Andrew Chen (a16z speedrun) on Agentic Engineering, Speedrun, and the Future of Building Software
- George Bandarian
- 5 days ago
- 2 min read
Season 2, Episode 1 of The Agentic Podcast with George Bandarian (Untapped Ventures)
AI isn’t just changing how we code - it’s changing what a “company” even is.
In the Season 2 premiere of The Agentic Podcast, George Bandarian, General Partner at Untapped Ventures, sits down with Andrew Chen, General Partner of a16z Speedrun to unpack what’s happening inside the AI wave.
From a16z’s $15B AI fundraise to the rise of agentic engineering (formerly vibe coding), this conversation explores how product, growth, retention, and company building are being rewritten in real time.
If you’re building or investing in AI-native startups, this episode sets the tone for what’s coming next.
Overview
Andrew shares why he believes the current AI wave could rival - or even surpass - the combined impact of mobile, internet, and cloud.
The conversation then moves into:
What actually gets founders accepted into a16z speedrun
Why early-stage investing is still mostly about people
The shift from “vibe coding” to agentic engineering
Why distribution channels feel crowded
And why retention curves in AI look scary, but can still produce explosive ARR growth
George and Andrew also discuss Nexxa.ai, a company they co-invested in, as an example of what strong early traction plus domain depth looks like inside the Speedrun ecosystem.
Key Takeaways
1. AI Could Be the Biggest Technology Wave Yet
Andreessen Horowitz recently raised $15B across new funds, reflecting the scale of opportunity investors see emerging from AI. According to Andrew, the impact of AI could rival - or even exceed - the combined effect of the internet, mobile, and cloud computing.
2. Getting Into Speedrun Is About Signal, Not Pedigree
Speedrun receives tens of thousands of applications each year, but acceptance isn’t about resumes or credentials. The strongest founders typically show one of two signals: real traction or exceptional ability demonstrated through past work, projects, or domain expertise.
3. The Best AI Startups Combine Domain Expertise + AI
Many successful companies in this wave aren’t simply adding AI features - they’re rethinking entire workflows. Nexxa.ai, a company George and Andrew co-invested in, is a good example: applying AI agents to complex industrial environments.
4. Distribution and Retention Are the Real Challenges
AI makes it easier to build products, but distribution hasn’t fundamentally changed yet. Many AI startups rely on social launches, early adopters, and product-led growth. Retention curves can look brutal early on - yet revenue can still grow quickly when power users or enterprise customers drive the majority of value.
5. Coding May Become as Easy as Creating Content
As AI coding tools improve, building software could start to resemble creating media. Smaller teams - or even solo founders - may launch large numbers of fast-built apps, including experimental or “disposable” software designed for specific use cases.
In that world, brand, distribution, and community could become stronger moats than code itself.
Watch / Listen to the Full Episode
Building in AI?
If you’re building an AI company at the pre-seed or seed stage and want to explore investment from Untapped Ventures, you can pitch us here.
Untapped Ventures writes checks of ~$1M and partners closely with founders early.
Comments