Skip to content
Book a demo

The runbook has the steps.The engineer has the reasons.

Code review catches what changed. It doesn't catch why the retry is there, why that library was never replaced, or which alert the team has learned to ignore. That reasoning sits with the engineer who made the call, and it leaves when they do.

Book a demoSee what comes back

What walks out the door

The commit says what changed. It rarely says why.

OakHive interviews them before they go. This is what comes back.

The retry nobody dares remove

One service retries on a specific error because of an incident years ago. The comment says what it does. The person who wrote it is the only one who knows what happens without it.

Handover guide. What lands on you first: The five things that need attention in week one.; The retry nobody dares remove: What was decided, and the reasoning behind it.; Known risks & exceptions: The one case that breaks the default process.; Who to ask, and when: Contacts for what isn't in this guide..

From the handover guide, written to whoever takes the role next.

The library that was never replaced

Staying on it was a decision with constraints behind it. Without those, the next team either rips it out and finds out why, or keeps it and can't defend it.

Guided session, Chapter 2 of 4. Chapters: The five things that land on you first; The library that was never replaced (open); The one exception that breaks the default process; When to call, and when to escalate.

From the guided session that walks the role chapter by chapter.

The alert everyone ignores

The team learned which alerts mean something. That sits with the people on the rota, not in the alert's description.

Ask Monica. A colleague: Is this alert safe to ignore? OakHive: During the nightly batch, yes. Outside that window she treated it as real and checked the queue depth first. A colleague: And who signed that off? OakHive: That wasn't covered in her interview, so I won't guess at it.

From the chat, answering after they have gone.

The deploy step that isn't in the runbook

There's an order to things that only matters when it's wrong. Whoever has done the deploy fifty times knows it; the runbook was last updated two rewrites ago.

Manager's summary, Section 4 of 4. Sections: What the role actually covered; Deploy steps that aren't in the runbook; Risks and gaps (flagged); What wasn't covered (flagged).

From the manager's summary, ready before the last day.

Who to interview

The engineer who has carried the service longest. Whoever gets called when a deploy goes wrong at night. The person new joiners are told to ask instead of reading the wiki.

The reasoning ships with the system this time.

Questions engineering teams ask

  • Documentation is what people were asked to write. OakHive interviews them instead: what they were responsible for, what breaks and how they fix it, the decisions they carried in their head, and who to call for what. It follows up when an answer is vague, which a template can't.

  • The interview. Answers in the chat come from what the departing engineer said, and the manager's summary lists what wasn't covered. No AI is perfect, so the documents show you where the interview ran thin.

  • The week someone resigns. From then on the clock runs on their notice period, and every week that passes is a week of answers you can't get back. Setup is their name and role.

All questions

Bring us the questionyour company keepsanswering twice.

Thirty minutes, no deck. You describe something your team has explained more than once, and we show you what OakHive does with it.

We are taking on a small number of companies at a time, so this is a conversation rather than a signup.

We reply within two working days.

What you send reaches the people who read these themselves, and is used to reply to you and prepare for the conversation. It is kept for 24 months, is never sold, and never feeds an advertising system. There is none here. Privacy Policy.