Startup Investor | Strategic Advisor | Cofounder of GovPredict (YC S14, Acquired) | Former Google, Quora, Sweetgreen & Capitol Hill
Austin, TX
Joined December 2008
- Tweets15.6K
- Following1.7K
- Followers2.8K
- Likes1.7K
Steve Johnston retweeted
Still chasing supplier updates through emails and manual follow-ups? There’s a better way.
In this episode, @ScottwLuton and Rick McDonald welcome Nick Gospodinov of Mandel AI to discuss how AI agents can automate supplier communication, catch disruptions, and reduce manual work while keeping people in control.
Listen now to see how agentic AI can help your supply chain work smarter: hubs.ly/Q04DLSw20
Steve Johnston retweeted
BREAKING: @PerseusDefense achieves direct hit of a moving target 3 weeks after first kinetic direct hit-to-kill 🤫
Steve Johnston retweeted
the full down range shot. bullseye 🎯
@PerseusDefense
My fiancée, watching @SteveKornacki’s Election Night coverage with me:
“It’s like a sport… a sport for nerds.”
No lies detected.
5
473
Steve Johnston retweeted
the second ever attritable missile to shoot a drone out of the sky, this time very far down range
Kinetic direct hit-to-kill for the win
@PerseusDefense
@ycombinator
Steve Johnston retweeted
the first ever attritable missile to shoot a drone out of the sky @PerseusDefense
Not too late to apply to the next batch of @ycombinator.
You never know: it could change your life.
One week left to apply to YC Winter '27!
YC founders get:
🔸 $500K in funding
🔸 $12M+ in tokens, credits & deals
🔸Partners who were founders first, and are in your corner for life
🔸 Access to the best founder community in the world
🔸 Demo Day in front of thousands of top investors
Apply by Nov 2: ycombinator.com/apply
Turns out @ycombinator is the YC for DC.
That's one of my biggest takeaways from this great conversation with @Airbnb Cofounder @jgebbia.
YC was transformational for Joe and Airbnb. Now he's bringing that same mindset (e.g. get close to users, question conventional wisdom, move quickly, build things people love) to Washington.
The opportunity is bigger than better government websites. It's about redesigning the user experience of being a citizen.
YC has changed technology, business, and culture. It's fascinating to see that impact extend into government, too.
Don’t discount your destination.
I recently talked with a founder who’s getting ready to fundraise. As we discussed what each of their products could become, it became clear he was underselling the size of the opportunity.
My biggest feedback: go bigger.
I’ve been thinking about this a lot lately. We tend to want to sound reasonable, set expectations we know we can meet, and not get too far out over our skis.
But sometimes that means selling ourselves short.
Counterintuitively, I think people are more likely to get behind you when you’re aiming for something really, really big.
Take up space. Be big and be bold.
Awesome that @SBADeputy_Bill stopped by @PerseusDefense. Texas is helping rebuild America’s Arsenal of Freedom! 🇺🇸
The Texas defense industrial base is growing and companies like @PerseusDefense are leading the way.
Perseus is a drone defense company less than two years old. They saw a gap and just went and built cool tech that is part of the emerging Arsenal of Freedom.
Great choice by @PhRMA to have Eric Cantor lead as the organization’s next president and CEO!
We are excited to announce Eric Cantor as our next president and CEO.
Eric is well respected among policy leaders in Washington and around the world, with a long track record of working effectively across the political spectrum in support of biomedical research and patients’ access to medicines. He joins PhRMA at a pivotal moment for America’s biopharmaceutical industry.
Read more: phrma.org/resources/phrma-se…
As advertisers look for incremental reach in the final weeks of a busy fall advertising season, one interesting channel worth knowing about is Taiv, a @ycombinator-backed startup I’ve gotten to know better this year and started advising.
Taiv helps advertisers extend TV campaigns beyond the living room, reaching audiences watching live TV (especially live sports) in bars and restaurants across the country.
If you’re looking for ways to add reach to a campaign this fall, feel free to DM me. Happy to connect you with the team.
1
284
Great discussion in Austin this week with @skupor and @JTLonsdale on bringing top talent and an entrepreneurial mindset to government. I especially appreciated their emphasis on incentives and encouraging more calculated risk-taking in the public sector. More of this!
6
249
Great to be back at @ycombinator Demo Day today. Some remarkable stats from @garrytan:
28K applications, <0.9% accepted
Nearly 200 companies
1 in 4 hard tech, 3x last batch
20 manufacturing
13 industrial
9 defense
1 in 3 founders has a PhD or taught in higher ed
162 papers, 5,500 citations
YC is where ambitious founders build. And today, that increasingly means hardware.
Software ate the world, and hardware will rebuild it.
3
218
Heading back to YC Demo Day tomorrow. Who are the investors I should make sure to meet while I’m there?
1
1
197
Steve Johnston retweeted
Who is Austin has a lux-looking bathroom we can use for a photo shoot?
Will trade you a Throne!
Steve Johnston retweeted
We built QCircuitEval: a benchmark for quantum code.
Here is the problem it was built for. Two quantum circuits can return the same measurement results and still be different programs. One correct, one wrong.
Many benchmarks for quantum code grade against a single reference solution, so a model can produce the right numbers without doing the work.
QCircuitEval checks both. What the circuit outputs, and how it was built. QCircuitEval says: I know you got the answer right, but show me the working.
70 tasks across Qiskit, Cirq, PennyLane, and NVIDIA CUDA-Q. Open-sourced, MIT license. GPT-5.6 Sol hits 92.7% pass@1. Seven of the 33 models and configurations we tested sit above 90%. Frontier models are already saturating the benchmark.
Two things I did not expect. Reasoning effort makes little difference to the frontier scores. And safety filters push some models to flat out refuse tasks they could otherwise complete.
Quantum computers will not produce new science on their own. Something has to write the programs that run on them. More and more often that something is a large language model. If those programs are wrong in ways neither we nor the model can see, the results are wrong too. Nobody finds out until much later.
We want researchers across academia and industry to set this standard with us. Quantum code generation depends on it. So does the first AI that uses a quantum computer to discover something new.
