Tools Converge. Action Doesn't.

On September 1, Clay made its Sequencer generally available. The post's own summary: it "brings lead sourcing, enrichment, and sequencing into one tool, so every email sends on current data." The pitch is freshness at send time. If your list, your enrichment, and your send live in three tools, the row is stale by the time the email leaves. Put them in one place and the send reads the row as it is right now.

That is good product strategy. Clay's July roadmap post said the quiet part: Sequencer 2.0 turns cold outbound into "one end-to-end workflow inside Clay," and the company describes itself as evolving into an end-to-end orchestration platform for go-to-market.

We are not here to grade the launch. We are here to read it next to the other launches this year, because they rhyme.

Everyone is absorbing the neighbor

Look at the last ten weeks.

Clay, a data company, shipped Account Research Agents on July 22 that maintain fields in your CRM or warehouse continuously instead of answering one question once. Then it shipped sending. Data absorbed research, then data absorbed sequencing.

6sense, an intent company, put its buying-stage predictions inside MCP-compatible agents on August 10: Claude, ChatGPT, Writer, Agentforce, plus people-search APIs and signal-triggered workflows. Intent absorbed the delivery surface. We wrote that one up as Intelligence Became an Ingredient.

Warmly, a visitor-identification company, describes itself in its own words as a platform that "identifies, qualifies, and engages," with chat, email sequences, and rep routing attached. That is the vendor's own positioning, and a clear statement of direction: visitor ID absorbed outreach.

And the CRMs keep adding intent panels and agents of their own, which is the same pattern from the other direction: the system of record reaching toward the signal.

Five categories, and each is eating the one next to it. Pick one below and see what it has swallowed since 2024.

What none of them absorbed

Read the launches again and look for the sentence that says who decided the action was allowed.

It is not there. Not because the vendors forgot it, but because it is not theirs to say. Clay's Sequencer will send on the freshest data in the category, and it still needs someone to decide that this segment gets this sequence. 6sense's predictions arrive inside your agent, and then someone decides what the agent may do with them. Warmly engages a visitor while they browse, inside rules a human wrote and approved.

We said earlier that seeing is consolidating and acting is still open. Then we said that once intelligence is callable from anyone's agent, it has become an ingredient. This is the third step: the tools themselves are consolidating, and the action is still open. Consolidated seeing and ingredient intelligence still end in the same place, a human's queue, where the approved action waits for someone to run it.

Why the empty slot stays empty

Acting on a signal requires three things that none of the launches claim, because none of them are features.

A boundary: which channels, which accounts, how many per day, what the action may never touch. Approval: someone with authority said yes to this action pattern, once, in a form that can be pointed at later. A record: what was seen, why it qualified, what ran, where it landed, when.

Those are governance properties. You cannot bolt them onto an enrichment column or a send button. They have to be the shape of the thing that acts. Chris writes tomorrow about what happens to agent adoption when those properties are present, so we will leave that argument to him.

The point for today is smaller: convergence made the empty slot more visible, not smaller. When five tools each did one thing, the missing owner could hide in the handoffs. When three tools each do three things, the missing owner is the only gap left on the shelf. We wrote about what happens when nobody fills it in Thirteen Agents, One Owner.

Draw your shelf

The exercise takes fifteen minutes.

List every GTM tool you pay for. Next to each, write two columns. Column one: what it sees. Column two: what it is allowed to do without you.

Column one fills fast. Visitor identity, intent topics, job changes, funding, tech stack, opens, reply sentiment. Column two is where the pen stops. Most rows say "sends what I loaded," or "alerts me," or nothing. A few rows say "engages automatically," and when you open the settings they mean "runs a rule someone set once."

The empty column is the job. Not the tool's job: yours, or whoever you hand it to. Every launch this year made column one longer and left column two alone.

Where Bryn sits

Bryn does not compete with the shelf. It runs Plays through it. A Play is an action pattern you approved once, with a boundary: when this kind of signal appears on this kind of account, run this sequence into these destinations, within these limits. Clay can be a destination. Your sequencer can be a destination. So can your CRM and your Slack. Bryn is not another dashboard to watch. It is the governed execution layer that runs Plays through your stack.

When a Play runs, Bryn writes the line itself, in its own voice, undramatic. The account and numbers here are illustrative:

I ran pricing-follow-up-v2 against acct_4471: pricing then comparison then repeat (7d), score 0.81 against threshold 0.70, sequence enrolled via Clay Sequencer, 14:02 UTC.

That line is column two, filled in, with the record attached. The tools around it can keep converging. The action has an owner now, and the owner can prove it.

The tools converged. The action still needs an owner.

Clay's Sequencer is a good product and a fair reading of where the category is going. So are the moves from 6sense and Warmly. None of them claim to own the approved action, and none of them should. That slot was never a feature. It was always the job.

Stop watching signals. Start running them.