Lots of people hacking together knowledge graph agents right now.
We've been building one from the ground up for months — custom agent framework, custom knowledge graph, running autonomously in production. It's live today.
https://t.co/sFcSxFEHSN
First research cycle is free, no card needed. 75% off for the first 100 signups.
https://t.co/sFcSxFEHSN
Curious what you'd point it at -- let me know in comments.
This blew our minds: Within three days it had built its own sell-discipline system from scratch — stops, trim triggers, rotation plans. None of it was in the prompt.
It caught structural bias in one of its own analyst tools — the valuation reviewer was flagging nearly every stock as "stretched." Rather than accepting the verdicts, it rewired its scoring system on its own.
It was profitable in both runs.
Then we ran Hanno (our financial world model) on the same 100 stocks, same window, same $10K starting capital.
⬇️
+20.8% return (vs +6.3%)
4.47 Sharpe (vs 1.37)
−3.5% max drawdown (vs −8.4%)
80% win rate across 20 closed trades
Everyone's building AI trading agents right now. We wanted to know how good they actually are — so we tested the one to beat.
We gave Anthropic's most capable LLM full autonomy over a 100-stock portfolio using their specialized financial analyst agents. Then we ran Hanno on the same stocks. 🧵
@trashhalo and the @Standd_io team built a financial world model, and set it up to short stocks while Wall Street said BUY.
The results: 70.6% win rate. 1.9x win/loss ratio. 54 signals. 1,801 companies. True out-of-sample, live trading, all of 2025.
We wrote up what we built, why it works, and why LLMs can't compete with it.
Read more: https://t.co/qGDfFInxjl
@jack and @roelofbotha published an essay this week with a contrarian answer to the most important question in tech right now: what is AI actually for. Their point of view resonates with how we're answering this question at @Standd_io.
Most companies are trying to figure out how to use AI to move faster through their existing workflows. But Block poses a more interesting question: what does a company look like when AI stops being a tool for efficiency gains and becomes the infrastructure through which a company operates? The essay is worth reading in full for their nuanced take on this.
What strikes me reading it is how much further this thinking can go. Block is applying this idea to how a company coordinates. We're applying it to how a company sees the world.
At Standd, we built from a simple observation: private market investors have traditionally been forced to develope expertise reactively, looking closely at something when circumstances demand it. Our autonomous agents do the inverse. They watch continuously, generating inferred signals about portfolio companies that don't exist in any database and that no one explicitly asked them to find. These signals inform proactive research and action. And they feed Standd PI, our world model, which builds causal understanding of markets, sectors, and positions over time. What the system develops is a documented point of view about markets, risk, and opportunites — one that compounds and deepens with every passing quarter.
The firms that define the next decade of private markets won't necessarily be the ones with the fastest analysts or the most deal flow. They'll be the ones that combine a winning strategy with a compounding data advantage — one that develops iteratively over time, until the gap between what they know and what everyone else knows becomes very hard to close. That's what becomes possible when AI stops being a tool people pick up and put down, and starts being the infrastructure through which a company sees the world.
OpenClaw agents die mid-task. No recovery, no visibility, no retry. We built the infrastructure layer that fixes that. We call it Exoclaw 🦀
https://t.co/hUgwpRRoRX
We are so excited to announce that we won this cool competitive process put on by First Row Partners' Ensemble Fund. We can't wait to get to work with this incredible group of investors!
The @FirstRowVC Ensemble Fund this fall selected @Standd_io as the winning investment last week, but the 'wins' were even broader for investors, groups, and founders. Thanks @geekwire for the article how this all went down.
Just got my first monthly payout from @BonsideInc.
Used it to fund my angel check into @Standd_io.
Did it all from my @carryhq_ Solo 401K.
This is the way.
WOW. What a moment for our team @Standd_io - It's been an honor to work with and learn from the amazing minds @AWS and we are so grateful to be a part of this occasion.
Shout out to the @togethercompute team for consistently delivering and being great people. Congrats on this exciting launch! You’ve got friends in #legaltech
RedPajama-INCITE-3B, an LLM for everyone: We are excited to share llama.cpp support! Efficiently run RedPajama on commodity CPUs!
This, plus new resources to fine-tune the model using LoRA, instruction-tuning tips and more on our blog: https://t.co/sv4qJQf448
It was such an honor to join @KateGoettel Jared Jaskot and @gsiskind to chat about AI and its implications for legal practice at the @federalbar Immigration Law Conference. What a fun and thought-provoking conversation!
2023 is gearing up to be a big year for DC-based legal tech startup @Standd_io, which recently won "Best Pitch" at @TechstarsSEA's demo day https://t.co/F7ppI3UkIL