The short answer for buyers
A risky AI consulting retainer has unclear outcomes, open-ended hours, vague ownership, weak exit terms, unapproved expenses or support promises with no response boundaries. A retainer can be sensible for ongoing work, but only when both sides can see what is included, reviewed and cancellable.
"Who owns the models
"Who owns the models, code, and data produced during the engagement?"
This is the single most important question. Some consultants retain ownership of everything they build, meaning you are effectively renting your own AI infrastructure. Others transfer full IP ownership to you.
What you want to hear: "You own everything we build for you. Full IP transfer on completion."
Red flag: "We retain ownership of the underlying framework, but you license it." This creates vendor lock-in and means you cannot take your AI assets to another provider.
"Can I take everything you build and hand it to another team?"
Related to ownership, but subtly different. Even if you own the IP, some systems are built in ways that only the original consultant can maintain. Ask specifically about documentation, code quality, and whether another competent team could pick up where they left off.
"How will we measure whether this is working?" Vague promises like "improved efficiency" or "better customer experience" are not enough
"How will we measure whether this is working?"
Vague promises like "improved efficiency" or "better customer experience" are not enough. Push for specific, measurable outcomes:
- What metrics will we track?
- What is the baseline we are measuring against?
- What does success look like at 3, 6, and 12 months?
- How often will we review performance data together?
A good consultant will help you define these upfront. A poor one will resist measurable commitments because they know they cannot deliver.
"What does your typical ROI timeline look like?"
Honest answer: most AI projects take 3-6 months to show meaningful ROI. Anyone promising returns in weeks is either overselling or doing something trivially simple that you could probably do yourself with ChatGPT.
"What happens when the retainer ends?" This question reveals more about a consultant's integrity than almost any other
"What happens when the retainer ends?"
This question reveals more about a consultant's integrity than almost any other. You want clear answers on:
- Transition support: Will they help hand over to your internal team or another provider?
- Documentation: Will you receive complete documentation of everything built?
- Data export: Can you export all your data in standard formats?
- Continued access: Will your systems keep running after the relationship ends?
If the answer to any of these is unclear or evasive, you are being set up for dependency, not partnership.
"What is the minimum commitment, and what is the cancellation process?"
Standard retainers in the UK AI consulting market typically require 3-month minimum commitments with 30-day cancellation notice. Anything longer than 6 months should come with exceptional justification and clear milestones.
Be wary of long lock-in periods with no performance clauses. If the consultant is confident in their work, they should be comfortable with reasonable exit terms.
"Are you building on open-source or proprietary technology?" Neither is inherently better
"Are you building on open-source or proprietary technology?"
Neither is inherently better, but you need to understand the implications:
- Open-source: More portable, larger talent pool for future maintenance, no licensing fees. But may require more customisation effort.
- Proprietary: May offer faster deployment and specialised capabilities. But creates dependency on the vendor and potentially ongoing licence costs.
The best consultants use a mix and are transparent about why they chose each component.
"What happens if the AI models you are using become unavailable or change pricing?"
This is a real risk. OpenAI, Anthropic, and Google regularly change their APIs, pricing, and capabilities. A good consultant will have contingency plans and architecture that is not tightly coupled to a single provider.
"Who specifically will work on our account?" Consultancies sometimes sell with senior partners and deliver with juniors
"Who specifically will work on our account?"
Consultancies sometimes sell with senior partners and deliver with juniors. Ask to meet the actual team who will do the work, not just the people in the sales meeting.
"What is your experience in our industry?"
AI consulting requires both technical expertise and domain knowledge. A consultant who has built AI systems for financial services may not understand the compliance requirements of healthcare, or the operational realities of manufacturing.
That said, do not dismiss consultants who are new to your sector if they demonstrate strong technical capability and genuine curiosity about your business. Sometimes fresh eyes spot opportunities that industry insiders miss.
"What exactly is included in the retainer fee?" Get specifics
"What exactly is included in the retainer fee?"
Get specifics. How many hours of work? Which team members? What about hosting costs, API fees, and third-party software? Are there any costs that sit outside the retainer?
"Can you show me case studies with verifiable results?"
Not testimonials. Case studies with specific metrics. Revenue increased by X%, processing time reduced by Y hours, cost saved of Z pounds. Ideally with client references you can contact directly.
An AI consulting retainer is not always the right move
An AI consulting retainer is not always the right move. Consider whether:
- You do not have a clear business problem: If you want AI because it is trendy, not because you have identified a specific opportunity, you are not ready for a retainer. Start with a one-off strategy session instead.
- You do not have internal capacity to engage: AI consulting requires your team's time and attention. If nobody internally can commit to working with the consultant, the engagement will stall.
- You need a one-off project, not ongoing support: A fixed-scope project with clear deliverables may be more appropriate and cost-effective than an open-ended retainer.
The right AI consultant will welcome every question on this list
The right AI consultant will welcome every question on this list. They will have clear, specific answers and will not pressure you to sign quickly. If a consultant gets defensive when you ask about ownership, exit terms, or measurable outcomes, that is the most valuable answer you will get.
Take your time. Do your due diligence. And remember that the best engagements are partnerships built on transparency, not contracts designed to lock you in.
Related buying guidance
For the related buying decision, read What are the red flags I should look for in an AI agency contract?.
For the related buying decision, read What are the red flags to check before hiring an AI consultant?.
Browse the problems guidance pillar for more buyer questions.

