An ICP defined on live data, every account scored for fit, customers and open deals suppressed, and the result pushed as a matched audience into LinkedIn and Meta or exported in the format each platform accepts. The infrastructure under targeting, run by autonomous AI agents.
LinkedIn Ads company and contact audiences, Meta Ads custom audiences, pushed from a list after one approval.
Google Customer Match, generic hashed CSV, plain CSV and JSON, formatted to what each platform imports.
Fit against the ICP with reasons, so the audience is the top of a ranking instead of the output of a filter.
Customers, open opportunities, competitors and the suppression list removed before any audience is built.
Industry, size and title filters inside the ad platform are the closest thing to a definition anyone has. The accounts that matter most, the ones hiring or switching stack this month, look identical to the rest inside those filters.
A CSV uploaded in March is still the audience in September. Customers are being retargeted, closed-lost accounts are being shown the launch ad, and refreshing means starting the export over.
The upload matched 40% and nobody knows why. Missing domains, stale emails and inconsistent company names each take a slice, and there is no tool between the CRM and the platform to fix them.
Industries, size bands, geographies, technologies, funding stages and signals held as a profile the agents apply, drafted from your site and your best customers and editable in plain language.
Domains resolved, emails verified, names normalized, duplicates merged and customers, open deals and competitors removed, so the audience uploads clean and matches high.
Company and contact audiences created in LinkedIn Ads and custom audiences in Meta Ads directly from a list, with match estimates shown and status tracked after the push.
Google Customer Match, hashed CSV, plain CSV and JSON exports formatted to each platform’s import, with the same suppression applied and the list kept for the next refresh.
Confirm the ICP or describe the audience in one sentence: "mid-market SaaS in North America running a modern CRM and hiring in RevOps".
Accounts found by firmographics, technographics, signals and similarity to your customers, then contacts by title and seniority where the platform matches on people.
Fit against the ICP with reasons, so the audience is the top of a ranked list and the threshold is a decision you can see.
Customers, open opportunities, competitors and the suppression list removed; domains, emails and names normalized for the match.
A matched audience in LinkedIn Ads or Meta Ads after approval, or a Customer Match or hashed CSV export for the platforms you upload to.
The list re-runs on a schedule, new accounts added, converted ones suppressed, the audience updated after the next approval.
Every company that fits the ICP and shows a buying signal this quarter, scored, suppressed against the CRM and pushed as a company audience so the campaign runs on accounts sales would actually want.
The decision-makers at target accounts resolved to verified contacts and pushed as a custom audience, so the ad reaches the people who sign, on the platform where they scroll.
The same scored, suppressed list formatted as a Google Customer Match file, ready to upload, refreshed on a schedule so the file in the platform stays current.
Customers, open opportunities and churned accounts as their own audiences, so acquisition spend excludes the people who already know you and retention campaigns reach only who they should.
| By hand | With AstroFabric | |
|---|---|---|
| Definition | The ad platform’s filters | An ICP on live data, scored with reasons |
| Signals | Invisible inside the platform | Hiring, funding, stack and intent decide who is in |
| Suppression | Remembered sometimes | Applied before every build, from the CRM |
| Match rate | Unknown and unexplained | Domains resolved and emails verified first, estimate shown |
| Push | CSV upload, per platform | LinkedIn and Meta direct, Customer Match formatted export |
| Freshness | The March export | Refreshed on a schedule, updated after approval |
Companies that fit the ICP and show a signal this quarter, scored and suppressed against the CRM, pushed as a LinkedIn company audience after approval.
Decision-makers at target accounts verified and pushed as a Meta custom audience, with the match estimate shown before the push.
A scored, suppressed list formatted for Google Customer Match every Monday, delivered to a shared drive with the changes since last week noted.
Three of a hundred. The full playbook library ships in every workspace, and anything you can describe becomes a mission.
Building, scoring and cleaning run inside the credit ceiling with no gate. Creating or updating an audience in an ad account is staged for approval exactly as previewed.
Customers, open opportunities, competitors and the workspace suppression list are removed before an audience exists, and the removed count is shown at the gate.
Google Customer Match and other platforms receive a formatted file that your team uploads inside the platform; the agents never operate those accounts.
Company and contact data come from licensed providers. The agents never scrape member networks.
The layer between your ICP and your ad platforms: accounts and contacts identified on live data, scored for fit with reasons, cleaned and suppressed against the CRM, then pushed as matched audiences into LinkedIn Ads and Meta Ads or exported for other platforms, with a person approving each push.
LinkedIn Ads and Meta Ads receive audiences directly after approval. Google Customer Match and any platform that imports a hashed or plain CSV receive a formatted export that your team uploads. Reddit Ads and others use the same export path.
Connect the CRM and the agents read customers, open opportunities and closed-lost accounts, add competitors and the workspace suppression list, and remove all of them before an audience is built. The count removed is shown at the approval gate every time.
Before the push, domains are resolved to the canonical website, company names normalized, emails verified through a waterfall of licensed sources and duplicates merged. The platform receives the identifiers it matches on, and an estimate is shown before you approve.
Yes. The list behind the audience carries its criteria, so a schedule re-runs discovery and scoring, adds new fits, suppresses converted accounts and stages the updated audience. The push itself still waits for approval, and the diff since the last refresh is posted to Slack with each run.
No. The agents build and push audiences and format exports; creative, budgets, bidding and campaign settings stay with your team inside each platform. The same boundary applies to the API, MCP server and CLI, so an agent you run can build an audience and cannot spend from an ad account.
Identifying accounts and enriching them is a few credits per row and a verified contact is a few credits; scoring, suppression, pushes and exports are free. Rates show before the run, misses on contact finding cost nothing, and plans start at $49 per month.
Describe the audience. A scored, suppressed list is ready to push into LinkedIn or Meta, or export anywhere, after one approval.