Frequently asked questions

Straight answers, including the ones that are awkward. Where we haven't decided something yet, we say so instead of inventing policy on a marketing page.

Product

Which tools do you support?

Claude Code — agent-session transcripts — today. That's it. We're pre-launch and would rather do one source well than list integrations we haven't built. If your team uses something else, tell us; that input genuinely shapes what we build next.

How is this different from DORA-style metrics or seat-usage dashboards?

Those measure output volume or license use. DORA tells you how fast code ships; a seat dashboard tells you who logged in. Neither tells you whether anyone is actually good with the tools. We observe practice — what people actually do inside their sessions: planning before coding, delegating to subagents, using skills and saved commands, tracking tasks, managing context, committing work. A team can have 100% seat adoption and healthy DORA numbers while most engineers use the agent as a fancy autocomplete. That gap is invisible in standard metrics, and it's exactly the thing we make observable.

Does it grade my code or my people?

No. Accrete tracks practice adoption — whether a behavior like plan-mode or subagent delegation appears in someone's sessions — not performance scores, code quality grades, or rankings. Deliberately. Scoring people makes them game the metric and hide the messy sessions, which are the ones you learn the most from. Adoption data answers a coaching question ("who hasn't tried planning yet?"), not an evaluation question ("who's the worst engineer?"). We built it for the first and won't sell it as the second.

Does "adopted" mean someone is good at the practice?

No, and we keep the two apart on purpose. A practice counts as adopted per session it appears in; volume of use is tracked separately. Adoption tells you someone has the habit. Proficiency is a different question, and we don't pretend a session count answers it.

Data & privacy

Is this surveillance?

It's the fair question, so here's the honest answer: Accrete observes work sessions, and any tool that does can be misused by bad management. What we've done about it: we measure practice adoption, not productivity scores or rankings; engineers run their own local explorer and see their own data on their own machine; and there are no performance grades to weaponize. What we can't do is fix a manager determined to use any data punitively. If your org's plan is to stack-rank engineers with this, we're the wrong tool and we'd rather you not buy it.

What exactly do you collect?

A small sync client on each developer's machine parses local Claude Code session transcripts into a local database, then pushes the parsed session data — not the raw transcript files — to your team's server. At ingest, each session is classified: which messages were actual human prompts versus tool output, which project the work belongs to, and which practice signals appeared. Full details on the privacy page.

Where does our data live? Is it pooled with other companies'?

One isolated database per company. No cross-company pooling, no shared org hierarchy, nothing common between tenants. Your data sits with your team's deployment and nowhere else.

Can engineers see their own data?

Yes. The same explorer that powers the team view runs locally on each engineer's machine, against their own sessions. They see what's collected before it syncs, and they get the personal-practice view for themselves — which is most of why senior engineers bring Accrete in bottom-up.

Why "durable capture"? What's the 30-day thing?

Claude Code purges raw transcripts after roughly 30 days. Accrete parses them into a durable local database before that happens, so your session history doesn't quietly evaporate. We haven't published a formal retention policy yet — pre-launch — so if you have specific retention requirements, ask us and we'll work it out directly.

Rollout

What does setup look like?

A small sync client on each developer's machine. It parses local transcripts and pushes parsed data to your team's server. No IDE plugin, no proxy in front of your model traffic, no change to how anyone codes. How it works walks through the pipeline.

Do we need the whole team to install it?

No, but the picture is only as complete as the machines that sync. The adoption matrix shows the people who are syncing; it can't say anything about the ones who aren't. Most teams we talk to start with a volunteer group and expand once people have seen their own data.

Will engineers push back?

Some will, and the skepticism is healthy. What's worked in practice: lead with the individual value — the local explorer gives each engineer visibility into their own practice before anything goes to a team server — and be explicit that there are no performance scores in the product. The origin story helps too: the founder built this to see his own sessions first. If you roll it out as a monitoring mandate from above, expect resistance; you'd have earned it.

Pricing & access

What does it cost?

We don't have public pricing yet — pre-launch, no point pretending otherwise. Design partners work directly with the founder, shape the practice catalog, and get early pricing. If you want a number, join the early access list or ask us directly; we'd rather quote you than publish a placeholder.

Can I use it solo?

Yes. The local explorer works standalone on your own machine, against your own sessions, with no team server involved. That's actually how the product started — see Accrete for engineers.

What happens to our data if we leave?

Your data is yours. It lives in one isolated database for your company, so there's no untangling it from anyone else's — handing it over or deleting it is structurally simple. We haven't finalized the contractual language yet (pre-launch, again), so we won't quote terms here that don't exist. The stance is settled even if the paperwork isn't: we don't keep, mine, or resell a departing customer's data.

What's a design partner?

A team that works directly with the founder during the pre-launch period: you get early pricing and a real say in the practice catalog; we get honest feedback from real sessions. Limited slots by design — it only works if we can pay attention to each team. Interested? Join the early access list and say so.