Docket (YC P25)

Centerpoint Connect runs full regression across 5 modules in under 30 minutes with Docket (YC P25). Centerpoint Connect is the all-in-one platform for commercial roofing contractors: CRM, project management, dispatch, estimating, client portal across Web, iOS, and Android. Their senior QA engineer had developed a Selenium + .NET test suite. But the team encountered familiar challenges as the product evolved: UI changes meant regular upkeep, CI integration remained on the roadmap, and new test authoring required a meaningful time investment. So, they brought in Docket (YC P25). Tests are now written in plain English. Docket runs them visually using screen coordinates and self-heals when the UI shifts. → Pre-release regression: down to under 30 minutes → Full platform coverage in weeks → QA freed up for exploratory work "Docket gave our team the confidence to ship faster without sacrificing quality." - William Riley, CEO, Centerpoint Connect Full case study in the comments 👇

eXp Realty validates Jira acceptance criteria across a $4.8B platform with Docket (YC P25) and 100+ team members are now using it. eXp Realty runs a complex tech stack serving tens of thousands of agents - dashboards, commission tracking, CRM integrations, AI tools. Every Jira ticket comes with acceptance criteria a QA engineer has to manually walk through. Hours per sprint, per engineer, on checks that could be described in plain English. Sherri Delbridge and her Software Assurance team brought in Docket. Now engineers write acceptance criteria as a Docket test in plain English. Docket runs the flow visually using screen coordinates (not DOM selectors), validates each criterion, and self-heals when the UI shifts. The compounding effect: one-off acceptance criteria checks accumulate into a permanent regression suite as a byproduct of normal sprint work. No separate automation project. → 100+ eXp team members on Docket today → Spreading across teams, including Zoocasa wiring Docket into CI/CD → Increasing QA velocity without increasing headcount "Docket basically acts as an extra QA engineer on our team." - Sherri Delbridge, Sr. Director, Software Assurance Full case study in the comments 👇

Building a car online should feel as smooth as driving one. With Docket (YC X25), we tested the Mercedes-Benz AG customizer by building a GLE 450 SUV end-to-end — all from a single natural language prompt. No scripts. No setup. Just real user behavior, automated. Because even luxury flows deserve reliable QA.

This one’s a founder favorite. With Docket (YC X25), we sketched a toy car on Stability AI's dreamstudio canvas, and then prompted it to turn the sketch into a Ferrari. End-to-end, we tested not just image generation but creative iteration and asserted the output succeeded. The coolest part? Docket itself drew the toy car. No scripts. No setup. Just natural language, real user behavior, and a touch of creativity.

We asked Docket (YC X25) to generate a video on Pika. Instead of the flow completing, we hit a paywall. That’s the reality of end-to-end testing: catching the unexpected before your users do.

Runway turns ideas into visuals — but only if every click, render, and export works flawlessly. With Docket (YC X25), we told Runway to generate a unique image and tested the entire flow end-to-end, using just natural language. No scripts. No setup. Just real user behavior, automated — from concept to final output. Because in creative AI, the smallest break can stall the biggest ideas.

Playground makes creating stunning visuals effortless. With Docket (YC X25), we told it to design a sticker and tested the entire flow end-to-end, using just natural language. No scripts. No setup. Just real user behavior, automated — from prompt to finished design. Because creative tools deserve creative QA.

Testing Gen AI isn’t just about buttons. We need to test subjective outputs. With Docket (YC X25), we acted as a synthetic user on Leonardo.Ai, giving only high-level instructions to test its model image generation flow end-to-end. No scripts. No setup. Just natural language — and a full check that the creative engine works as expected.