A developer builds a useful AI tool, then moves on. Suddenly nobody knows where the code is, which account pays for the AI API, or how to change the prompts. This happens more often than you might think — and it is completely avoidable.
What "your AI system" actually includes
- Source code and its repository
- Prompts and configuration — often the most valuable part
- Data: documents, training examples, embeddings and logs
- Accounts: cloud hosting, AI provider, domain and email services
- API keys and secrets
- Trained models and fine-tuning files, if any
- Documentation on how it all fits together
Common ownership traps
- The repository lives in the developer's personal account.
- The AI and cloud accounts are on the developer's credit card.
- No written agreement transfers intellectual property to you.
- Prompts and settings exist only on the developer's laptop.
- Nothing is documented.
How to protect your business
| Asset | Best practice |
|---|---|
| Code | Company-owned GitHub/GitLab organisation; developer added as a member |
| Cloud & AI accounts | Created and billed in your company's name |
| Secrets | Stored in a password manager or secrets vault you control |
| IP | Contract or NDA with explicit transfer of ownership |
| Knowledge | README, architecture notes and a handover session |
The handover checklist
- Access to every repository, account and dashboard confirmed
- Keys rotated after the developer leaves
- Deployment steps documented and tested by someone else
- A short recorded walkthrough of the system
If you cannot deploy, change and pay for it without the original developer, you do not fully own it yet.
At Appri Infotech we work in client-owned repositories and accounts, sign NDAs and transfer full ownership as standard. See how our dedicated developers work.