Depth Handover Pack
Someone just resigned.
What leaves with them?
Depth finds the code only they understood, captures what the code can’t say, and confirms someone else can run it, before their last day.
$300 per person handed over, on any plan. Refundable until the work begins.
- Payments retry · criticalNobody after FriSession with PriyaBen · not yet
- Ledger · criticalAt risk after FriSession with PriyaCara · not yet
- RefundsNobody after FriSession with PriyaBen · not yet
- Data exportAt risk after FriSession with PriyaDev · not yet
The numbers
People leave all the time. What they knew rarely gets measured.
3.6×
the average knowledge loss, in the worst periods, as people leave large projects.
17%
less understanding of what they'd just built, for engineers coding with AI.
What a handover usually looks like
A two-week notice, as engineers describe it.
Day 1
The resignation. “Can you write a handover doc?”
Day 3
Two hour-long walkthrough calls, recorded. Nobody watches them twice.
Day 8
The doc is half done, and most of it says what the code already says.
Last day
Laptop returned. Accounts closed, and with them, sometimes, the recordings.
Day 40
3am. A failure the doc doesn't cover. The person who knew is gone.
Day 43
A senior engineer has spent three days reverse-engineering it, and is now the only one who knows.
“When somebody explains to you, you think you understood everything.”
A development manager, in a McGill study of 27 developers and managers on what happens when people leave (Robillard, FSE 2021). The timeline above is a composite of what they described.
It happens to the best-known code there is
Before Heartbleed in 2014, OpenSSL, which secures much of the web, had one full-time developer and about $2,000 a year in donations.
Most teams have a smaller version of this in their own code. The question is whether you find out before the person who holds it leaves, or after. (MIT Technology Review)
Why handovers fail
Not for lack of effort. For lack of proof.
Nobody knows what's at risk.
The handover covers what the person leaving happens to remember, not where the risk is.
Docs hold the what, not the why.
Anyone can read what code does. What leaves is why it's this way, what was tried and dropped, and what must never change.
Nobody checks it worked.
People overrate their understanding until they try to explain it. Handovers end when the meetings do.
Two weeks is short. Layoffs give none.
And moving teams or a long leave are rarely treated as handovers at all, though the knowledge goes all the same.
How Depth Handover works
A handover that ends with proof, not a doc.
01
Plan from evidence
In about a minute: the modules whose status drops when they go, riskiest first, and who should take each over, with the reasons. Two successors on critical modules, so no one becomes the next single point of failure.
Payments retryNobody after FriBen, CaraLedgerAt risk after FriCaraBen reviewed 14 of Priya’s changes here · fits in 9 working days
02
Capture what the code can't say
A short written session per module, about 20 minutes, with questions grounded in their own code: intent, history, landmines, what to check at 3am. Depth checks every answer against the code and writes a brief they approve.
Why it’s this way · retry.ts:31–44
You stop after three attempts and flag the order for review. Why three, and why review instead of failing the checkout?
03
Learn by doing
Successors follow a walkthrough of the changes that show the module best, explaining each step back in their own words, and get a 20-minute agenda for the person leaving from whatever they weren't sure about.
2How a retry stays safe
Every attempt sends the same idempotency key. Priya flagged this as the module’s main landmine.
Got it
04
Verified before the last day
Handover checks and a 3am incident scenario, answered in their own words. When they get there, the module turns Healthy again on the map, and the report shows what was handed over, to whom, and what's still at risk.
3am check
Charges for one bank have failed since 02:10. Where do you look first?
Solid.Ben is covered on Payments retry.
Built for the people doing it
Respectful of the person leaving. Useful to the people staying.
For the person leaving
- About 20 minutes per module, written whenever suits them. Skip anything.
- Their name on every brief, and nothing shared until they approve it.
- They see everything their manager sees about the handover.
- Never used to judge anyone, least of all on their way out.
For the people taking over
- A brief and a walkthrough instead of an hour-long recording.
- A pairing agenda built from whatever they weren't sure about.
- Answers private to them: managers see covered or not yet.
- Refreshes at 30 and 60 days, so it sticks.
Not just resignations
Every way knowledge walks out.
Leaving
Resigning, retiring, a contract ending: the full handover.
Moving teams
Knowledge fades from people who move on, even when they stay reachable.
Long leave
Parental, medical, a sabbatical: a lighter cover plan, and their own standing kept.
Already gone
A layoff or a sudden exit: briefs rebuilt from pull requests and reviews, clearly labelled.
Handover Pack
$300
per person handed over, one-off, on any plan
- Every module they were key to
- Up to two successors per module
- A session and an approved brief per module
- Walkthroughs with explain-backs
- Handover checks with a 3am scenario
- Refreshes at 30 and 60 days
- A daily digest and the final report
- Briefs kept as the module's docs
Charged in rupees through Razorpay at the day’s rate, excluding taxes. Refundable in full until the first session starts (policy). For comparison: a new engineer takes about 4.8 months to reach normal productivity.
Questions
What people ask first.
Does the person leaving have to take part?
It works best when they do, at about 20 minutes per module, written whenever suits them. If they'd rather not, or they've already gone, Depth rebuilds the brief from the code, pull requests and reviews, and says so on the brief.
Do you store our code?
No. Code is read from GitHub inside the step that needs it and never stored. Briefs and walkthroughs hold prose and links to exact lines at a commit.
Who sees what?
The manager sees the plan, each module's status, and covered or not yet for each successor. The person leaving sees all of that too. Successors' answers stay private to them, and session answers become a brief only after the person leaving approves it.
How long does it take?
About a minute to plan, about 20 minutes per module for the person leaving, and about 45 minutes per module for each successor. It's built to fit a two-week notice, and the plan tells you on day one if yours won't.
Do we need Slack?
Slack carries the nudges and the manager's daily digest. The work itself happens in Depth, in the browser, so it works without Slack too.
Which AI writes it, and is our data used for training?
Anthropic's most capable models write what people read and review the answers, and every claim in a brief is checked against the code before anyone sees it. Customer data is never used to train models.
What happens to the briefs afterwards?
They stay on the module's page in Depth as its docs, marked with how much the code has changed since, so the next person to join benefits too.
Someone resigning? Start their handover today.
About a minute to plan. Everyone involved hears about it in Slack, and the person leaving sees everything you see.
Book a 30‑minute demo
We’ll walk through a handover on a repository like yours.