Your users already have ChatGPT open in the next tab.
So why would they use your AI feature? I find out where it loses them to the general assistants, then fix it with behavioral design, conversion work and code.
The free Leak Check takes your AI feature’s usage numbers and shows you right away where users drop off and how many more weekly users a fix could bring.
The Next-Tab Test
Every AI feature now competes with an assistant your users already pay for. Whether your AI works is only half the question. The other half: for this job, would your users rather switch tabs?
I run your feature's real jobs through ChatGPT, Claude and Gemini and score every answer blind against what your users would accept: the answers are shuffled and unlabeled while I score them. Then I check the five points where AI features lose people.
Find
Users notice the feature at the moment they need it.
First win
The first try produces something worth keeping, fast.
Trust
Users can tell when to rely on it, and recover when it gets things wrong.
Habit
It becomes the default for the job, not an occasional experiment.
Pay
The price makes sense next to an assistant they already pay for.
Four steps, each one deeper
Start with the free Leak Check and go as far as you need. The report's price is credited toward the audit.
-
Free
Next-Tab Leak Check
Enter your AI feature’s usage numbers and see where users drop off on the way to weekly use: Find, First win, Trust or Habit. You get your biggest leak, what fixing it could bring and the next step that fits.
-
$149
AI Feature Gap Report
You send screenshots or a recording of your AI flow and five real tasks with your feature's outputs. I score them blind against the three assistants, review the flow and give you ten ranked fixes.
-
$2,900
AI Feature Adoption Audit
A logged-in walkthrough of how users find, try, trust, repeat and pay for your AI feature, with the benchmark on ten real jobs. You get the one constraint holding adoption back, fixes as tickets, prompt and copy rewrites, and a readout.
Founding price for the first 5 clients, then $4,900. 5 of 5 spots left.
See the audit -
from $12,000
Adoption Fix Sprint
I ship the top fixes with your team: the first-run experience as working code or a prototype, rewritten prompts and copy, tracking for AI activation, two or three experiments, and a before-and-after benchmark.
Why me
Most people who audit AI features look at one layer: the screens, the model or the pricing. Adoption breaks between the layers, so I work across all of them.
- I build AI products. From a clinical AI coaching platform to AI-written email conversations. My fixes come as working code, not slides.
- 30 years of software development. I read your prompts, your stack and your event data, not just your screens.
- I have run my own funnel. I built my business on search traffic and conversion, so I know what a lost click costs.
- Behavior change is my training. BJ Fogg's Behavior Design, Cialdini's work on influence, CBT and motivational interviewing. Adopting a new tool is a behavior change.
NN/g UX Master Certification, CXL Certified Optimizer, Pragmatic Institute Product Master. Executive AI programs at MIT, Stanford and Oxford.
Questions
Do you need access to our product?
Not for the Leak Check, which works from numbers you enter. For the report you send screenshots or a screen recording and five real tasks. For the audit I need a test account.
Who is this for?
Software companies that have shipped an AI feature or an AI product and want more people to use it, keep using it and pay for it. It’s built for B2B SaaS teams and AI-native startups with paying users.
What if ChatGPT simply does the job better?
Then you'll know which jobs to stop competing on, and where your product has an edge a general assistant can't copy: your data, your place in the workflow, your defaults. That is often the most useful finding.
How is the scoring blind if you run the tests yourself?
A script shuffles the answers, labels them A to D and removes names and formatting that give away where they came from. I score each one against the quality bar you set for the task, and only then see which tool wrote which. You get every answer and the key, so you can check each score.
What happens to the material we send?
It is stored privately, only I see it, and it is deleted 90 days after you send it. I'm happy to sign your NDA before you send anything.
How do payment and invoices work?
The report and the audit are paid upfront by card through Stripe, and you get an invoice with your company details and VAT ID. Sprints and the Advisor are invoiced.
Who does the work?
I do, from the first email to the readout. No account managers and no juniors.
Would your users switch tabs?
Find out for free. Enter your AI feature’s numbers and see right away where users drop off.
Run the free Leak Check