First Commit

Pair programming with AI gets you 20 to 30% faster. Robert Brennan builds for the workflow that gets you 10x.

In this episode of First Commit, Eliya Elon and Laura Hamilton sit down with Robert Brennan, CEO and co-founder of OpenHands — the open source research project that became one of the most starred AI coding agent repos on GitHub. Robert has spent over a decade in commercial open source. Early on, as Cursor blows up and every competitor pivots toward an IDE, his team makes the opposite call: the IDE is an old form factor, and what developers need is an interface for managing fleets of agents. He walks through what that bet looks like two years in, from the banks that refuse to build on closed source to the critic model his team trains to grade its own agents.

Join us as we explore:

• The Dark Factory: What software development looks like when agents write, review, merge, and monitor code end to end — and the two jobs Robert thinks stay human.
• 20% vs. 10X: Why the pair programming model caps out around a 30% gain, and where the order-of-magnitude lift actually sits.
• The Per-Token Trap: Robert on refusing usage-based pricing while competitors post eye-popping revenue at negative gross margin.
• Open Source as Distribution: Why large banks tell OpenHands they will only build on infrastructure they own in the long term.

40% of searches now end without a click, and on many sites AI agents already outnumber human visitors. So who are you actually marketing to?

In this episode of First Commit, Eliya Elon and Laura Hamilton sit down with Aviv Shamny, co-founder and CEO of Limy AI. The problem Aviv and his co-founders chase is one most marketers haven't priced in: AI agents now discover, fetch, and convert on products across the web, while the old signals (clicks, impressions, organic rankings) erode underneath them. Limy builds the connection layer between brands and agents, operating server-side where agents actually work, so companies can see and influence how they show up in machine-driven answers.

Join us as we explore:

• Machine intent over clicks: Why Aviv chases structured machine intent instead of the impressions-and-clicks approximations marketing has always lived with.
• Server-side, not clickstream: How Limy operates where agents actually work, the server layer, rather than overlaying human browsing data onto machine logic.
• Making a site agent-ready: What actually earns agent visibility and why there's no silver bullet.
• Prime years over the beach: Why Aviv picks marketers over "insurance vibes," and what a Fortune 100's real monetization moment looks like.

Follow Notable Capital:
Find Us On X: @notablecap
Find Us On LinkedIn: @Notable Capital
Visit Our Website: https://www.notablecap.com/

Math might be the last programming language AI hasn't conquered.

In this episode of First Commit, Eliya Elon and Laura Hamilton sit down with Carina Hong, founder and CEO of Axiom, the company building an AI mathematician — and, fresh off a $200M Series A at a $1.6B valuation led by Menlo Ventures, one of the most-watched bets on "verified AI." Hong leaves a Stanford PhD in March 2025 to chase a specific window: reasoning models maturing, coding capability compounding, and the formal-proof language Lean finally becoming robust enough for industry-scale work. While frontier labs skip formal math as too data-scarce to bother with, she bets the opposite — that math is special, that it transfers the way code does, and that a system which proves its own work is worth building long before anyone is paying attention.

Join us as we explore:

• Formal vs. informal AI: A model that answers in Lean can be verified by simply running the program — Hong explains why that guarantee changes everything you can build on top of it.
• Where proofs pay off: Chip design, software verification, and cryptography get tractable the moment a plain-English spec becomes something you can actually prove.
• Research taste as strategy: Hong hires across mathematicians, formal-methods veterans, and young high-agency engineers, and makes the case that taste is the one thing a research org can't get wrong.
• The founder's obsession: The "I couldn't do anything else" moment, plus her theory on why cheap capital fragments a field instead of cracking it.

Follow Notable Capital:
Find Us On X: @notablecap
Find Us On LinkedIn: @Notable Capital
Visit Our Website: https://www.notablecap.com/

What does it take to build AI agents that enterprises will actually trust with their most critical workflows — and what happens when the company building those agents has no security clearance to see them deployed?

In this episode of First Commit, Eliya Elon and Laura Hamilton sit down with Mayada Gonimah, co-founder and CTO of Thread AI. Mayada spent her career at the deepest end of enterprise infrastructure — building prime brokerage systems at Goldman Sachs, PCI-compliant payment rails at the New York Times, and software for the Department of Defense at Palantir — without always having the clearance to see her own work run in the field. That constraint forced a discipline that now defines Thread AI: build infrastructure so reliable it functions correctly wherever it's deployed, under whatever conditions, whether you're watching or not. Thread AI builds AI-native orchestration for enterprises running mission-critical, multi-step agentic workflows — think end-to-end invoice processing, insurance claim routing, compliance monitoring — in financial services, healthcare, and public safety.

Join us as we explore:

• The orchestration gap: Why tools like Temporal solved durability but not non-determinism — and why that gap became Thread AI's founding thesis in spring 2023.
• Selling to enterprises early: How Thread AI closed contracts with some of the largest regulated organizations in the world before product-market fit was fully established — and why multi-cloud and compliance weren't optional.
• The "applied AI" role: What Thread AI's forward-deployed applied AI engineers actually do, why strong distributed systems fundamentals beat AI hype on the hiring scorecard, and why the last 10% of production always wins.

Follow Notable Capital:
 Find Us On X: @notablecap
Find Us On LinkedIn: @Notable Capital
Visit Our Website: https://www.notablecap.com/
Subscribe on Spotify: https://open.spotify.com/show/43j3cMkcKIgHuXu1tFHWpz?si=a326d0cd016f47c0
Subscribe on Apple Podcast: https://podcasts.apple.com/us/podcast/notable-perspectives/id1392649094

What does it actually take to validate a startup idea — and why should you make it harder on yourself?

In this episode of First Commit, Eliya Elon sits down with Dor Fledel, co-founder and CEO of Spera Security. Built in just under 18 months, Spera — an identity security posture management company — was acquired by Okta for a reported ~$100M. Dor then led Okta's SecureAI, PAM, and ISPM product lines before recently joining Anthropic to help enterprises adopt AI securely. A Talpiot program alumnus and Unit 8200 veteran, Dor brings an intelligence-trained rigor to how he thinks about building, validating, and committing.

Join us as we explore:

• Friction as a Feature: Why deliberately making it harder for customers, co-founders, and employees to say yes is actually the sharpest signal you have about whether your idea is real.
• Meeting People Where They Are: How Dor learned to stop pitching identity security to people who were already convinced — and start positioning for the world they've already heard ten times.
• The Personal Stack: Why the hardest co-founder conversation often has nothing to do with equity or roles — and everything to do with whether someone's partner will move across an ocean in 18 months.

Follow Notable Capital:
Find Us On X: @notablecap
Find Us On LinkedIn: @Notable Capital
Visit Our Website: https://www.notablecap.com/ 

What if your browser knew your entire workday — and could quietly handle the parts that drain your focus?

In this episode of First Commit, Eliya Elon and Laura Hamilton sit down with Regina Lin, co-founder and CEO of ThirdLayer, the company building Dex — an AI workspace that lives inside Chrome and acts as a true knowledge worker for the browser. 

Regina started building Dex as a college side project with her co-founder Kevin while juggling a math degree and years of competitive piano performance — originally just trying to get GPT-4o to order DoorDash at midnight. The insight that it could replace entire categories of manual, browser-bound work crystallized when they snuck into a Boston AI conference with fake lanyards and ran into someone managing hundreds of UiPath bots. What Dex became is a hybrid agent that pulls context directly from the browser while executing actions through precision tool calling — not screen scraping, not MCP — built for the knowledge workers who spend every minute of their day living in tabs.

Join us as we explore:

• ICP Discovery: How Regina landed on go-to-market professionals as the core user — people whose entire workday is implicit, scattered across tabs, with no to-do list because their to-dos are the tabs.
• Hybrid Agent Architecture: Why Dex combines browser context with custom tool calling for maximum accuracy — and why the "pure browser agent" framing misses what actually makes it fast.
• The Education Problem: Why giving people agents isn't enough — and how human creativity, not technology, turns out to be the real bottleneck to adoption.
• Lessons from Piano: How eight-hour practice days, public masterclass critiques, and improvising a concerto ending she'd forgotten trained Regina for the pressure of building in real time.

Follow Notable Capital:
Find Us On X: @notablecap
Find Us On LinkedIn: @Notable Capital
Visit Our Website: https://www.notablecap.com/ 

Watch on YouTube: https://www.youtube.com/@notablecapital

This week, we’re joined by Alexis Gauba, Co-Founder of Raindrop, an AI native observability platform built for agents in production.

Alexis breaks down why operating agents is fundamentally different from monitoring traditional software. As systems shift from deterministic code to probabilistic behavior, dashboards alone are not enough. Teams need to detect unknown issues, track signals like forgetting, and understand long agent trajectories across millions of AI events.

We discuss why agent observability has become essential over the past year, what makes agent infrastructure distinct from prior platform shifts, and when internal tooling stops scaling. Alexis also explains Raindrop’s approach to production monitoring, combining explicit signals with automated detection to help teams not just find issues, but fix them.

Episode chapters:

• 2:05 - Founding Raindrop
• 3:54 - Building with Close Friends
• 5:44 - What Raindrop Actually Does
• 7:45 - The Reliability Challenge of Agents
• 9:55 - Monitoring Agents at Scale
• 14:45 - Experiencing the Pain Firsthand
• 18:00 - The Rise of Agent Infrastructure
• 22:17 - Internal AI Use Cases
• 24:07 - Hiring for Initiative and Ownership
• 28:30 - The Power of Multitasking
• 32:06 - Quick Fire Round 

This episode is brought to you by Grata, the leading deal sourcing platform for private equity. Grata’s AI powered search, investment grade data, and intuitive workflows help you find and win the right deals faster. Visit grata.com to book a demo.

This episode is also sponsored by Overlap, the AI powered app that uses LLMs to surface the best moments from any podcast. Overlap reads full transcripts, finds the most relevant clips, and stitches them into a personalized stream of insights. Tap into podcasts as a real information source with Overlap 2.0, now available on the App Store.

This week, we’re joined by Anish Agarwal, CEO of Traversal, an AI-native site reliability platform helping teams detect, diagnose, and remediate incidents before they spiral into prolonged downtime.

Anish shares how Traversal is tackling the full lifecycle of reliability work: identifying what caused an incident, determining which signals actually mattered during alert triage, and helping teams plan how their infrastructure should evolve over time. As modern systems grow more complex, spanning microservices, serverless, and multi-cloud environments, the surface area for failure continues to expand, especially as AI-generated code accelerates change faster than humans can reasonably keep up.

We talk about why observability has produced some of the largest outcomes in software, why the traditional dashboard-first model is breaking down, and why the next generation of tooling needs agents that can search, reason over, and act on observability data, not just display it. Anish also breaks down where “self-healing systems” are real today, where expectations need to be reset, and why many AI-SRE products risk building faster horses instead of rethinking the experience entirely.

Episode chapters:

• 1:50 — Anish’s background and early building
• 2:30 — The origins of SRE and why it emerged
• 8:40 — Research-driven thinking in product and engineering
• 10:07 — Building for large enterprise environments
• 13:00 — AI-driven code and the growing infrastructure surface area
• 16:11 — Preparing infrastructure for an AI-native future
• 20:10 — Fear, trust, and operating critical systems
• 23:05 — From detection to automated action
• 24:20 — How the SRE role is changing
• 26:05 — Hiring and building the right team
• 29:08 — Raising capital from Sequoia and KP
• 32:38 — Reflections and lessons learned
• 35:50 — Quick-fire round 

This episode is brought to you by Grata, the leading deal sourcing platform for private equity. Grata’s AI powered search, investment grade data, and intuitive workflows help you find and win the right deals faster. Visit grata.com to book a demo.

This episode is also sponsored by Overlap, the AI powered app that uses LLMs to surface the best moments from any podcast. Overlap reads full transcripts, finds the most relevant clips, and stitches them into a personalized stream of insights. Tap into podcasts as a real information source with Overlap 2.0, now available on the App Store.

This week, we’re joined by Ivan Burazin, co-founder of Daytona - a company rethinking developer environments for an AI-native world.

We talk about how Daytona creates real value for developers, why the most advanced agent companies are emerging bottom up, and the idea that agents should be treated as first class users with reliable access to compute. Ivan also shares some of Daytona’s most demanding use cases including AI scientists in chemistry and pharma running agents inside massive sandboxes with hundreds of CPUs.

We also cover what reliability looks like when customers define success as whether the system works at all, how Daytona maintains sub 90 millisecond latency while spinning up millions of environments per day, and how they support always on usage with no fixed schedule. Finally, we dig into go to market lessons from putting engineers on the front lines early to why Daytona prioritizes in person engagement and intentional events as the most authentic way to build trust with developers.

Episode chapters:

• 1:43 – Market timing and why now
• 3:18 – Building for developers and dev tools
• 4:55 – Global developer communities
• 7:55 – Computers as infrastructure for agents
• 10:32 – Product-led growth
• 12:17 – Enterprise use cases and adoption
• 16:25 – Managing cost, performance, and latency
• 19:20 – Hiring for resiliency at scale
• 20:40 – Internal AI use cases at Daytona
• 24:25 – Creating a bottom-up go-to-market motion
• 27:42 – Hiring and scaling developer relations
• 29:55 – Partnerships and ecosystem strategy
• 31:00 – Quick-fire round 

This episode is brought to you by Grata, the leading deal sourcing platform for private equity. Grata’s AI powered search, investment grade data, and intuitive workflows help you find and win the right deals faster. Visit grata.com to book a demo.

This episode is also sponsored by Overlap, the AI powered app that uses LLMs to surface the best moments from any podcast. Overlap reads full transcripts, finds the most relevant clips, and stitches them into a personalized stream of insights. Tap into podcasts as a real information source with Overlap 2.0, now available on the App Store.

This week, we’re joined by Ishan Gupta, co-founder of Juicebox — a company redefining how recruiting works in an AI-native world.

Ishan shares how they pivoted from earlier ideas to focus on the highest value part of hiring: identifying the right people and getting them into process. With Juicebox, you describe what you want in natural language and the platform searches more than 800 million professional profiles, surfaces the best matches, and engages candidates through recruiting agents.

Their breakout feature, Autopilot, uses LLMs to semantically evaluate and stack rank candidates based on nuanced criteria, which quickly drove organic growth and strong PMF. We talk about why their data makes the product so sticky, how they see recruiting evolving over the next five years, what led them to raise their Series A, and the culture they are building around moving fast and being intellectually honest.

Ishan also shares what they are building next with their memory layer, which will make it clear what the AI is learning over time and how it improves future searches.

Episode chapters:

1. 1:44 - Competitive programming
2. 4:35 - Choosing the company name
3. 5:55 - YC and the pivot
4. 8:00 - How Juicebox differs from traditional recruiting
5. 9:51 - The killer feature
6. 14:30 - What makes the product sticky
7. 17:15 - Is recruiting a zero sum game
8. 19:10 - The macro view of hiring
9. 22:05 - Raising a later Series A
10. 24:10 - Product expansion
11. 26:45 - Quick fire round 

This episode is brought to you by Grata, the leading deal sourcing platform for private equity. Grata’s AI powered search, investment grade data, and intuitive workflows help you find and win the right deals faster. Visit grata.com to book a demo.

This episode is also sponsored by Overlap, the AI powered app that uses LLMs to surface the best moments from any podcast. Overlap reads full transcripts, finds the most relevant clips, and stitches them into a personalized stream of insights. Tap into podcasts as a real information source with Overlap 2.0, now available on the App Store.