MyAgentStack.io Is Live. The Map of Every AI Agent in PR, Marketing, and Comms.
I built it this weekend because the agent layer is changing your business faster than anyone is tracking it. Bookmark this. Send it to your team. Submit yours.
Hi Team PR@ctical.
I don’t know what you did this weekend. I had bronchitis. So I rested, and I built. And in the middle of the building, it hit me: I have made SO much over the last twelve months that I have done a bad job of reminding you what already exists. You can’t use a hub I never linked to. You can’t bookmark a glossary I never named. You can’t send your CMO to a research desk you never knew was there.
So before I tell you about the new thing, let me tell you about the things that have been waiting for you this whole time.
Then I’ll tell you about MyAgentStack.io, which is the new thing, and which I think is going to change how you build your AI workflow this quarter.
🏠 The AskSarah.AI Hub. Bookmark Every Link.
Five things live, working, and updated by me (with a lot of AI and a lot of human sweat) every week:
Daily AI Briefings. Hand-curated, sourced, ready by 5 AM. The one I open before my coffee.
NEW: The Research Hub. Every stat, study, and report I trust, in one place. Stop sourcing from scratch. Stop quoting research you can’t verify.
Live Show Recaps. Two shows now: The Visibility Equation and 6MOA. The replays are where the real receipts live.
NEW: The AI Glossary. New terms added weekly. Last week’s word was Tokenmaxxing. This week’s, you’ll see in a minute. (Spoiler: B2AI, courtesy of Visa.)
The Learn Hub. Hours of structured guides for CMOs, founders, and comms leaders trying to catch up.
Now, the new thing.
🚀 BUILDING & LAUNCHING: MyAgentStack.io
What it is
MyAgentStack.io is a curated, public repository of every AI agent that touches PR, marketing, comms, and media. One page. One search. Every agent.
Not tools. Not models. Agents. The things that go do work for you, autonomously or semi-autonomously, inside your workflow. Research agents. Pitching agents. Listening agents. Visibility agents. Briefing agents. Shopping agents. Procurement agents. The ones already inside your Slack, your Ads Manager, your CMS. And the ones your competitor is already running that you haven’t heard of yet.
Why I built it (and why this week)
The agent explosion is real, and it is not slowing down. Visa just declared the B2AI era this week (more on that below). Microsoft Copilot Checkout is buying for 500,000 merchants. Meta’s AI Business Assistant just went global. Anthropic ran an internal experiment where Claude agents closed deals with zero human approval. Every one of those moves is an agent doing the work that used to require a person on your team or a contract with a holding company.
There is no map. There is no central index. There is no easy way for a CMO to walk into a Monday meeting and say “here are the eight agents we should be testing this quarter and why.”
So I built one.
How to use it this week
If you are running a comms or marketing function, do these three things on Monday morning:
Open MyAgentStack.io and scroll the full list. Don’t filter. Just look. You will find five agents you didn’t know existed.
Pick one to pilot. Just one. Something tied to a workflow you do every week (briefings, pitching, AI visibility audits, ad ops, content distribution).
Bring it to your next standup with a one-page brief: what it does, what it would replace, what it would augment, and what the 14-day pilot looks like.
That’s it. You don’t need a new tech stack overhaul. You need to start running agents inside the workflows you already own.
What you’ll find on it
Categories on day one include:
Research and listening agents
Pitching and media outreach agents
AI visibility and citation agents (GEO, AEO, answer share)
Newsroom and editorial agents
Briefing and morning-digest agents
Ad ops and creative agents
Procurement and shopping agents (the B2AI layer)
Custom-built / vibe-coded agents from the community
Every entry has: what it does, who built it, what it costs, where to access it, and (where I have it) a real-world use case from someone running it in production.
Submit your own
If you are vibe-coding, building, or shipping an agent that touches our industry, send it to me. I review every submission. The bar is utility, not polish. If it solves a real problem for a real PR or marketing operator, it goes in.
Why this matters for your business
The companies six months ahead on AI right now are not the ones with the biggest budgets. They are the ones with the clearest agent inventory. They know what they are running, what they are testing, what they are sunsetting, and what they are about to buy. Most of you cannot list your current agent stack on one piece of paper. That is the gap.
MyAgentStack.io is the start of closing it.
Bookmark it. Send it. Submit to it. Let’s go.
⚡ THE BRIEF
1. Visa renamed commerce. It’s now B2AI.
What happened: Visa CMO Frank Cooper III declared the B2AI era this week. Visa’s research (with Morning Consult, surveying 2,000 consumers and 512 business decision-makers) shows 71% of businesses are willing to optimize products for AI agents, and 53% would let agents negotiate prices with other AI systems on their behalf.
What to do: Rewrite your top 10 product or service pages this week as a structured bullet sheet a shopping agent can quote back word for word. Specs, price comparables, differentiators, use cases. Skip the brand brochure prose.
Why it matters: When the buyer is a bot, paragraphs get skipped. Agents lift bullets and structured facts. If your page reads like a press release, you are invisible to the procurement layer that is now shopping for your customer. → Visa research summary
2. Anthropic let Claude agents run an office marketplace. Zero human approval. 186 deals closed.
What happened: Anthropic gave 69 employees Claude agents to buy and sell inside an internal marketplace for one week. The agents struck 186 deals across 500 listings, totaling just over $4,000. Sellers paired with Claude Opus 4.5 earned more per item than humans selling alone. Buyers using agents paid less.
What to do: Ask ChatGPT or Claude to shop for your top product under a real price point this week. If the agent cannot identify your brand, price the offer, and complete the transaction without you, your product page is broken for the agent buyer.
Why it matters: This is the first published in-house experiment proving agents can close transactions in production. It is not a research thought experiment. It is a working system. Your buyer’s procurement team is about to test the same thing on you. → Anthropic Project Deal
3. ChatGPT ads now reach logged-out users. Minimum spend dropped 75%.
What happened: OpenAI quietly extended ChatGPT advertising to unauthenticated users this week. Ads appear native inside chat responses, not as banners. The minimum advertiser commitment dropped from $200,000 to roughly $50,000. Canva and JobCopilot were among the first confirmed advertisers in logged-out sessions.
What to do: If your CMO has been waiting to test ChatGPT inventory because of the price floor, that excuse is gone. Brief your media team to model a $50K test against your top funnel campaign this week.
Why it matters: This is the first time mid-market budgets can buy ChatGPT inventory. The audience just expanded to every casual ChatGPT user, not just paid subscribers. Inventory is bigger, the floor is lower, and your competitor is going to test it before you do. → Search Engine Land · AdExchanger · OpenAI’s announcement
4. Microsoft Copilot Checkout hit 500,000 merchants. The agent is doing the buying.
What happened: Microsoft expanded Copilot Checkout to over half a million merchants and pushed it into the mobile app. Ask Copilot to redo your living room and it researches, picks pieces, and pays inside the chat. Urban Outfitters and Etsy are live.
What to do: Run a Copilot or ChatGPT shopping query for your top SKU today. If your product or a competitor is missing, screenshot the gap and send it to your e-comm lead before noon.
Why it matters: Half a million merchants is past pilot scale. If your product feed isn’t structured for agent retrieval, your category is getting bought without you. (Shameless plug: this is the entire reason we built Published Monthly at Zen Media.) → Microsoft news
5. MIT Sloan: AI is widening the gap, not closing it.
What happened: MIT Sloan Management Review published “How AI Helps the Best and Hurts the Rest” on April 20. The study reverses earlier findings that AI levels the playing field. Among entrepreneurs facing ambiguous problems, AI made the top performers significantly stronger and made everyone else worse.
What to do: Audit your team’s AI proficiency this week. Identify your top 20%. That is where to invest training time, tooling budget, and air cover. The other 80% needs a structured upskilling program, not access.
Why it matters: This is the screenshot-test stat for the entire quarter. The “AI is democratizing” thesis just collapsed. Your top performers pull further ahead with AI. Your bottom performers fall further behind. The gap is now a permanent line item on your org chart. → MIT Sloan Review
📊 This week’s RECEIPTS, REPORT DROP, and PLATFORM MOVES are below the fold for paid subscribers →
