Show HN: Acceptify – AI personas that run user acceptance tests on your product(acceptify.ai)
acceptify.ai
Show HN: Acceptify – AI personas that run user acceptance tests on your product
https://acceptify.ai/
Hey Show HN! I built Acceptify to solve a problem I ran into while building my other app.
I’m the founder of a SaaS tool called Fortress. Early on, I had a classic issue: not enough real users yet, but I still needed feedback on whether features actually worked from a user perspective. QA and unit tests told me if the code worked — not if the experience made sense.
So I built an internal tool that creates AI personas and agents to simulate different kinds of users interacting with the product.
How it works:
1. You define a user story and acceptance criteria (similar to how product teams already write them)
2. You create AI personas (e.g. technical power user, first-time non-technical user, impatient user, etc.)
3. You give an AI agent a goal like “complete signup” or “create a project”
4. The agent explores the product autonomously, navigating, clicking, and making decisions
5. Acceptify evaluates what passed, what’s uncertain, and where behavior didn’t match expectations
It’s essentially an attempt at automated user acceptance testing that focuses on behavior and expectations, not just DOM checks or scripted flows.
Some things it surfaces well:
• Flows that technically work but are confusing
• Missing validation or feedback states
• Places where navigation or UX assumptions break
• Gaps between acceptance criteria and actual behavior
This started as an internal tool for Fortress, but I realized other product teams probably face the same “too few users, too much guesswork” problem, so I turned it into a standalone product.
It’s still early, and I’m very interested in feedback from HN folks especially around:
• How this compares to existing QA/UAT workflows you use
• Where you’d expect this to fail or hallucinate
• What would make this genuinely useful vs. just a demo
Happy to answer questions and share more technical details about how the agents and evaluation work.
3 comments
Interesting approach. The core problem you're solving - "not enough real users yet but need behavioral feedback" - is very real for early-stage products.
One gap I'd watch for: AI personas will tend to be too rational. Real first-time users do things no persona model would predict. They misread buttons, get distracted halfway through a flow, form completely wrong mental models of what your product does. The "impatient user" persona is a good start, but actual impatient humans are more creatively confused than any simulation.
We're working on the other side of this at Test by Human (testbyhuman.com) - actual people screen-recording their first time through your product with voice narration. The combination could be strong: use Acceptify for fast iteration loops during development, then validate with a real human before shipping.
Curious how the agent decides when to "give up" on a confused flow vs. keep trying. That threshold seems like it would matter a lot for the quality of the feedback.
One gap I'd watch for: AI personas will tend to be too rational. Real first-time users do things no persona model would predict. They misread buttons, get distracted halfway through a flow, form completely wrong mental models of what your product does. The "impatient user" persona is a good start, but actual impatient humans are more creatively confused than any simulation.
We're working on the other side of this at Test by Human (testbyhuman.com) - actual people screen-recording their first time through your product with voice narration. The combination could be strong: use Acceptify for fast iteration loops during development, then validate with a real human before shipping.
Curious how the agent decides when to "give up" on a confused flow vs. keep trying. That threshold seems like it would matter a lot for the quality of the feedback.
Great question, and your combination of AI and human validation is definitely something that could be strong.
As far as the 'give up' signal, the Agent has several triggers that could stop it from executing, as well as a max number of actions it can take before giving up. Examples of the triggers are page has not loaded after 30s, button clicks or inputs are not responding, etc. If the agent is given the task of 'complete the onboarding process', it will evaluate after each action whether or not it has sufficient information or needs to continue. Once it determines that it has completed the task, it summarizes the actions taken into the final report, including screenshots where applicable.
Are there additional triggers that you think would be helpful? Should the 'max steps' be user-customizable (i.e. maybe you allow it to take up to 50 actions before failing vs. 20 or something like that)
Thank you for your input! It's much appreciated!
As far as the 'give up' signal, the Agent has several triggers that could stop it from executing, as well as a max number of actions it can take before giving up. Examples of the triggers are page has not loaded after 30s, button clicks or inputs are not responding, etc. If the agent is given the task of 'complete the onboarding process', it will evaluate after each action whether or not it has sufficient information or needs to continue. Once it determines that it has completed the task, it summarizes the actions taken into the final report, including screenshots where applicable.
Are there additional triggers that you think would be helpful? Should the 'max steps' be user-customizable (i.e. maybe you allow it to take up to 50 actions before failing vs. 20 or something like that)
Thank you for your input! It's much appreciated!