We presented this project at innovationintelligencehackathon.com and today we are putting a url out for you to test and give us feedback. after talking to teams building agents that research companies to create outreach, we found few things that are very important to them. one of them is recency. When we prompted agents with outdated alert about companies, 4 out of 7 times, they came up with outdated insights, even when they had recent news in context. A bigger model didnt fix it. So we built search for ai-agents with usefulness score on top. So, when you search each page comes with score+price.

this solve 3 problems;

1. token, SOTA agents read all pages to make sense of the useful ones

2. cost, when agents pay for pages they read, its is cost-effective to read only useful ones

3. quality, our score consider the task/query at hand and score useful pages, increasing quality of answers.

we want to find out fast if we are really up to something here. we'd love to hear from you.