We shipped an agent that mines your Attio calls for content. Here is how it works.

A sales call is the best content research session you will ever run.

The prospect tells you exactly how they describe their problem. The exact words. The before and the after. The objections they had walking in. The vocabulary specific to their space that your marketing team did not know they used. Most of that disappears after the meeting ends.

The Content Mining Agent stops it from disappearing. It reads every call transcript in your Attio workspace, extracts the content seeds, writes them to a Notion backlog, and sends the top five to Slack every Friday morning.

This post is what it does, how it runs, what it costs, and how to install it on your own workspace.

What it extracts

Five categories per call.

1. Pain quotes. The exact words the prospect used to describe their problem. Verbatim, not paraphrased. Marketing's best opening line was always the customer's own sentence.
2. Before-and-after stories. What the team had before. What they are looking for now. What changed in their world that made this a problem this quarter, not last year.
3. Objections. What the prospect pushed back on. How it was resolved or how it stayed unresolved. Both shapes are content.
4. Vocabulary. Industry terms the prospect used that your team has been writing around. The Content Mining Agent surfaces them so your site and copy start using the same words your buyers use.
5. Use cases. What the prospect wants to do with the product that was not in your original pitch. The fastest signal for new positioning, new pages, and new features.

Each extraction becomes one row in a Notion database the agent auto-creates on first run.

Where it writes

A Notion content backlog. One database, one row per seed, every row linked back to the source call in Attio.

The agent tags every seed by format: Blog, Case study, LinkedIn, Tweet, FAQ, Talk. A writer opens the database, filters to the format they are working in, clicks through to the call in Attio to hear the moment in context, and writes from raw material instead of from a blank page.

The Notion database is yours, on your workspace, on your Notion account. We do not host it. We do not see your calls. The agent runs entirely inside your stack.

The Friday digest

Every Friday morning, the agent posts the top five seeds from the week into a Slack channel of your choice.

The digest is not a recap. It is a hand-off. Five quotes, ready to write, with the customer, the call link, and the suggested format. Content planning meetings stop starting from zero.

We picked Friday because that is when most teams plan content for the next week. The digest lands before the planning meeting, not after.

How it runs

The Content Mining Agent ships as a Claude Code skill. The customer installs it once, schedules it once, and never touches the infrastructure again.

The runtime stack is short:

• color:var(--color-text-heading)]">The skill lives in a public GitHub repo at [github.com/gmaramigin/attio-content-mining-skill. Markdown only. No Python runner. No Anthropic API key. No service to host.
• The reader is the Attio MCP server, already connected to your Claude Code. The agent calls the call-recording search and read endpoints to pull transcripts.
• The writers are the Notion MCP and the Slack MCP, also already connected on the customer's Claude Code.
• The cron is /schedule inside Claude Code. The scheduled run fetches the latest SKILL.md from raw GitHub each fire, so improvements to the agent propagate without the customer touching anything.

The customer's Claude Code subscription is the entire runtime. We do not host the agent. We do not hold an API key. If we disappear tomorrow, the skill keeps running on its own.

This is the rule for every agent in our catalog. Pipeline Hygiene works the same way. Deal Focus works the same way. The customer owns the runtime, the data, and the schedule.

Why a content backlog beats a content calendar

Most content programs run from a calendar. Topics get planned in advance. Writers fill in the topics. The output reads like everyone else's because the topics came from a competitor scan, not from customer mouths.

A content backlog runs from raw customer language. The team writes what the customer just said, in the words the customer used, with the proof of the call sitting one click away. The output sounds like a person who knows the customer, because it came directly from the customer.

The Content Mining Agent is the conversion from "calls we did" to "backlog of things to write" without a person sitting through every transcript. The work the agent does in twenty minutes is the work that used to take a content marketer a full afternoon.

What it costs

Build. A one-time build inside your workspace. Includes the skill installed on your Claude Code, the Notion database created, the Slack channel wired, the schedule set, and a working session to tune the format taxonomy to your team.

Run. Optional monthly retainer if you want us watching the output, tuning prompts, and adjusting the taxonomy as your positioning shifts.

Claude Code consumption. The agent runs on your existing Claude Code subscription. No separate Anthropic API key. No usage-based bill arrives from us.

Pricing for both options lives on the pricing page.

How to install

Three steps if you want to install it yourself.

1. Clone the skill repo into your Claude Code skills directory. git clone https://github.com/gmaramigin/attio-content-mining-skill ~/.claude/skills/attio-content-mining-skill.
2. Connect the MCPs for Attio, Notion, and Slack on your Claude Code. Each takes about five minutes the first time.
3. Schedule the run. Inside Claude Code: /schedule "Run the attio-content-mining-skill in scheduled mode" --cron "0 9 * * 5". That is 9 AM Friday in your local timezone. The skill handles the rest.

If you want the agent live this week without doing the install yourself, book a call and we will set it up on your workspace.

Why this is the lowest-effort agent we ship

Most teams have a year of call recordings already in Attio and have never read them past the week they happened. The signal is sitting there. The blocker has always been the effort of listening to every call.

The Content Mining Agent removes the blocker. The calls happened anyway. The transcripts are already in Attio. The agent is the small piece that turns existing data into a content backlog the team will actually use.

If your team is sitting on a year of recordings and not extracting anything from them, this is the place to start.