01What we are building
Most AI products stop at giving advice. Emily has to finish the work, and get it right. That means operating reliably in browsers and desktop apps, staying on track through long tasks, recovering on her own when something breaks, and always respecting the boundaries of permission and security.
02What you will do
You will own the core, and hardest, part of Emily, from the first prototype to how it performs in production.
- Design and refine Emily's agent loop: planning, tool use, and acting in real web pages and desktop apps.
- Build the evaluation system that turns "did she get it right?" into numbers we look at every day.
- Solve the hard problems: reliability on long tasks, recovery from errors, memory and context.
- Turn model capability into product experience people can feel, and own the results after launch.
03What we value
These draw on teams we admire, such as Replit, and they are the standards we hold ourselves to every day.
- Mission first: ask whether it makes Emily more reliable before asking whether it is easy.
- Think boldly, ship in small steps: question convention, but get every step live and learn from it fast.
- Extreme ownership: from idea to launch to production issues, it is yours, and the bar is high.
- Seek the hard truth: go looking for the most painful problems and the bluntest feedback early.
- Product sense: care about what people actually experience, not just metrics and demos.
- Self-directed and clear: you move without being pushed and can explain complex things simply.
04How we decide
We care more about what you have built than about the names of your school or last employer. When we talk, bring something you made and are happy to go deep on: a project, some code, a hard problem you solved. Technical interviews mean building together on a real problem, not reciting answers.
05Worth knowing first
If any of these give you pause, this may not be the place for you, and that is fine.
- We move fast and plans change often as we learn.
- We would rather try and fail than finish something mediocre.
- Building is always harder than critiquing what others built. We need builders.
- We do not glorify overtime. What matters is what got done, not hours at a desk.
Want to talk?
Send an email: who you are, what you have built, and the one piece of work you most want us to see.
Innovate Link Inc. · Palo Alto, California, United States