How a Brief Is Built

An opinion with a number on it

On Tuesday, Brad Webb, Civic's CGO, opened Briefs, by Bryn, as a Public Preview. Bryn, by Civic, is the agent that builds them. Name a company and Bryn reads its site, scores it against the kind of company Bryn is built to find (in this preview the customer is Bryn; the same read, scored against your customer, is being built into Bryn), finds the tools it runs and the people who would sign off, writes the first follow-up it would run, and sends the Brief, inside a minute. The site puts it in one line: "Bryn scores the buyers on your site and runs the follow-up; a Brief is that read on one account."

Most research tools stop at the score. A number arrives, a color goes with it, and how the number was made is left to trust. That is an opinion with a number on it. A Brief ends differently. Its last section, Section 07, is headed "This page is a Play that already ran", and it is the reason the score can be checked rather than believed.

Section 07, the Play that already ran

A Play is a rule you set once: when this happens, do that. Bryn calls that a Play. Every Brief was built by one of Bryn's own rules, and it ends on the receipt: the clock of each step, from the request to the action. Section 07 lays the rule out in four lines. The request was watched. The company was scored. The page and the email were the action. The record was written. Those are the same four steps Bryn runs on a customer's own traffic, here pointed at a request from a stranger.

The request is where it starts, and the request itself is a signal: something a buyer just did, like coming back to your pricing page. Bryn files it where the interest belongs. When someone at company A asks for a Brief on company B, the interest is A's. Bryn notes it against A, and A's team gets the Slack ping when the Brief is done. B's own page in Bryn stays B's, and Bryn reads nothing from it until it has checked, by B's domain, that the page is B's.

Under the rule runs the Play trace: one row per step, each with the clock, from the request to Bryn's own follow-up in Slack, and the time elapsed. Under the trace sits the receipt, one row per section as it landed. Here is one trace for two invented companies. Someone at larkspur.example asked about quillfeather.example. Every stamp and value is illustrative.

1. "10:02:11" Request received, and filed on the asker's record.
2. "10:02:13" Site read: the home page, the pricing page, the docs. A site reader.
3. "10:02:19" Company scored, against that profile. The scorer. The score lands first.
4. "10:02:27" Tool stack read: the page, the tag manager, the company's own domain and mail records. A tag reader.
5. "10:02:36" People found: who would sign off, with a title read from a source. A people finder.
6. "10:02:49" Page assembled, section by section, in the company's own colors. The assembler.
7. "10:02:52" Email sent to the person who asked. The sender. The email is the Brief, not a notification.
8. "10:02:53" Slack follow-up posted: Bryn's own, once the work is done.
9. "10:02:54" Record written. Section 07 closes, forty-three seconds after the request.

Step through the same rows below. Each row shows its clock stamp, the step in plain words, the kind of tool, the time elapsed, and the section of the Brief it landed.

Section 07, row by row.

Step through one illustrative Play trace. Left: the trace row and what it means in plain words. Right: the Brief, with the section that row landed highlighted. Use the row buttons or Prev and Next; Left and Right arrows, Home and End work while a row button has focus.

(illustrative run: someone at larkspur.example asked about quillfeather.example; every clock stamp is invented)

This page is a Play that already ran

The request was watched. The company was scored. The page and the email were the action. The record was written. Under the rule, the trace. Under the trace, the receipt.

1. "10:02:11" request received ⬩ signal ⬩ +0s ⬩ filed on: the asker's record, larkspur.example

   Someone at larkspur.example asked for a Brief on quillfeather.example. The request is a signal, something a buyer just did, and Bryn notes it against larkspur, and larkspur's team gets the Slack ping when the Brief is done. Quillfeather's own page in Bryn stays quillfeather's, and Bryn reads nothing from it until it has checked, by domain, that the page is quillfeather's.

   Lands: the Brief's header, asked for.

2. "10:02:13" site read ⬩ site reader ⬩ +2s ⬩ source: quillfeather.example: home, pricing, docs

   Bryn reads the company's own pages first: the home page, the pricing page, the docs. The brand comes from here too, the logo, the palette, the type, so the Brief arrives in the company's own colors. This is where the brand and key pages section comes from.

   Lands: brand and key pages.

3. "10:02:19" company scored ⬩ scorer ⬩ +8s ⬩ source: the site read, against that profile

   The score lands first. Bryn scores the company against that profile, with the same weights and the same bar a customer's own account would have to clear, and prints what it looked at and what would move the number. Does it look like a buyer.

   Lands: the score.

4. "10:02:27" tool stack read ⬩ tag reader ⬩ +16s ⬩ source: the page, the tag manager, the company's own domain and mail records

   The page, the tag manager, and the company's own domain and mail records say which tools are already running. Each tool found becomes a tile, with a word for where it was proved. A tool with no proof behind it is not a tile.

   Lands: tool stack.

5. "10:02:36" people found ⬩ people finder ⬩ +25s ⬩ source: public profiles and titles

   The people who would sign off, each with a title read from a source. Initials only in this illustration. Nobody is added to make the list look fuller.

   Lands: people who would sign off.

6. "10:02:49" page assembled ⬩ assembler ⬩ +38s ⬩ source: every row above, section by section

   The assembler puts the sections in order, in the company's own colors, and writes the first follow-up Bryn would run from everything above. A section that comes back with a gap is repaired on its own, and the sections already landed stay put.

   Lands: the first follow-up Bryn would run.

7. "10:02:52" email sent ⬩ sender ⬩ +41s ⬩ to: the person who asked

   The email goes to the person who asked. It is the Brief, not a notification: the score, the tools that connect, one insight, and the first follow-up, in the company's colors.

   Lands: email.

8. "10:02:53" Slack follow-up posted ⬩ sender ⬩ +42s ⬩ Bryn's own follow-up

   Bryn's own follow-up lands in Slack, once the work is done. You don't have to check Bryn. Bryn checks in with you, once the work is done.

   Lands: Slack.

9. "10:02:54" record written ⬩ Section 07 ⬩ +43s ⬩ the rule, this trace, the receipt

   The record is written and Section 07 closes: the rule that built the page, this trace with the clock on every row, and the receipt, one row per section as it landed. Forty-three seconds from the request, inside a minute (illustrative).

   Lands: 07, this page is a Play that already ran.

Forty-three seconds, in this illustration. The run has a sixty-second wall, and the page finishes inside it or says which section is still coming while the rest keeps moving. A section that comes back with a gap is repaired on its own, and the sections already landed stay put. Ask about the same company again inside thirty days and you get the same Brief back, not a second run. Open it later and the score has re-read itself, with a line headed "Since this Brief was sent" beneath it; that part is Tuesday's story, in Brad's Let me be Brief...

Why the person who signs off reads it first

The person who approves a Brief, or approves the outreach a Brief leads to, is not the person who wanted the score. They are the person who will be asked, later, why the company was contacted. Section 07 is the answer they can give. It says what ran, in what order, at what time, and from where, and it says it in rows a second person can reconstruct from.

We have made this argument twice before, about Bryn's own record. Your Rules, Readable held Bryn's profile, weights, and rules to a three-question test: can you read it, change it, and see what your change did. Show Me a Blocked Action ran a vendor test in public: one account the agent was not allowed to touch, the rule that blocked it, and the record line. A Brief is the same habit applied to research. The record is the pitch, and the pitch holds up because it was not written to be one.

Read the trace before the score and the score changes character. It stops being a verdict and becomes a sum you can check, with the clock still on it.

One step, one line

Behind the trace is one rule, and it cuts both ways. Every step that ran writes a row. Every section the Brief shows traces back to one of those rows, with its clock stamp. If a section's row has not landed, the Brief shows the gap and repairs it; it does not fill the space with a guess.

So a Brief's tool stack shows the tools the tag reader actually found, each with the place it was proved, on the page, in the company's own domain and mail records, or in its copy, and no more, however short that leaves the row. The people who would sign off are the people the people finder could place with a title it read from a source, not a padded list of likely names. A blank tile is honest. A filled tile with no row behind it is a guess wearing a uniform.

This is the same principle Bryn applies to its own account records, where a field a source stops providing is removed rather than left to go stale, as Brad described in Bryn's First Big Leap. An account page and a Brief are the same kind of document. Both would rather show you less than show you something nobody vouches for.

The same habit, everywhere

A Brief is one account, read on request. The rest of Bryn is every account, read as it arrives. Bryn watches your site and product for buyers, works out who they are, decides if they're worth acting on, and runs the follow-up you approved, within minutes. Every run is logged, so the people who sign off can read it.

The follow-up it runs is a Play, and a Play's run leaves the same record a Brief does: a trace with the clock on every row, because the alternative is a result nobody can explain a week later. That is also why the Slack row is in the trace. You don't have to check Bryn. Bryn checks in with you, once the work is done. Bryn is not another dashboard to watch. It is the governed execution layer that runs Plays through your stack. In plain terms: you write the rule once, Bryn carries it out inside the tools you already use, and nothing runs without a record of who approved it.

Two pieces this week sit beside this one. Brad Webb's Let me be Brief... is the announcement: what a Brief is, what is in one, and how to get one. Chris Hart's Let Your Agent Set Up Your Tools told buyers to look for three things before handing an agent the setup job: a draft, a human go-live decision, and "a record of what changed." Section 07 is the research version of his third question. It is what a record of what changed looks like when the thing that changed is what you know about a company.

Get a Brief

Run one on a company you are about to talk to. The site says what comes back: "A Brief reads your site and names what's already running, what connects today, and what Bryn would replace." Read Section 07 first. If you would have done the same steps by hand, that is work you can stop doing by hand, which is where Brad Webb's series on the one-person growth team went yesterday with The Work You Stop Doing.

Briefs, by Bryn, is a Public Preview at civic.com/briefs. Any email address signs in. Bryn, by Civic, is the agent that built it.

Then take the habit with you. The next time a tool hands you a score, ask it for the trace: what did it do, in what order, at what time, and from where. If it cannot show you, the number is an opinion, and you can treat it as one.

Stop watching signals. Start running them.