tl;dr On August 10, 6sense shipped an MCP server and intent APIs that make its buying-stage predictions callable from Claude, ChatGPT, Writer, and Agentforce. When the strongest see-layer vendor turns its intelligence into an ingredient for anyone's agent, intelligence has stopped being the product. What stays scarce is the layer that can take an approved action on the signal and leave a record.
In August, 6sense announced four product releases that move its intelligence out of its own suite: an MCP server that makes predictive buying stages and keyword intent callable from Claude, ChatGPT, Writer, and Agentforce with no custom integration, people search and enrichment APIs, a programmatic intent data API, and configurable signal-triggered workflows.
It is a smart set of releases, and it is worth reading for what it says about the category rather than about the vendor.
For a decade, the intent data business worked like a walled garden: the predictions lived inside the vendor's suite, and you went there to look at them. The suite was the product. The data was the moat.
An MCP server inverts that. The predictions now travel to wherever the buyer's agent lives. Any team, on any stack, can call the same buying-stage model from the same endpoint. The garden wall just became a serving window.
That is what commoditization looks like from the inside, and it is rational. When every GTM team is standing up agents, the vendor that refuses to be callable gets left out of the conversation entirely. Being an ingredient beats being a museum.
STACK UNBUNDLER (interactive: pick where the see-layer's intelligence lives; the stack redraws and the scarce layer is named) Where intelligence lives, what that means, and what stays scarce: 1. Inside the vendor's suite: you go to the dashboard to look at predictions. Differentiation comes from having access at all. Scarce: access to the data. 2. Behind an API: your warehouse and tools pull the predictions. Access stops differentiating; plumbing does the work. Scarce: what you build on top. 3. Inside any agent, via MCP: any team's agent calls the same model. Intelligence is an ingredient priced per call. Scarce: governed execution: profile, approved Play, routing, record. Qualitative; describes market structure, not any single vendor's terms.Here is the strategic read for a growth team. When intelligence is callable, composable, and priced per call, it stops being something you can win with, because the two competitors in your deal cycles can call the identical endpoint.
We made this argument a year out from the fact in Seeing Is Consolidating. Acting Is Still Open. The see-layer consolidates and commoditizes; the act-layer stays open. The August releases are the see-layer agreeing.
Notice what is not in the release: execution. The intent data arrives inside your agent, and then everything hard is still yours. Which accounts matter against your profile. What action is approved. Who runs it, where it lands, and what record exists afterward.
THE SHELF AND THE KITCHEN (static figure: the see layer as a shelf of per-call ingredient jars feeding one execution kitchen) Everyone shops the same shelf. The dish is made in the kitchen. THE SHELF, the see layer, now callable by anyone: intent feeds, enrichment, buying stages, keywords, technographics, hiring data. Each priced per call. THE KITCHEN, what stays yours: - Your profile: which accounts matter - Your Plays: the approved actions - Your routing: where the action lands - Your record: what ran, why, and when Two competitors can buy the same jars. They cannot buy your kitchen. Illustrative, not a benchmark. Prompted by 6sense's August 2026 releases making intent data callable via MCP and APIs.The Monday exercise is a short vendor conversation. Ask three questions.
Where can this data act, not just appear? A feed that lands in a dashboard is a museum piece. A feed that lands in an agent is an ingredient. Neither is an outcome.
What action can it trigger without a human re-keying it? If the answer involves a person copying an account name from one tool into another, the data is not driving anything. The person is.
What record exists after it acts? If a signal fired and something happened, you should be able to point at the line that says what, why, and where.
Bryn sits on the answer side of those questions. Bryn is not another dashboard to watch. It is the governed execution layer that runs Plays through your stack: it takes signals wherever they live, including rented ones, evaluates them against the profile you defined, runs the Play you approved, and logs every step. Ingredients are welcome. The kitchen is the product.
Stop watching signals. Start running them.
Source: https://www.civic.com/field-notes/intelligence-became-an-ingredient