Why AppThentic
Every company runs on decisions nobody wrote down. Here is what that costs, and what AppThentic changes.
Your systems record what happens. They do not record why: who decided, what they were protecting, and whether that reason still holds.
- payouts/scheduler.py:58Seller payouts wait 24 hours after deliveryWhy? Unknown
- risk/rules.ts:112Payouts above ₹50,000 wait for a manual checkWhy? Unknown
- refunds/policy.py:42Refunds close 30 days after deliveryWhy? Unknown
Each one is a single line of code. The reason behind it is in nobody's code.
That reason lives in people’s heads, in a Slack thread from 2021, in a ticket closed by someone who left. When it is lost, every change to the rule becomes a guess.
What it costs
For the business
| Finding | Source |
|---|---|
| The median employee has been with their employer 3.9 years | US Bureau of Labor Statistics, January 2024 |
| 33% of staff at acquired tech companies leave in their first year, against 12% of comparable hires | Daniel Kim, MIT, “Predictable Exodus” |
| Knight Capital lost about $460 million in 45 minutes when new code reused an old flag whose meaning nobody active remembered | US SEC, 2013 |
| Westpac paid A$113 million in penalties across six cases, including charging interest above what its contracts allowed. The court called systems failures “a common feature” of all six | ASIC, 2022 |
For engineering
| Finding | Source |
|---|---|
| Developers spend about 58% of their time understanding existing code, not writing new code | Xia et al., IEEE Transactions on Software Engineering, 2018 |
| Finding information is the top source of lost time for developers; half lose 10 or more hours a week | Atlassian, State of Developer Experience, 2025 |
| More than two-thirds of developers hit a knowledge silo at least once a week | Stack Overflow Developer Survey, 2022 |
Put together: the people who know why a system works the way it does will mostly have moved on within four years, the rules they left behind carry real money, and the people who remain spend more of their week reconstructing intent than building.
Why the usual fixes don’t hold
Documentation goes stale. It is written once, at the moment people have the least time, and nothing tells anyone when it stops being true.
Wikis and search find documents, not decisions. They return the page that mentions refunds, not the reason the window is 30 days.
AI assistants read code well, and still cannot know the reason. A model can tell you that seller payouts wait 24 hours. It cannot tell you that the hold covers the payment gateway’s fraud check, which can take up to a day to flag an order, because that reason was never written down.
Handover documents list systems, not reasons. The leaving engineer writes what they owned. The why behind each rule leaves with them.
What AppThentic does instead
AppThentic treats every business rule as a fact that needs a source, an owner and an approval.
- It reads your systems: code, pull requests, tickets, docs, threads, support tickets and schemas. It finds the rules that decide what customers pay, get and are allowed to do.
- It asks the person who decided: a few specific questions they answer with a typed line or a voice note, or a call where they share their screen and walk om through the flow. It follows the answer to whoever actually knows. See how AppThentic collects answers.
- It keeps the answer with its source, owner, date and status, and asks again when the system or the reason changes.
The result is a memory of why your business runs the way it does, that anyone can query before they change something.
Who uses it
- Executives call om and get answers about how the business actually works, confirmed by the people who own them.
- Security, risk and audit teams get a trail of who decided what, and when.
- Engineering leaders see which parts of the system are understood, and which depend on one person.
- Engineers check a rule’s reason and owner before they change it.
Next: Quickstart shows how a team rolls AppThentic out.