Employee-built software needs a named owner and a named successor from the moment other people depend on it, because visibility tells you an app exists and ownership tells you who fixes it. Chris Stokel-Walker reports in a recent article that engineering teams are inheriting tools they never built, and this piece follows that one claim: what happens to the software after the person who made it leaves.
Stokel-Walker reports that engineering inherits what employees build
Stokel-Walker describes how AI coding tools let people with no software background build their own applications, and how Ramp and Stripe have built internal AI tools that let employees outside engineering create applications from company data. He cites a Retool survey in which 60% of builders had made something outside IT oversight in the past year. Christina Troy, senior manager of IT at Tines, tells him that IT teams “often only find out about the software and the connectors used when something breaks.” Tines’ answer is a platform where employees build and IT and security teams can see which apps reach which tools and data.
He then raises the case this piece takes up. When the employee who built a tool leaves and the tool has become part of how the business runs, the responsibility tends to fall on an engineering team that had no part in building it and would not have designed it that way. Instead of clearing its own backlog, that team starts a new one. He also reports research by Edona Elshan of the Vrije Universiteit Amsterdam finding friction between professional developers and the employees now building for themselves.
Visibility answers what exists, and ownership answers who is accountable
The Tines approach Stokel-Walker describes solves a discovery problem. A platform that shows which apps have access to which data lets IT find duplication and risky connectors before an incident. It does not, on its own, say whose job it is to patch an app when a dependency breaks, rotate its credentials, or retire it. Those are ownership questions, and the inheritance problem in the article is what happens when nobody answers them while the builder is still around.
Security standards drew this line long before AI coding tools arrived. NIST SP 800-53 Revision 5 treats the inventory and the owner as separate requirements. Control CM-8 asks organizations to keep an inventory of system components that accurately reflects the system and includes all of its components. Its enhancement CM-8(4), titled Accountability Information, goes further: the inventory must include a means of identifying, by name, position or role, the individuals responsible and accountable for administering those components. A list of apps satisfies the first requirement. Only a recorded owner satisfies the second.
An owner record for employee-built software needs a successor field
The departure case shows why a single owner name is not enough. When the owner is the builder, the record expires on the builder’s last day. CM-8(4) allows accountability to be recorded by position or role, which suggests a practical fix: record both the person who built the tool and the role that takes it over, such as the builder’s manager or the business unit that relies on it. When someone leaves, offboarding then includes a step that already exists for laptops and access badges: transfer every tool they own to the named successor, or schedule it for retirement.
The default owner should sit in the business unit, with engineering taking over a tool only by explicit decision. If engineering accepts a tool, the decision should come with the hours to maintain it, because otherwise the new backlog Stokel-Walker describes arrives with no budget attached. If engineering declines, the business unit keeps the tool and its risk, or retires it. Either way the decision is recorded before the builder leaves, when the person who understands the tool can still explain it.
Ownership rules for employee-built software fail if they reach every script
There is a boundary to this argument. Retool’s own AI build vs. buy report, based on a survey of 817 builders conducted in late 2025, found that 60% had built outside IT oversight and 25% did so frequently. Only 17% said they had been blocked from building something IT had not sanctioned, and the report itself calls that figure deceptively low because IT cannot block builds it does not know about. Rules that require paperwork for every personal spreadsheet would push more building out of sight, which is the behavior Elshan describes when she says people slip away from compliance guidelines.
The owner record should therefore attach at a trigger. A reasonable set of triggers is: the tool gains a user other than its builder, it reads or writes production data, or it holds credentials to a company system. Below that line, building stays free. Above it, the tool gets an owner, a successor role, and an entry in the inventory. This is the same principle I argued in AI governance is now an engineering problem, published in August 2026: controls that run inside the workflow hold up, and controls that live only in policy documents do not. Here the workflow is the build platform and the offboarding checklist.
Why This Matters for Engineering Leadership
Engineering leaders are the people most likely to receive orphaned employee-built software, so they have the strongest reason to insist on ownership while builders are still in their roles. The practical request to make of IT, security and business leaders is small: add an owner field and a successor role to whatever inventory already exists, set the triggers that make them mandatory, and add tool transfer to offboarding. Engineering then inherits software by decision, with time budgeted for it, instead of by default on the day something breaks.
FAQ
What is employee-built software?
It is software made by people outside engineering, often with AI coding tools, for their own team’s work. Internal apps, automated workflows and dashboards are common examples.
How common is software built outside IT oversight?
In a Retool survey of 817 builders conducted in late 2025, 60% said they had built something outside IT oversight in the past year and 25% said they did so frequently.
Why is visibility not enough?
A visibility platform shows which apps exist and what data they touch. It does not say who is accountable for fixing an app, or who takes it over when its builder leaves.
What does NIST say about component ownership?
NIST SP 800-53 Revision 5 control CM-8(4) asks organizations to record, in the component inventory, a way to identify the individuals responsible and accountable for administering each component.
Does every spreadsheet or script need an owner record?
No. This piece argues for a trigger: record an owner and a successor once a tool gains users beyond its builder, touches production data, or holds credentials to company systems.
Who should own employee-built software?
The builder’s own manager or business unit should own it by default. Engineering takes ownership only by an explicit decision, with the time to maintain it budgeted.
