Hey everyone 👋
We want to start a thread on something we’ve been building a lot with small and mid-sized teams: a company brain. That means one place in Airtable where the business’s context lives (clients, projects, processes, decisions, SOPs), so that both people and AI agents can read from it and write back to it.
The problem we keep seeing is that most SMBs have their knowledge spread across Slack threads, Google Docs, and a few people’s heads. When someone leaves, or you try to plug in AI, there’s nothing solid for it to stand on.
Here’s what’s been working for us so far:
- Separate tables for each type of knowledge. For example, Clients, Projects, Processes/SOPs, Decisions log, and FAQs, all linked together instead of kept in one giant table.
- A “source of truth” field and an owner on every record, so people know whether something is current and who keeps it up to date.
- AI agents that read from the base, like a lead responder or an ops assistant, so they answer with real company context instead of guessing.
- Agents that write back, for example logging call summaries, updating project status, or flagging records that look stale.
- Interfaces built for each team, so people actually use the brain day to day instead of treating it as an archive.
Some things we’re still figuring out, and would love your take on:
- How do you keep the brain from going stale? Do you use automations, review cycles, or something else?
- Where do you draw the line between what goes in Airtable and what stays in docs?
- Is anyone connecting their base to an LLM or agent? What does your setup look like?
- How are you handling permissions when AI has read/write access?
Share your setups, screenshots, or even failed attempts. We’ll post some of our base structures in the replies too.