from tools such as <a href="https://humantic.ai/">Humantic AI</a>, and built-in Bombora and Apollo intent data to understand personality, intent and firmographic context.</p><p>Then Claude can turn:</p><blockquote><em>“This account has a pricing objection.”</em></blockquote><p>into something more actionable:</p><blockquote><em>“This account is actively researching the category, the VP of Marketing values peer validation, and they’ve raised implementation concerns twice.”</em></blockquote><p>Now we’re building campaigns around <strong>real buyers</strong>, not assumed personas.</p><h3>3. Analyse: Find the Problems Actually Costing Revenue</h3><p>This is where the system gets interesting.</p><p>Most teams rank objections by frequency.</p><p>That’s not necessarily the right metric.</p><p>A question might appear in 40% of calls but rarely stop a deal.</p><p>Another might appear in 15% of calls and kill 30% of your enterprise pipeline.</p><p>So I ask Claude to rank objections by <strong>revenue impact</strong>, not frequency:</p><blockquote><em>“Look at my deals from the last two quarters. For each objection theme, calculate closed-won vs. closed-lost rate, total revenue lost and average deal cycle. Rank the themes by revenue blocked.”</em></blockquote><p>Now marketing can answer a much better question:</p><p><strong>Which buyer problem should we solve first?</strong></p><p>The output usually narrows the list to a few high-impact themes.</p><p>Those become campaign priorities.</p><h3>4. Orchestrate: Turn Buyer Problems into Campaigns</h3><p>Now we have:</p><ul><li>recurring buyer objections</li><li>revenue impact</li><li>funnel stage</li><li>account intent</li><li>buying committee intelligence</li></ul><p>The next step is turning those insights into action.</p><p>This is where <a href="https://www.clay.com/">Clay </a>can connect the dots between research, enrichment and workflow automation.</p><p>For example:</p><p>A call surfaces an implementation objection → the account meets your ICP criteria → Clay enriches the buying committee → buyer intelligence is added → Claude creates a campaign brief around the specific objection.</p><p>I can ask:</p><blockquote><em>“For the top three revenue-blocking themes, find active pipeline accounts at the evaluation stage. Identify the relevant buying committee contacts and create a campaign brief for each theme, including messaging angles, content format and recommended channels.”</em></blockquote><p>The important part isn’t that Claude writes the brief.</p><p><strong>It’s what the brief is based on.</strong></p><p>It’s grounded in what buyers actually said, the deals affected and the people involved.</p><h3>5. Activate: Reach the People Who Raised the Objection</h3><p>The final step is getting the message to the right people.</p><p>Traditional ABM (account-based marketing) often tells you that <strong>Company X</strong> engaged.</p><p>Contact-level activation lets you get closer to:</p><p><strong>Jane, VP Marketing, at Company X engaged with the content addressing the implementation concern she raised on the sales call.</strong></p><p>With tools such as<a href="https://www.influ2.com/"> Influ2</a>, the campaign can target named buying committee members rather than treating the account as one anonymous audience.</p><p>The activation prompt can be as simple as:</p><blockquote><em>“For active pipeline accounts where implementation timeline is the primary objection, identify the relevant buying committee contacts and create a campaign brief using the case study addressing that concern.”</em></blockquote><p>Now the loop is closed.</p><p><strong>Buyer signal → insight → revenue impact → campaign → named contact → engagement.</strong></p><h3>The Prompt That Brings It Together</h3><p>Once the connections are live, I can ask Claude to run the analysis across the stack:</p><blockquote><em>“Review my call notes and CRM deal data from the past 30 days. Identify the top three objection themes by revenue impact. For each, show affected accounts, deal stage, buying committee profiles and the best content format to address the objection. Then create a campaign brief with messaging angles, channels and target contacts.”</em></blockquote><p>That’s the real shift.</p><p>Claude isn’t just helping the marketing team <strong>produce content faster</strong>.</p><p>It’s helping answer a more strategic question:</p><p><strong>Given what our buyers are telling us, which campaign has the best chance of moving revenue right now?</strong></p><h3>You Don’t Need the Full Stack</h3><p>You can start with three things:</p><p><strong>Claude + your CRM + sales call intelligence/notetaker.</strong></p><p>Connect the data. Start tagging recurring objections. After a few weeks, analyse the patterns.</p><p>Then add enrichment, intent, buyer intelligence and activation as the workflow matures.</p><p>You don’t need to replace every existing marketing process either.</p><p>Brand and thought leadership content still have a place.</p><p>But campaigns designed to influence pipeline should have a different starting point:</p><p><strong>buyer intelligence.</strong></p><h3>Conclusion</h3><p>Your sales calls contain some of your most valuable marketing research.</p><p>Your CRM contains the commercial context.</p><p>Your buyer intelligence tools add the “who” and “why.”</p><p>Claude connects the dots.</p><p>The result isn’t simply faster content production.</p><p>It’s a marketing system where campaigns start with <strong>what buyers actually said</strong>, are prioritised by <strong>what is costing you revenue</strong>, and reach the <strong>people who need to hear the message</strong>.</p><p>The best campaign idea might not be in your next brainstorming session.</p><p>It might already be sitting in yesterday’s sales calls.</p><p>If you’re learning Claude and AI in Marketing, or building with AI and stuck somewhere, get in touch.</p><ul><li><strong>DM me on LinkedIn</strong> → <a href="https://www.linkedin.com/in/katya-tarapovskaia/">linkedin.com/in/katya-tarapovskaia</a></li><li><strong>Or email at </strong>katya@youstellar.com</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a299a137fce1" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Connect 24 MCPs to Claude. Automate Your Marketing Stack in Under an Hour.]]></title>
            <link>https://medium.com/@katya_tarapovskaia/connect-24-mcps-to-claude-automate-your-marketing-stack-in-under-an-hour-e9b81b271722?source=rss-35b9ab9537f1------2</link>
            <guid isPermaLink="false">https://medium.com/p/e9b81b271722</guid>
            <category><![CDATA[claude]]></category>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[marketing]]></category>
            <category><![CDATA[automation]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Katya Tarapovskaia]]></dc:creator>
            <pubDate>Fri, 21 Aug 2026 11:27:04 GMT</pubDate>
            <atom:updated>2026-08-21T11:27:04.940Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*9qv_oeiS1tF5DgmlGplz1A.png" /><figcaption>www.youstellar.com</figcaption></figure><p>Most ABM and GTM teams don’t have a data problem.</p><p>They have a <strong>context and automation problem</strong>.</p><p>Intent data sits in one platform. CRM information lives somewhere else. LinkedIn engagement is disconnected from outreach activity, campaign reporting and sales follow-up.</p><p>The result? <strong>Marketers spend too much time moving data between tools and not enough time acting on it.</strong></p><p><strong><em>Model Context Protocol (MCP)</em></strong> offers a different approach.</p><p>MCPs connect Claude to the systems your team already uses, allowing it to read data, write data and run actions through approved tools. Instead of treating each platform as an isolated dashboard, Claude can help connect the signals across your stack.</p><p>That creates a more coordinated ABM workflow:</p><p><strong>Find the signal → Understand the buyer → Enrich the account → Prioritise the opportunity → Activate the next best action</strong></p><p>This article covers the MCP landscape I’ve mapped for ABM teams, how MCPs work in Claude, and how to extend them with Skills and scheduled routines.</p><h3>What is an MCP?</h3><p>An MCP is a connection between Claude and an external tool, platform or data source.</p><p>It gives Claude access to capabilities that are not available from the model alone. Depending on the integration and permissions, an MCP may allow Claude to:</p><ul><li>Read CRM, intent, campaign or analytics data.</li><li>Search records, documents or account information.</li><li>Create or update records.</li><li>Run workflows or other actions.</li><li>Combine information from multiple connected systems.</li></ul><p>Without an MCP, you might need to export data from several platforms, combine it manually and then ask Claude to analyse it.</p><p>With MCPs connected, you can ask Claude questions such as:</p><blockquote><em>“Which target accounts showed intent in the last seven days, engaged with our campaigns, and have an open opportunity in the CRM?”</em></blockquote><p>The value is not simply faster analysis. The value is <strong><em>connected context</em></strong><em>.</em></p><p>Claude can reason across the information required to make a marketing decision instead of analysing one isolated data source at a time.</p><p>Thanks for reading Katya Tarapovskaia’s Substack! Subscribe for free to receive new posts and support my work.</p><h3>MCP is not the same as an API</h3><p>An API allows software to communicate with another system.</p><p>An MCP provides a standard way for AI applications to discover and use tools exposed by that system.</p><p>You can think of an MCP as an AI-ready access layer between Claude (or another AI) and your technology stack.</p><h3>The final ABM MCP stack</h3><p>I’ve mapped <strong>24 MCPs and MCP-connected tools</strong> across the ABM workflow.</p><p>The stack covers account intelligence, outreach, orchestration, content, analytics, CRM and paid media.</p><p><strong>1. ABM and intent</strong></p><ul><li>Apollo.</li><li>Trigify.</li><li>Humantic AI.</li><li>ZenABM.</li><li>6Sense</li></ul><p><strong>2. Outreach</strong></p><ul><li>Smartlead.ai.</li><li>La Growth Machine.</li><li>HeyReach.</li><li>Instantly</li></ul><p><strong>3. Orchestration</strong></p><ul><li>Clay.</li><li>HubSpot.</li></ul><p><strong>4. Content</strong></p><ul><li>Google Drive.</li><li>Notion.</li><li>Figma.</li><li>Canva.</li></ul><p><strong>5. Analytics and intelligence</strong></p><ul><li>BigQuery.</li><li>Ahrefs.</li><li>Supermetrics.</li><li>Waldo.</li></ul><p><strong>6. CRM</strong></p><ul><li>Attio.</li><li>Pipedrive.</li><li>Salesforce.</li></ul><p><strong>7. Paid media</strong></p><ul><li>Windsor.ai.</li><li>Meta.</li></ul><p>This is not a recommendation to connect all 24 tools on day one.</p><p>It is a map of the types of systems that can contribute to an AI-assisted ABM workflow.</p><p>The right stack depends on your existing architecture, data quality, permissions, budget and operating model.</p><h3>What each layer contributes?</h3><ol><li><strong>ABM and intent</strong></li></ol><p>These tools help identify accounts, people and signals that may indicate buying activity.</p><p>For example:</p><ul><li>Apollo can support account and contact research.</li><li>Trigify can surface relevant LinkedIn activity and engagement signals.</li><li>Humantic AI can provide buyer intelligence and communication context.</li><li>ZenABM can connect account-level engagement, intent and ABM campaign data.</li></ul><p>The objective is to move beyond a static target-account list and understand where attention may be warranted now.</p><p><strong>2. Outreach</strong></p><p>Outreach MCPs connect signals and research to execution.</p><p>They can help teams:</p><ul><li>Build outreach lists.</li><li>Personalise messages.</li><li>Coordinate LinkedIn and email sequences.</li><li>Monitor replies and engagement.</li><li>Route promising conversations to sales.</li></ul><p>The important distinction is between generating a message and actually sending one. Sending messages, adding contacts to sequences or changing campaign settings should require explicit approval and clear governance.</p><p><strong>3. Orchestration</strong></p><p>Clay and HubSpot sit close to the centre of the workflow.</p><p>Clay can support enrichment, research and multi-step workflows. HubSpot can provide CRM, marketing and pipeline context.</p><p>Together, they can help Claude connect:</p><p><strong>Account data → Enrichment → Campaign activity → Pipeline context → Next action</strong></p><p><strong>4. Content</strong></p><p>Google Drive, Notion, Figma and Canva provide access to the content and creative layer.</p><p>Claude can use these tools to:</p><ul><li>Find relevant case studies.</li><li>Review campaign assets.</li><li>Draft content briefs.</li><li>Check brand guidance.</li><li>Create editorial plans.</li><li>Repurpose approved content for specific accounts or segments.</li></ul><p>The best use case is not creating more content for its own sake. It is helping teams find and apply the right content at the right stage of the buying journey.</p><p><strong>5. Analytics and intelligence</strong></p><p>BigQuery, Ahrefs, Supermetrics and Waldo provide analytical and market context.</p><p>They can support:</p><ul><li>Website and campaign analysis.</li><li>SEO and keyword research.</li><li>Cross-channel reporting.</li><li>Competitive intelligence.</li><li>Market and messaging monitoring.</li></ul><p>This layer helps answer whether activity is creating meaningful movement — not just generating clicks or impressions.</p><p><strong>6. CRM</strong></p><p>Attio, Pipedrive and Salesforce represent different CRM environments, but the principle is the same: connect AI to the system where account, contact and opportunity context is managed.</p><p>Claude can help identify:</p><ul><li>At-risk opportunities.</li><li>Accounts with renewed engagement.</li><li>Missing or inconsistent data.</li><li>Pipeline influenced by marketing activity.</li><li>Follow-up tasks that need attention.</li></ul><p><strong>7. Paid media</strong></p><p>Windsor.ai and Meta can connect paid-media performance and campaign activity to the wider ABM workflow.</p><p>This opens possibilities such as:</p><ul><li>Comparing campaign performance across channels.</li><li>Identifying accounts exposed to or engaging with campaigns.</li><li>Finding inefficient spend.</li><li>Recommending audience or creative changes.</li><li>Connecting media activity with account and pipeline outcomes.</li></ul><p>Any workflow capable of changing budgets, pausing campaigns or launching ads should be treated as a controlled action — not an automatic default.</p><h3>Four ways to connect tools to Claude</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*b7L-jA9nx1-DRMN3.jpeg" /><figcaption>www.youstellar.com</figcaption></figure><p>MCP is one way to connect Claude to external systems, but it is not the only one.</p><p><strong>1. Official Claude connector</strong></p><p>An official connector is supported directly through Claude’s connector or integration experience.</p><p>It generally provides the simplest setup and a more familiar authentication process.</p><p>Use it when:</p><ul><li>The vendor officially supports Claude.</li><li>You want a managed connection.</li><li>You need a straightforward setup for non-technical users.</li></ul><p>The limitation is that functionality depends on what the vendor has built into its connector.</p><p><strong>2. Custom MCP connector</strong></p><p>A custom MCP connects Claude to an MCP server that may not appear in Claude’s directory.</p><p>This can provide similar capabilities — reading data, writing data and running actions — but the server URL may need to be added manually.</p><p>Use it when:</p><ul><li>The vendor provides an MCP server.</li><li>The integration is not yet listed in Claude’s directory.</li><li>You need access to a specialised tool.</li></ul><p>The main consideration is trust. Teams should verify the server, permissions, authentication method and data handling before connecting it.</p><p><strong>3. Command-line interface</strong></p><p>Some tools expose a CLI rather than an MCP.</p><p>Claude Code can use the CLI to operate the tool through the terminal.</p><p>This is useful for:</p><ul><li>Technical workflows.</li><li>Repeatable commands.</li><li>Local files and scripts.</li><li>Tools that do not yet have an MCP.</li></ul><p>The limitation is that it generally requires Claude Code or terminal access and may be less accessible for marketing teams working primarily in Claude’s chat or browser experience.</p><p><strong>4. Direct API connection</strong></p><p>You can connect Claude directly to an API or build your own MCP around an API.</p><p>This provides more control, but it usually requires more technical setup.</p><p>It may involve:</p><ul><li>API authentication.</li><li>Data mapping.</li><li>Error handling.</li><li>Rate limits.</li><li>Permission management.</li><li>Maintenance when the API changes.</li></ul><p>For most marketing teams, an official or vendor-maintained MCP is preferable when one is available.</p><h3>MCPs, Skills and routines</h3><p>Connecting an MCP gives Claude access to tools.</p><p>It does not automatically give Claude your team’s operating model.</p><p>That is where <strong>Skills</strong> become useful.</p><h3>What is a Skill?</h3><p>A Skill is a reusable set of instructions, rules and workflows that tells Claude how to use a tool or complete a task.</p><p>A Skill can define:</p><ul><li>Which MCPs to use.</li><li>Which fields to retrieve.</li><li>How to interpret the data.</li><li>What thresholds to apply.</li><li>What format to use for the output.</li><li>What actions require approval.</li><li>Where the result should be saved or shared.</li></ul><p>For example, a generic HubSpot MCP gives Claude access to CRM records.</p><p>An ABM pipeline-review Skill could tell Claude:</p><ol><li>Retrieve open opportunities.</li><li>Check account engagement.</li><li>Compare activity with the previous 14 days.</li><li>Flag accounts with declining engagement.</li><li>Exclude accounts with a closed-lost status.</li><li>Recommend a next action.</li><li>Never update a CRM record without approval.</li><li>Format the output as a sales briefing.</li></ol><p>The MCP provides the capability.</p><p>The Skill provides the method.</p><h3>Why Skills matter for marketing teams?</h3><p>Without Skills, users may ask similar questions in different ways and receive inconsistent outputs.</p><p>With Skills, the organisation can standardise how recurring work is completed.</p><p><strong>Examples include:</strong></p><ul><li>Weekly ABM account review.</li><li>Intent-to-outreach workflow.</li><li>Campaign performance analysis.</li><li>Competitive intelligence briefing.</li><li>Sales handoff document.</li><li>Content recommendation workflow.</li><li>Pipeline-attribution review.</li></ul><p>Skills help turn individual prompts into repeatable operating processes.</p><h3>What does it mean to stack MCPs?</h3><p>Stacking MCPs means using multiple MCPs in the same Claude session or workflow.</p><p>For example:</p><ul><li>Trigify identifies a LinkedIn signal.</li><li>Humantic AI adds buyer context.</li><li>Clay enriches the account.</li><li>ZenABM checks account engagement.</li><li>HubSpot provides pipeline status.</li><li>La Growth Machine prepares an outreach sequence.</li><li>Notion stores the account brief.</li><li>Slack alerts the account owner.</li></ul><p>The benefit is that Claude can connect multiple steps without requiring the user to manually copy information between platforms.</p><p>However, more MCPs do not always mean better results.</p><p>Too many tools can create:</p><ul><li>Ambiguous source-of-truth issues.</li><li>Conflicting account or contact records.</li><li>Slower workflows.</li><li>Higher context usage.</li><li>Unclear permissions.</li><li>Greater risk of accidental actions.</li></ul><p>A strong stack is usually modular:</p><p><strong>Signal layer → Intelligence layer → System of record → Activation layer → Reporting layer</strong></p><h3>How routines work</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Qn4EkHEbcVHHAY7F.png" /><figcaption>www.youstellar.com</figcaption></figure><p>A routine is a repeatable workflow that runs on a schedule.</p><p>Instead of asking Claude to perform the same task every Monday, you can create a workflow that uses MCPs on a recurring basis.</p><p>Examples:</p><ul><li>A daily hot-account alert.</li><li>A weekly ABM report.</li><li>A monthly campaign review.</li><li>A Friday pipeline and engagement reconciliation.</li><li>A scheduled competitive-intelligence briefing.</li></ul><p>A routine generally combines:</p><ol><li>A trigger or schedule.</li><li>A prompt or Skill.</li><li>One or more MCPs.</li><li>An output destination.</li><li>Approval rules.</li></ol><p>For example:</p><pre>Every Monday at 9:00 AM:<br><br>Use ZenABM, Trigify and HubSpot to identify target accounts with new intent <br>or engagement in the previous seven days.</pre><pre>Create a report containing:<br>- Top 10 priority accounts.<br>- New signals.<br>- Current CRM stage.<br>- Account owner.<br>- Recommended next action.<br>- Data gaps.</pre><pre>Save the report to the ABM Weekly Reports page in Notion.</pre><pre>Do not send messages, update CRM records or change campaigns automatically.</pre><p>A more advanced routine could prepare a Slack notification, but it should still use an approval step before posting externally.</p><h3>The difference between a prompt and a routine</h3><p>A prompt is something you ask Claude to do now.</p><p>A routine is something Claude is configured to do repeatedly.</p><p>A Skill defines how the work should be performed.</p><p>Together, they create a reusable operating process:</p><p><strong>MCP = access</strong><br><strong>Skill = method</strong><br><strong>Prompt = instruction</strong><br><strong>Routine = repeatability</strong></p><h3>Start read-only</h3><p>The most important implementation principle is to separate <strong>insight</strong> from <strong>action</strong>.</p><p>Begin with read-only workflows such as:</p><ul><li>Account research.</li><li>Campaign analysis.</li><li>Pipeline review.</li><li>Content recommendations.</li><li>Competitive monitoring.</li><li>Weekly reporting.</li></ul><p>Only enable write actions once you have tested:</p><ul><li>Data accuracy.</li><li>Account matching.</li><li>Duplicate handling.</li><li>Permission boundaries.</li><li>Approval steps.</li><li>Audit logging.</li></ul><p>Actions that should normally require approval include:</p><ul><li>Sending an email or LinkedIn message.</li><li>Adding a contact to a sequence.</li><li>Updating a CRM record.</li><li>Creating or assigning a task.</li><li>Posting to Slack.</li><li>Changing ad budgets.</li><li>Pausing or launching campaigns.</li></ul><p>The more consequential the action, the stronger the approval process should be.</p><p>Thanks for reading Katya Tarapovskaia’s Substack! Subscribe for free to receive new posts and support my work.</p><h3>A practical rollout plan</h3><p>You do not need to connect all 24 tools at once.</p><h3>Phase 1: Establish the foundation</h3><p>Start with:</p><ul><li>One CRM.</li><li>One account or intent source.</li><li>One reporting destination.</li></ul><p>For example:</p><p><strong>HubSpot + ZenABM + Notion</strong></p><h3>Phase 2: Add context</h3><p>Connect:</p><ul><li>Trigify.</li><li>Humantic AI.</li><li>Clay.</li><li>BigQuery or Supermetrics.</li></ul><p>This gives Claude more context for prioritisation and analysis.</p><h3>Phase 3: Add activation</h3><p>Connect:</p><ul><li>La Growth Machine or Smartlead.ai or Instantly</li><li>HeyReach.</li><li>Meta.</li></ul><p>Keep actions approval-based while you test the workflows.</p><h3>Phase 4: Add repeatability</h3><p>Create:</p><ul><li>ABM review Skills.</li><li>Account-research Skills.</li><li>Sales-handoff Skills.</li><li>Weekly reporting routines.</li><li>Competitive-monitoring routines.</li></ul><p><strong><em>This is where MCPs move from individual experiments to team processes.</em></strong></p><p>If you’re learning Claude and AI in Marketing, or building with AI and stuck somewhere, get in touch.</p><ul><li><strong>DM me on LinkedIn</strong> → <a href="https://www.linkedin.com/in/katya-tarapovskaia/">linkedin.com/in/katya-tarapovskaia</a></li><li><strong>Or email at </strong>katya@youstellar.com</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e9b81b271722" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>