# How to choose an AI engineering partner
Choose a partner on evidence that they can take an AI system into production in an environment like yours, not on the quality of their demonstration.
## Define the outcome before the shortlist
For a leader commissioning a first production AI system, start by writing down the decision or task the system will support, who will use it and what must be true for it to count as working. Include the data it will rely on and the review it will need. Partners can only be compared against a defined outcome.
## Ask for production evidence
Ask each partner to show a system they built that is still in use, and to explain how it is monitored, who owns it and what happened when it failed. A polished prototype shows that a model can do a task. It does not show that the partner can handle access controls, exceptions, evaluation and support.
## Meet the people who will do the work
Ask who will build the system and how much of their time it will have. Senior people who appear in the sales process and then leave are a common source of delay. A partner whose engineers work directly with your operators will usually find the real constraints sooner than one that works from a written specification.
## Test their approach to governance
Ask how they decide where human review is needed, how they record the evidence behind an output and how they evaluate the system before and after release. In regulated work these are acceptance criteria, not documentation to add at the end. A partner who cannot answer in specific terms is unlikely to build them in.
## Agree ownership and exit at the start
Confirm who owns the code, models, prompts and data, where the system will run and what your team will receive at handover. Ask for a plan that leaves your people able to operate and change the system. A short paid discovery phase with a defined output is a lower-risk way to test the working relationship than a large first contract.
## Related reading
- [How to put document AI into an operational workflow](/solutions/guides/implementing-document-ai)
- [Document AI vs OCR: what does your workflow need?](/solutions/guides/document-ai-vs-ocr)
- [AI consultancy or an in-house AI team?](/solutions/guides/ai-consultancy-or-in-house-team)
- [Document processing for lenders](/solutions/lending-document-processing)
- [Legal case intelligence](/solutions/legal-case-intelligence)
- [AI knowledge search for professional services](/solutions/professional-services-knowledge-search)
- [AI-assisted case triage and review](/solutions/ai-case-triage)
[Discuss your requirements](/contact?solution=choosing-an-ai-engineering-partner)
HTML: https://rodan.io/solutions/guides/choosing-an-ai-engineering-partner