Uncovering
AI UX Audit
Uncovering — A specialized audit of the AI layer of your B2B SaaS product, whether it's already in production or about to launch.
Discover what isn't working in your AI experience, why, and what to change.
The problem
You shipped the AI feature. And something isn't clicking.
- —Users aren't using the feature as much as you expected.
- —The agent answers correctly, but the product experience doesn't follow.
- —Adoption numbers are below plan, and it's not clear whether the problem is the AI or what surrounds it.
- —Your product works, but the AI feels disconnected from the rest.
The problem is rarely the model. It's how the interaction is designed, and everything around it.
The solution
A specialized audit of your product's AI layer.
Not a generic UX audit. Not a classic flow review. A focused look at the architecture decisions that determine whether your AI gets used, understood, and trusted by your users.
What I analyze
Your AI lives inside a product. The audit looks at both — because one depends on the other.
The AI experience
Flows, interactions, tone, handoffs, failure modes.
The experience around the AI
Onboarding, discovery, where the feature lives in the product, how it's presented to the user.
Trust signals
How the product communicates certainty, uncertainty, and limits.
Adoption friction
Where users abandon, avoid, or distrust the system.
Coherence over time
Whether the agent remembers, forgets, or contradicts itself in ways that confuse users.
Capability vs. interface
The gap between what the AI can do and what the interface allows.
How I audit
A 40+ criteria framework, applied to your real product.
Each audit combines classic UX heuristics reinterpreted for AI interactions, AI-specific heuristics that don't exist in traditional UX, and transversal dimensions that affect the entire experience.
Some of the questions I answer:
Does the agent appropriately communicate when it's certain, when it doubts, and when it doesn't know?
Can the user interrupt, correct, and reject suggestions without losing the thread?
Does the UI promise things the model can't deliver, or hide capabilities it does have?
Does the onboarding explain what the agent can and can't do before the user hits the limits?
Do human-AI handoffs preserve context, or does the user have to repeat everything?
Is the agent's memory appropriate — neither too much nor too little — for this product?
The audit isn't done by trying the product for a few minutes. It includes:
- ✓Free exploration by every relevant role — operator, manager, admin.
- ✓Directed use cases — 8 to 15 representative tasks of real use, walked through systematically.
- ✓AI-specific adversarial tests — ambiguous inputs, contradictions, edge cases, system-breaking attempts.
- ✓Review of the experience around the AI — onboarding, discoverability, integration with the rest of the product.
- ✓Granular documentation of each finding, with evidence, severity, and concrete recommendation.
Deliverables
Deliverables that get acted on, not archived.
Prioritized report
Findings ordered by impact, with concrete next steps for each. Not a document that gets filed — one that gets used.
Live walkthrough
We walk through the findings together. You ask questions, we discuss, we adjust. Your team leaves with clarity, not with a PDF.
Functional prototype (when applicable)
Built with Claude Code and the actual model. Clickable. Real behavior, not a mockup. So you see the recommendation before implementing it.
Format and pricing
One audit, one clear scope.
USD 1,500 · 2 to 3 weeks
- ✓40+ criteria framework applied to your real product
- ✓Prioritized findings report
- ✓90-minute live walkthrough
- ✓Functional prototype when applicable
- ✓One week of email follow-up support
The audit is also the lowest-risk way to start working together. Most engagements begin here and grow into Feature Design or Fractional ownership once we both see how the work fits.
Book a scoping call →How it works
Five steps, from the first call to follow-up.
Scoping call (free, 30 min)
We talk about the product and the problem. I confirm whether the audit fits and whether prototype applies. Book now →
Proposal and kickoff
Within 24-48 hours you receive the proposal with scope and timeline. On signing, we start.
Audit
I investigate the product, the flows, and the AI layer in depth, following the framework.
Delivery
Report, walkthrough, and prototype when applicable.
Follow-up (included)
One week of email support for questions while you act on the recommendations.
Why me
I'm not a UX designer who also looks at AI.
20+ years
Designing and building software.
8 years
UX as an explicit discipline.
AI + product
My current focus, in the intersection of both.
I'm not a UX designer who "also looks at AI." I'm not an engineer who does wireframes. I design AI interactions that work in reality, not just on screen, and I prototype them with the actual model before you write production code.
See case study → →FAQ
What people usually ask before starting.
What if the prototype doesn't apply to my case?
I replace it with a deliverable that does move the needle: detailed flow diagrams, interaction specs, or strategy documents. We decide together in the scoping call.
Do you work with companies outside Latin America?
Yes. The scoping call and deliveries adapt to the team's time zone. Most asynchronous work happens in writing.
What if my audit needs more than 3 weeks of work?
For products with multiple AI features or significant complexity, we can scope a custom engagement. I'll let you know in the scoping call if that's the case, and we agree on scope and pricing before starting.
What if I need to design AI from scratch, not audit something existing?
That's a different engagement, closer to AI Feature Design than Audit. Let's discuss it on the scoping call — I can guide you even if it doesn't fit Uncovering exactly.
What access do you need to the product?
Test users for each relevant role, with representative data loaded. URL of the environment to audit (staging or production), and a point of contact for specific questions during the audit. Details we agree on at project start.
Start with a conversation.
30 minutes. Free. No commitment.
I'll tell you if I can help, what the audit will cover, and what to expect.
Book your scoping call →