Airtree details its Claude Cowork rollout and claims research now takes minutes

Airtree has made Claude Cowork part of its shared working infrastructure, using it to gather research and prepare board materials across teams. In a new Anthropic customer interview, partner Jackie Vullinghs says research that once took two days now takes minutes—freeing time, in her account, for deeper diligence and founder relationships.

Published October 30, the edited Q&A describes an existing deployment at the venture firm, which backs founders in Australia and New Zealand. The time savings are Airtree’s reported experience, published by its AI supplier, rather than independently verified measurements.

The obstacle was scattered information

Airtree’s starting problem was not a lack of data. Board reports, portfolio updates and market research drew on information spread across Google Workspace, Slack, Notion and other tools. Dashboards, extra Slack channels and consolidated reporting had not solved the work of finding and combining it. Vullinghs describes that preparation competing with time spent getting to know founders.

Some employees had begun using Claude Code to automate routine tasks, but teaching everyone to work through a terminal—a text interface for computer commands—looked difficult. Cowork’s interface lowered that barrier. Airtree’s data lead built a one-hour course for different functions, progressing from basic use to connected tools and Skills: reusable instructions for recurring work.

Skills made by one employee could then serve colleagues, rather than remaining personal shortcuts. Airtree’s previously published public course shows the training sequence behind that approach. The firm says it uses the same beginner-friendly material internally and that it takes about an hour.

• Read and organize folders, then describe the desired outcome rather than each step.
• Build reusable workflows, with a bonus lesson on connecting workplace tools.

A board meeting becomes a repeatable workflow

The board-preparation Skill reads a portfolio company’s Google Drive materials, searches Slack for updates, checks competitor and market news, and examines comparable public companies. It compares the findings with earlier preparation to track changes, commitments and unfinished business. A separate Skill adds company-specific context, including financial metrics and recurring issues to watch.

After the meeting, the Board Summary Skill checks captured notes against the preparation documents. It produces a detailed record saved to Google Drive and a roughly 250-word Slack summary, sent when the user says “Send.” Vullinghs says the strongest results come from routine tasks with a narrow, explicit scope, rather than handing investment judgment to AI.

Beyond board work, daily pre-meeting research that took an hour now takes minutes, Airtree says. Its marketing team reports saving two hours a month by supplying platform-results CSV files for dashboards and summaries. Vullinghs says company research also lets investors enter first meetings with a deeper understanding of a founder’s business, instead of spending the conversation on introductions.

Review stays mandatory as the task list grows

Airtree requires human verification and review of all Cowork-generated content before decisions or publication. IT security, risk and compliance managers assessed the tool before use; the firm records usage and requires responsible-AI training. Those rules sit alongside connections to systems including Gmail, Affinity, Granola, Notion and Slack, where the firm holds sensitive information.

The next proposed workflows go further. Vullinghs describes a planned internal hackathon, with ideas including a portfolio snapshot portal, travel briefings and fund-analysis reports explaining changes between periods. These remain proposals in the interview, not demonstrated extensions of the current time savings.