Everyone claims to "do AI" now. The gap between a partner that ships trustworthy, production-grade software and one that ships an impressive demo is enormous. These are the questions that reveal which is which.
Questions to ask
- 1Who owns the code and IP? (The answer should be: you, 100%.)
- 2How do you guarantee quality when AI writes code? (Look for human review, tests, security.)
- 3How do you prevent AI hallucinations in features? (Retrieval grounding, evals, guardrails.)
- 4Can I see working software weekly? (Real partners demo continuously.)
- 5How do you handle security and our private data?
- 6What happens after launch — support, monitoring, iteration?
Green flags
They talk about architecture, testing and security — not just speed. They show you real, shipped products and measurable outcomes. They are clear that AI accelerates the work but humans own the quality. They give you full ownership and no lock-in.
Red flags
"The AI writes everything" with no mention of review or testing. No portfolio, no references, no measurable results. Vague answers on IP ownership or data handling.
#AI#Hiring#Business#Engineering
EG
Elena García
Head of AI
Leads generative-AI engineering at CRUDTree — RAG, agents, evals and the unglamorous work that makes AI features trustworthy.