Joe Magyer (@Magyer) header
Joe Magyer (@Magyer) profile picture

Managing Partner at @SeaplaneVC. 2X founder. Host of the @InvestNStartups podcast.

Austin, Texas
Joined February 2010
Photo by Joe Magyer (@Magyer)Photo by Joe Magyer (@Magyer)Photo by Joe Magyer (@Magyer)Photo by Joe Magyer (@Magyer)Photo by Joe Magyer (@Magyer)Photo by Joe Magyer (@Magyer)Photo by Joe Magyer (@Magyer)Photo by Joe Magyer (@Magyer)Photo by Joe Magyer (@Magyer)Photo by Joe Magyer (@Magyer)
Joe Magyer retweeted
Introducing roots by micro1. The future of intelligence will learn from the real world. We’re making it easier for anyone to be a part of that process and earn from it. roots is an app that lets you contribute real-world data to help train AI systems, from photos and videos to screen recordings, wearable data, and more. Privacy and security are a priority as we build roots. This includes removing sensitive personal information from submissions, requiring consent when recording others, and securely storing data. Explore available projects, capture or upload your contributions directly through the app, and get paid for approved work. New opportunities will be added regularly. Join the waitlist via link in comments.
63
64
36
570
81,885
Joe Magyer retweeted
Honestly, shame on the CIA for not getting Jensen a patriotic state issued baddie.
173
1,466
110
38,237
876,479
If you don't like datacenters then get off the internet
JUST IN: 56% of voters say AI & data centers were a major or crucial factor in their midterm vote for Congress.
160
219
16
3,418
84,783
Joe Magyer retweeted
If America tries to attack its leading technology companies, the country will suffer. Technology is our greatest asset for increasing the pie. This must be bipartisan.
4
1
8
1,706
Joe Magyer retweeted
Today, Trata is excited to share the revised Hedge-Bench: our benchmark tying open-ending reasoning to prediction capabilities. We constructed 92 real-world environments derived from hedge fund analysts’ reasoning traces. Agents reason from frozen information snapshots, and their forecasts are evaluated against future real-world outcomes. No model scores above 50% (GPT-6 Astra being the top performer at 44.2%). For frontier, state-of-the-art models, these results highlight a persistent and wide gap between producing plausible-sounding reasoning and making accurate, economically valuable forecasts. In a semi-verifiable domain like finance, where outcomes are delayed and reasoning is difficult to verify, intermediate expert reasoning traces are crucial. We post-trained @Meta's Muse-Glimmer-30B on 165 environments with per-turn credit assignment using an LLM-as-a-Judge. We observe evidence of generalization: held-out forecasting accuracy at high effort improved by 10.06 percentage points, while out-of-distribution evaluation scores improved by 3.94 points on @OpenAI's GDPval, 2.29 on @Surge’s GDP.pdf, 7.07 on @Harvey’s LAB, and 4.96 on @Samaya_AI’s FrontierFinance under their respective metrics.
13
12
3
92
11,999
Joe Magyer retweeted
Just spent the last 7 days on the front lines of the vulnpocalypse fending off an active cyber attack on one of our products. Some thoughts: - Software factories are going to be required equipment for everyone. There’s just no way to keep up with the pace of highly automated adversaries without one. - We were lucky because we had spent the last few months getting our software factory running well. It is nice to have for product development but proved to be absolutely essential this week. - At the minimum you need to have a coding server in the cloud pre-provisioned with read only access to all your logs and all the services you depend on for observability. With that in place you are a prompt away from reasonably good situational awareness, ongoing monitoring with loops, and ability for coding agents to validate fixes against your actual, observed attack patterns. - Shared coding server also unlocks multiplayer incident response where everyone can see every session and jump in as needed. Sharing session links that can be forked is much higher bandwidth than re-explaining things. - You don’t need to be Ramp/Stripe/Uber scale to have a working software factory in the cloud. We are a team of 3. Ours is built with an @opencode server (thanks @thdxr!), a forked version of compound engineering optimized for automated workflows (thanks @kieranklaassen!) and a multiplayer agent in slack that handles crons (we have our own harness but pick your favorite here). - To get our software factory up and running well, we had to stop all new product development for ~6 weeks and just focus on getting the validation systems working that enabled us to reliably shipping PR without human code review. This felt very painful in the moment but was totally worth the eventual speedup in velocity. - KYC for @AnthropicAI & @OpenAI cyber programs is now easy and quick (at least for defensive access), but a day felt like a lifetime this week. Submit your app now before you really need it. - Getting Mythos access unlocked a deeper level of analysis that helped us shift from reactive mitigation to proactive patching. Wish we had done this a few weeks earlier. Get started now!
72
155
59
1,876
263,034
I had a great time talking startups and tech with my friend @Magyer on his podcast. Check it out!
The best deals aren't competitive. That's the view of my friend and early-stage investor, @alexrubalcava. We talked on @InvestNStartups about VC groupthink, the evolution of moats, embracing technical risk, only investing in greenfield markets, and more. Technology risk creates opportunity. Many VCs avoid technically difficult startups, leaving room for investors able to assess and underwrite those risks. As AI makes software easier to build, Alex sees more potential in frontier tech and physical-world applications. AI is redefining software moats. Faster development weakens traditional defensibility. Alex favors software in regulated or high-stakes settings where accuracy, auditability, accountability, and human oversight still matter. Sales and product judgment matter more. Alex evaluates founders on whether they can ship, sell, and hire. With building easier, sales velocity and deciding what to build—and when—have become more important. Venture suffers from groupthink. His public-markets background shapes a contrarian approach grounded in valuation discipline, independent judgment, and finding opportunities overlooked by consensus-driven investors. Small funds can deliver big returns. Exceptional LP outcomes don't require trillion-dollar exits or mega-funds. An overlooked multibillion-dollar exit can be transformative for a smaller venture fund. Capital efficiency reduces funding risk. Long runways give startups time to prove their case, especially when follow-on investors favor AI narratives. Alex cites profitable, fast-growing non-AI companies struggling to raise despite strong fundamentals. Greenfield markets offer an edge. Alex prefers startups serving needs no existing technology addresses over crowded VC-backed categories. Shipping products, winning early customers, and showing demand turn unknown markets into investable opportunities. Investing expertise must evolve. Alex highlights secondary share sales as a path to LP liquidity, Amplify's founder network as a diligence resource, and the importance of steadily expanding ones circle of competence. Persistence needs experimentation. Productive founders test new products and sales tactics rather than repeating failed approaches. Alex recalls a founder who shut down a struggling startup and returned half its capital instead of grinding on. Conviction beats allocation. Alex argues the best opportunities may have just one term sheet. At Stage VP, 42 of his 45 investments were noncompetitive, showing how leading overlooked rounds can create an edge. Alex is very talented and a great investor. I hope you enjoy this one! investinginstartups.com/post…
1
7
659
Joe Magyer retweeted
I would rather go back to the BCS than a 24-team playoff.
202
2,834
144
23,592
325,169
Joe Magyer retweeted
To our knowledge, the most impactful bug ever found in crypto. Good work @cayden_liao!
Our AI found a bug that let anyone mint infinite XRP, with $94B at risk. It had been hiding in one of the most audited codebases in crypto for nearly a decade, and earned us the maximum $250K bounty. Here's how it went unnoticed, and how we found it.
7
5
2
95
7,167
Our AI found a bug that let anyone mint infinite XRP, with $94B at risk. It had been hiding in one of the most audited codebases in crypto for nearly a decade, and earned us the maximum $250K bounty. Here's how it went unnoticed, and how we found it.
282
356
195
2,631
477,770
Joe Magyer retweeted
So, let's say combined lab ARR is actually $126B vs $146B... The annualized growth rate for 2026 is 577% instead of 724%. I get the danger of relying on non-GAAP metrics, but IF THIS IS ACCURATE, it is not "popping anything."
73
63
14
673
76,824
Joe Magyer retweeted
thanks for having me Julian & for this U.S. first AI question. 🇺🇲
"There are about 150 companies signing up every single day." Ali shut down a $7M run-rate startup to build what's now a $4B company. Solo Founders Podcast ep 31 is live with @aliansarinik of @micro1_ai. 04:10 The first thing he ever sold online 17:30 Three signals it's time to move on 24:40 The all-in Loom that deleted the revenue 31:45 "Only the founding team can inject risk" 35:50 He never knew co-founders existed 43:40 Threads, final actions, zero meetings 53:30 Inside the AI data wars 1:01:50 Why micro1 won't sell to America's adversaries 1:07:30 The case for and against going solo
12
10
4
88
8,224
Joe Magyer retweeted
licensing data will become one of the most common revenue line items for companies worldwide, similar to how ads became a default monetization model. for many companies, it will become core to the business.
22
17
8
331
26,208
Founders, I am investing even earlier now! Got an idea you can’t leave alone? Let’s get it off the ground and get you from 0 to 1. I’m excited to launch Cold Start from @yonderventures: a $10k starter check to help launch your marketplace or network-effect business. Work directly with me, tap into the Yonder mentor network, and get $15k+ in partner perks. Solo or with cofounders. An idea is enough to apply. Tell me why this is the thing you want to build above all else. Rolling applications. A response within two weeks. Let’s build!
🤖 Made with AI
56
26
4
299
20,727
Joe Magyer retweeted
to win the AI race, we shouldn’t export American Intelligence 🇺🇸🇺🇸🇺🇸
17
15
8
173
180,418
Joe Magyer retweeted
Today we’re launching micro1’s PII transformation model, flow-transform 1.0, delivering frontier-level performance across detection, identity synthesis, and transformation of personally identifiable information. On PrivacyBench, our model reaches 96.0% F1, outperforming every detection baseline we tested, including Tonic Textual, Claude Opus 4.8, Sonnet 4.6, Microsoft Presidio, Haiku 4.5 and GLiNER2. Some of the most valuable training data for frontier AI models lives inside fully functioning companies. It captures years of real work across decisions, communications, tools, handoffs, exceptions and the relationships connecting them. The problem is that this data is also full of PII. Traditional redaction makes the data safe, but it also destroys the very workflows and relationships frontier models need to learn from. flow-transform 1.0 solves this by turning enterprise operational data into high-fidelity training data for frontier models by replacing real-world identities without flattening the reality the data captures.
91
109
33
476
287,906
Joe Magyer retweeted
We’re sharing a deeper look at flow, our next-generation data platform for delivering predictable units of intelligence improvement at scale. This article transparently lays out the full flow architecture including its data generation models, expert workflows, and models for quality control and evaluation. At its core is a flywheel where human expertise continuously compounds. Experts set the standard for models to generate and evaluate training data, then their corrections feed back into both the data and the models producing it. Each round strengthens the next, making intelligence gains more predictable while driving costs substantially lower. Explore the full framework below. micro1.ai/blog/introducing-f…
14
15
6
88
23,249
Joe Magyer retweeted
the hardware embodiment of frontier models like Claude and GPT is the most urgent AI safety problem in front of us today. we simulated two very simple use cases using claude both in simulation and using robot arms. in one, claude spilled toxic liquids in a lab. in another, the force it used to place an animal toy into a basket was strong enough that it could have physically harmed sensitive material—or anything else in its path. these are simple experiments, using models out of the box today. as researchers increasingly give frontier models arms, legs, and access to the physical world, we need to urgently build and assess guardrails around what these systems can and cannot do. models escaping sandboxes or compromising enterprise security infrastructure are serious concerns. however, hardware embodiments introduce something fundamentally different: an AI system can make a mistake in the physical world, and the consequences may not be reversible. this is not a future safety problem. the capabilities exist today. we’ve released a report detailing this & solutions we propose. link in comments below.
53
72
16
246
185,113