Joined May 2021
1,066 Photos and videos
What if you had a system of agents that worked like a real equity research team? Not a model summarizing filings. A team that builds a thesis, then argues about it before anything ships. We built it, and the hard part wasn't the finance. 🧵
1
6
297
The fix came from how real teams work. Defend the thesis, update when the evidence is real, don't fold just because someone pushed. Now the analyst has to accept, reject, or partially accept each critique, with reasoning. The lesson nobody warned us about: LLMs are agreeable by default. Useful almost everywhere. Here it was the main failure mode. Getting research worth reading meant engineering disagreement into the system on purpose.
1
2
131
We expected the finance to be the hard part. It wasn't. The hard part was conviction. An agent that holds a view, updates on evidence, and doesn't go soft under pressure. Full breakdown 👇 you.com/resources/building-a…
2
131
Yesterday, our Head of Dev Rel @MarianeBekker spoke at @she_TO with engineering leaders from @mongoDB, @Salesforce, and @Disneyplus. When women are given a seat at the table, the whole industry builds better products. 💪 #womeninSTEM
1
5
618
tool_backend = os.getenv("BROWSER_BACKEND", "youcom") we're proud to partner with @OpenAI and support open-source agent tooling as the default browser backend.
3
6
19
2,934
Spotted at the midtown tunnel! 👀 Hallucinations? Elsewhere. Latency? Speedy. Integrate, ship, explore. @youdotcom does more. #newyorkforever
1
2
12
2,448
The best conversations at any conference happen off the floor. 🕹️ In one week, we're taking over an arcade at the @databricks Data AI Summit. Join us for a night designed for AI builders, hosted by the @youdotcom Head of Dev Rel, @marianebekker. Let's play some games and swap ideas. 🎮 RSVP in the comments👇
2
1
7
1,198
Join us at Tesla tomorrow at 10am for breakfast with founders and AI builders. 🍳 We'll have panels and discuss agents, infra, evals, production systems. Real talk from teams at @Workday, @Salesforce, @Snowflake, @Cisco, @IBM, and @SAP. Then, @Tesla demos and a light show. 🚗⚡ RSVP at the link in the comments 👇
1
2
10
827
Ran a macro research agent across all 27 EU members using the @youdotcom Finance Research API @LangChain Deep Agents. 5 subagents, each with one job. ~20 API calls. 45 min. $2.20. 🧾 Ireland showed a 12.3% GDP growth in 2025. It was pharma tariff front-loading, not a boom. The trace in LangSmith showed exactly how it got there. Read more in the comments 👇
3
8
1,240
AI agents > chatbots. Tomorrow at 11am PT/2pm ET we're showing how to combine Hermes Agent You.com Web Search APIs for: 🔷 Real-time web research 🔷 Competitor monitoring 🔷 Knowledge management 🔷 Workflow automation Join @dmitry (@buildwithRemy) and the @youdotcom team. Don't miss out! RSVP in the comments 👇
3
3
15
901
The tool layer is where agentic search quietly breaks. Not because the model is bad, but because the interface makes it do too much work. 🧵
23
51
117
32,946
The payloads do the heavy lifting instead. Goal-conditioned extraction, compact link encoding, inline citation markers. The model calls a familiar function and gets back exactly what it needs. Full inference budget goes to research, not to the scaffolding around it. Two principles, three tools, one cleaner way to build.
1
5
1,090