Theory Ventures

Legacy ERPs enforce a consolidated stack, but different functions have very different needs. Finance is all about conformity to accounting standards, while ops are all about configurability for workflows. To give its customers the best of both worlds, @doss_hq has partnered with leading finance/accounting platforms - both incumbent ( @Intuit ) and startups ( @RilletHQ& @Meet_campfire ):

Attackers only have to be right once. Defenders have to be right every single time. Our Head of AI @BEBischof sits down with @DropzoneAI’s Edward Wu to unpack the implications of this asymmetry in cyber defense: youtu.be/fDZ8vC1kzEw?si…

Catch our Head of AI @BEBischof advocating for a new optical interface for evaluation and prompt optimization: eyeballs.

Enterprises now run on AI agents. So do hackers. Join @ttunguz for Office Hours with Jonathan Jaffe, CISO at @Lemonade_Inc, who will share what it means to build, secure, and operate AI systems when both sides are automated. 📅 Thursday, May 21st 📷 8:00 AM PDT / 11:00 AM EDT gatsby.events/theory-venture…

@HessianFree @motherduck @lancedb @AICouncilConf Ya that duck lookin' fly

@Theoryvc @motherduck @lancedb @AICouncilConf ive never seen someone with so much drip!

From Theory, @motherduck, and @lancedb: we salute you @AICouncilConf, and the conference attendees. If you missed our boat party, we hope you’ll join us in the future. 🫡

In 2024, teams expected 500% efficiency gains from AI but recorded 0%. The 2026 Theory GTM Survey is live with five new hypotheses, from AI-augmented reps vs. full replacement to whether buyer-side AI is the real disruptor. Take the survey to receive our data: tomtunguz.com/2026-gtm-surve…

Did your favorite dev tool get declared dead last week? Our Head of AI @BEBischof joined @DevInterrupted to cover what makes data agents work, why the industry keeps eulogizing perfectly good tools, how Theory builds operational VC software, and the origin story behind rip-grep Listen: open.spotify.com/episode/5BGDDu…

I went on Dev Interrupted to talk to Andrew about why this year, you should be in the Inference Systems track at @AICouncilConf This is your chance to actually learn what's going on on the other side of those API requests. This is where you can learn about what you're going to be expected to understand in 2027 as an AI practitioner. I love the application layer, but the future is inference systems. Catch me tomorrow at AI Council.

@yacinelearning @AICouncilConf @BEBischof @yaroslavvb @togethercompute Ya homie, I gotchu. Email Bryan

@AICouncilConf @BEBischof @Theoryvc @yaroslavvb @togethercompute where can I get a ticket man I'm in town

The Inference Systems track at AI Council 2026 is where the people rebuilding the stack share what they're doing. Curated by @BEBischof, Head of AI at @Theoryvc. Here's the lineup: →@yaroslavvb, Principal Researcher at @togethercompute: "What Comes After Deep Learning?" → Neil Movva, Co-Founder at Sail Research: "Great Infra for Background Agents" →@vikhyatk, CTO at M87 Labs (@moondreamai): "No Dropped Frames: Designing a VLM Around a Latency Budget" →@johnpdickerson, CEO at @MozillaAI: "Do the Boring Stuff to Make Open Source AI Win" →@CompleteSkeptic, Co-founder & CEO at @typesafeai (co-inventor of ChatGPT): "AI: Too Good to Be True, Too Bad to Be Useful" → Sriram Vishwanath, Professor and Founder at @GeorgiaTech: "Beyond Next-Token Prediction: Joint Embeddings, World Models, and Why Natural Language Isn't Enough" → Omead Pooladzandi, Co-Founder & Co-Head of Research at @PrismML: [talk title forthcoming] Thanks for curating a great track, Bryan! See you at AI Council 2026 in SF tomorrow! 🎟️aicouncil.com

What does a CISO who once took on Facebook think about AI security? Before becoming CISO at @Lemonade_Inc, Jonathan Jaffe was the named plaintiff in a major privacy class action against Facebook -a case that ended its Sponsored Stories ad product. The throughline: paying close attention to where technology gets ahead of its guardrails. Now he's focused on a new version of that question: What does it mean to build, secure, and operate AI systems when both sides are automated? Join @ttunguz for Office Hours: 📅 Thursday, May 21st ⏰ 8:30 AM PDT / 11:30 AM EDT gatsby.events/theory-venture…

Ever notice that so-called better models are often… worse? When Anthropic announced Opus 4.7, they said it was the best coding model they’ve ever released. Yet developers found 4.7 to perform far, far worse than 4.6 at coding workflows. What’s happening here? Partner Andy Triedman unpacks this phenomenon and what labs + agent companies can do in the face of these “degrading changes.” Read: theoryvc.com/blog-posts/deg…

"Why are we still creating feature pipelines by hand?" Chang She @changhiskhan, CEO and Co-founder of @lancedb and original co-author of the pandas project, sees a world where software engineers have stopped writing most of their code manually but data teams are still debugging distributed pipeline failures at 2 AM and watching their OpenAI bills spike over the weekend. In his PyAI Conf 2026 talk, Chang breaks down the ten biggest pain points of feature engineering for AI and makes the case that our data infrastructure was built for the last generation. Full talk here: youtube.com/watch?v=dv4vZQ…

Our AI engineer @adam__conway is a man of Taste. Your coding agent... not so much. Adam attempts the impossible: splitting the atom to explain taste in software,  and help agents eff the ineffable. theoryvc.com/blog-posts/tas…

Can ducks swim in muck? Our Head of AI @BEBischof probed @motherduck CEO @jrdntgn on why unclean data is still a problem, whether tables still matter, and how stored context keeps you from dredging your data lake every time:

If AI coding agents made one thing clear, it's that every business should expect custom-made, self-updating software. But no enterprise will want to vibe-code their core system of record. And legacy, decades-old ERPs are not designed to be modified by coding agents. @doss_hq is closing the gap for AI-generated enterprise apps, giving agents well-defined building blocks to construct and modify system-critical workflows. Our Partner Andy Triedman sat down with CEO @wileycwj to unpack what that looks like in practice:

@ttunguz spoke with @AnitaRamaswamy at @theinformation on moats in AI cybersecurity: “Imagine you have a 100,000-person organization. All those people have to come into IT, and it has to go and delete the CrowdStrike agent and then put on the Anthropic agent. What a pain. I’m not going to go through and spend six months of my security team’s time replacing this bit of software, especially if the gain is nothing. So that’s why I think there’s some defensibility. It’s just inertia.” Read more: theinformation.com/articles/crowd…

Join us for a free-flowing chat on data and AI as we sail along the Bay. 🚢 During AI Council week in San Francisco, we're hosting a private gathering with @motherduck and @lancedb. We’ll be convening a high-signal ship of builders, operators, and technical leaders shaping the future of AI systems and data infrastructure. 📅 Wednesday, May 13 at 6:15pm PDT Some conversations are just better on the water. Apply to attend: luma.com/dxsyixj5

"What's changing in enterprise AI adoption, and where is real value being created? Partner Andy Triedman will join the VC stage at Startup Grind to discuss how AI is reshaping enterprise software and what early stage builders must do to differentiate in this space. Join him next week (April 28 4-5pm) at Fox Theatre in Redwood City: startupgrind.tech/agenda?date=20…"

No watered-down cocktails here... just a great evening floating the Bay and talking data and AI. We're hosting a private event with @motherduck and @lancedb during AI Council week in San Francisco. Small group. No traditional event format. Just builders, operators, and technical leaders in a free-flowing environment designed for real conversation as we sail the Bay. 📅 Wednesday, May 13 ⏰ 6:15–9:00pm PDT 📍San Francisco Sailor outfits not required. Apply to attend: luma.com/dxsyixj5