Raphael Mosaic

Enterprises which win the AGI race will build their own internal Mercor.

Until now there was no way to build internal human datalabs without surveilling your workforce.

Introducing @alexandriainc

hosting Dr. Robert Fink CTO of @HelsingAI and co-inventor of @PalantirTech foundry
and @cory founder of @zfellows
for fireside chats at Alexandria Munich this week

20 participants each, apply below

the most complete manifesto of where we are standing with BCI

Bloomberg Terminal desktop sessions could be used as trajectories to train long-running quant trading agents.

This Saturday, Alexandria and Maximilian Eichler are hosting a research circle on this topic at @EarlybirdVC London office.

Most conversations about computer-use agents focus on execution: can a model navigate a terminal and click the right buttons?
We think the more interesting opportunity is the dataset.

Senior quant traders leave behind detailed traces of how they research: what they query, where they shift attention, which hypotheses they test and how they reach decisions.
This research circle focuses on the question of how we can create long-running agentic clones of the best quant traders based on their long-horizon Bloomberg Terminal usage.

There will be no presentations and no fixed program. This is not a hackathon. It’s three hours of collaborative discussion and brainstorming around questions like these:
- How should RL environments for Bloomberg Terminal research be designed? - What should constitute the observation space, action space, episode, reset mechanism and research horizon?
- How should trajectories represent quant semantics, including instrument identity, research universe, timestamp, data vintage, hypothesis, constraints and decision rationale?
- Which Bloomberg Terminal workflows are most valuable to capture, including idea generation, security screening, data retrieval, event analysis, model iteration, portfolio construction, risk analysis...?

Bring a laptop if you would like to look up papers, inspect tools or test ideas during the discussion.
We’ll meet in a boardroom with smaller spaces for focused conversations.

The event is limited to 20 participants. Apply via the link below. A background in machine learning or quant/commodity trading is preferred.

Ad Astra

sf needs more galas which embrace long horizon conversations:
@miramurati seems like having out of distribution high integrity and conviction on AGI with humans in the center.
@mkratsios47 gave sharp insights on how the administration is backing individual scientists instead of scientific orgs (similar to talent investing in vc), so that labs need to compete for the scientists with the money instead the other way around. Germany needs to level up and get a similarly sharp and ambitious CTO/CSO for the country, @DoroBaer is definitely not it…

great job: @JoinFAI @poastmilque