What happens to AI-built tools when the person who built them leaves

Every company has a version of this story. The operations coordinator who built the scheduling spreadsheet everyone depends on gives notice. On her last day someone asks how the formulas work, and nobody else knows. A week later the spreadsheet breaks and the team goes back to the whiteboard.

AI makes this story more common and more serious. It is now easy for one motivated employee to build a working app in an afternoon. That is good news. But if the app was built in their own account, on their own card, with only their own understanding of how it works, the company does not really have the tool. It has borrowed it. This guide explains what happens to AI-built tools when their builder leaves, why it happens, and the plan that keeps every tool working and owned by the company.

Why this is a bigger problem now

Employees are not waiting for permission to use AI. Microsoft and LinkedIn's 2024 Work Trend Index found that 78% of AI users bring their own AI tools to work. Deloitte's 2026 UK survey found that 63% of UK working adults knowingly use GenAI for work, and 31% of those users do so without their employer's knowledge. PagerDuty's 2026 survey found that two-thirds of office professionals who use AI at work have used AI tools they believed were not permitted.

Many of those tools now build software, not just answer questions. Pax8's Q3 2026 survey of small and mid-size businesses found that 67% of AI-engaged businesses have built one or more AI solutions now in regular use, and 35% discovered shadow AI, meaning AI in use without company approval, in the past six months alone. Some of those solutions live in company accounts. Many live in someone's personal login.

When the person stays, a personal-account tool looks harmless. When they leave, it becomes a problem in four ways at once.

The four ways a tool leaves with its builder

1. The account belongs to the person

Most AI app builders are sold to individuals first. Someone signs up with an email address, builds an app, and shares a link. If that email is personal, or the workspace is under their name, the app, its data and its settings belong to that account. The company may have a claim on paper. It does not have the login.

2. The bill belongs to the person

Many AI builders charge by plan and by usage, often on whatever card the builder entered. When the builder leaves and cancels, or the card stops paying, hosting can stop, credits run out, and the app goes dark. Some platforms also tie hosting to the same balance. Lovable's pricing page, for example, says its credits pay for building, hosting with Cloud and AI features inside your apps from one balance. That model works fine for a company account. On a former employee's personal account, it means the company's tool runs on someone else's balance.

3. The knowledge belongs to the person

Even inside a company account, a tool that only one person understands is fragile. Why does this field exist? Why does that approval skip the manager on Fridays? Without a record of how the tool was built and changed, the next person has to guess.

4. The access stays with the person

The most serious risk is the quietest. A former employee who still has a login to a personal-account tool may still be able to see customer records, schedules, prices or employee details inside it. RedAccess reported in May 2026 that it reviewed about 380,000 vibe-coded assets, about 5,000 were built for corporate purposes, and 40% of those exposed sensitive data. Tools built outside company control are the hardest to find and the hardest to lock down.

What each kind of setup means at offboarding

Check the admin controls of whatever platform your team uses. Some put the controls a company needs at offboarding on their top tier. Lovable's plan documentation, for example, lists SCIM, the standard way to add and remove users automatically from a central directory, and audit logs as Enterprise only. That is not unusual in the category, and it is worth knowing before you choose.

The offboarding plan for AI-built tools

Use this plan every time someone who built or owns a tool leaves, or changes roles.

Before their last day

1. List every tool they built or own. Pull it from your company workspace if you have one. Ask them directly too, including anything built in a personal account.
2. Assign a new owner for each tool. Pick someone who uses it and understands the work it supports. Name a backup as well.
3. Hold a 30-minute handover per important tool. The builder walks the new owner through what it does, who uses it, what breaks, and any rules that live only in their head.
4. Move anything outside the company workspace in. If a tool lives in a personal account, rebuild or transfer it into the company workspace now, while the builder is still there to explain it.
5. Write a one-paragraph description of each tool. What it does, who uses it, where its data comes from, and who owns it.

On their last day

1. Remove their access in one action. Their tools should stay. Only their login should go.
2. Rotate anything they could see. Shared passwords, connection keys to other systems, and any login stored in a tool they managed.
3. Confirm no company data sits in their personal accounts. Ask, and record the answer.

The week after

1. Review the change record. Look at their recent changes to each tool. Anything unexpected gets a second look.
2. Check the tools still work. Have the new owner run each tool's main workflow end to end.
3. Update your tool list. New owner, new backup, date of handover.

How to prevent the problem before anyone leaves

Offboarding is the moment the problem shows up. The fix is in how tools are built from day one.

• One approved place to build. Give people a company workspace where building is easy, so they have no reason to use a personal account. Banning AI without offering a better option pushes it into the shadows.
• Every tool has an owner and a backup. No tool goes live with only one person who understands it.
• Roles, not shared logins. Decide who can build, who can change tools, and who only uses them. Shared logins make offboarding impossible to do cleanly.
• A record of every change. Who asked for what, what changed, and when. This is what lets a new owner pick up a tool with confidence.
• A living tool list. One page listing every tool, its owner, backup and users. Review it each quarter.
• Company billing only. Tools that the business relies on are paid for by the business, with spending caps and alerts the owner sees.

Example: a 40-person property management company discovers, during an exit interview, that the maintenance coordinator built the vendor dispatch app the whole office uses in a personal AI account. The app holds tenant names, unit numbers and vendor rates. Following the plan above, the company spends the coordinator's last week rebuilding the app in its own workspace with the coordinator in the room, assigns the office manager as owner and a property manager as backup, and removes the coordinator's access on the last day. The personal-account app is retired with its data deleted, and the coordinator confirms it in writing.

How to talk to your team about it

Builders are often the most valuable people on the team. They saw a problem and fixed it without being asked. The way you raise ownership should reward that, not punish it.

• Lead with thanks. "This tool saves the office hours every week. We want it to keep working no matter who is here."
• Make the inventory no-blame. If listing a personal-account tool gets someone in trouble, nobody lists anything, and the risk stays hidden.
• Offer a better home, not a ban. Moving a tool into the company workspace should feel like a promotion for the tool: proper roles, proper billing, a backup owner.
• Credit the builder. Keep their name on the record as the original builder. People build more when their work is seen.
• Set the rule going forward. From now on, tools the business relies on are built in the company workspace. Say it once, plainly, and make it easy to follow.

Problems the category ships with

AI builders are designed to let one person go from idea to working app quickly. That strength creates the single-builder risk this guide is about, and some of the category's public incidents show what happens when control is loose. In July 2025, Fortune reported that a Replit AI agent wiped SaaStr's live database, holding data on more than 1,200 executives and over 1,190 companies, during a code freeze. Replit responded by promising automatic separation of development and production databases and a planning-only mode. The lesson for offboarding is simple: when it is unclear who controls a tool and what can change it, problems are harder to prevent and harder to trace.

Offboarding checklist for AI-built tools

1. Pull the list of every tool the person built, owns or changed.
2. Ask directly about tools in personal accounts.
3. Assign a new owner and a backup for each tool.
4. Hold a handover for each important tool and write its one-paragraph description.
5. Move any personal-account tool into the company workspace.
6. Remove the person's access in one action on their last day.
7. Rotate shared passwords and connection keys they could see.
8. Confirm in writing that no company data remains in their personal accounts.
9. Review their recent changes in the change record.
10. Run each tool's main workflow with the new owner and update the tool list.

How FORGE handles it

FORGE is built so tools belong to the company from the first minute. Every app lives inside your company, not inside a person's account, and people join only by invite. Owners, admins, builders and app users each have a role that decides what they can see and change, so there are no shared logins to untangle.

When someone leaves, removing them ends their access immediately and every tool they built stays with the company. A complete record of who did what shows the new owner how each tool was built and changed, so the knowledge does not walk out the door. Every piece of every tool is checked for quality and security before it is saved, which means the tool a new owner inherits is one they can trust.

Billing is the company's, not a person's: $49 a seat a month with 3 builds, app users free, and caps and alerts at 80 and 100 percent so nothing bills by itself. Our team onboards every company with a workflow workshop, and we set up roles and ownership with you from the start. Read about security, see how it works, or talk to us.

Questions people ask

Sources

1. Pax8: Q3 2026 SMB AI Pulse Report
2. Deloitte UK: GenAI Workforce Survey press release (Sept 2026)
3. PagerDuty: shadow AI workplace survey 2026
4. Microsoft Work Trend Index 2024: AI at work is here
5. RedAccess: shadow AI builders security report
6. Lovable pricing
7. Lovable plans (docs)
8. Fortune: Replit AI agent wiped a company's database (July 2025)