Table of Contents
AI is revolutionizing startups and science, giving rise to leaner teams, deeper innovation, and a bold new era of biological engineering, operational leverage, and human expression.
Key Takeaways
- AI enables startup teams to scale massive impact with fewer people by automating workflows in hiring, R&D, and deployment.
- Friedberg’s company Oh is advancing CRISPR-powered plant engineering using genome models trained on global agricultural data.
- Modern biology now relies on AI-driven predictions of genetic outcomes, protein synthesis, and crop viability at scale.
- Organizational AI immersion—through hackathons and workflow mandates—ensures every team builds with and through AI.
- The cost of sequencing a genome has dropped by nearly one million-fold, unlocking biocomputation as an accessible design domain.
- AI is speeding up drug design and enabling biological systems to be edited like software, with predictive testing for function.
- AI-generated content will allow individualized cultural experiences—films, music, and narratives built on personal tastes and timelines.
- Friedberg insists startups embed AI at the foundation, not as an overlay, to compete in future markets.
- AI is not shrinking the role of startups—it's expanding their ambition, compressing timelines, and unlocking billion-dollar domains.
Building an AI-First Startup Culture
- David Friedberg returned to lead Oh in late 2023 after observing its internal breakthroughs in genome-editing and machine learning for plant trait optimization.
- He mandates every headcount request be evaluated against AI capability. If AI can perform the function, no hire is approved.
- AI scripts at Oh now handle initial resume screening, sourcing, and fit scoring—automating what used to take recruiters days into mere hours.
- In scientific workflows, AI agents read research papers, evaluate lab conditions, and propose detailed experiments using internal datasets and literature.
- This isn’t about cost-cutting—it’s about scale. Lab productivity has increased 10x without proportional increases in staff.
- Leadership teams engage in regular AI hackathons to maintain fluency in prompt engineering, model fine-tuning, and decision tree optimization.
- Company culture includes peer-training in AI tools, incentivizing employees to discover, share, and improve workflows with open-source models and custom APIs.
How AI Shrinks Big Problems and Amplifies Ambition
- Friedberg argues that traditional limits on startup ambition—capital, scale, timeline—are being erased by AI.
- He illustrates this with examples like rare-earth mineral extraction, where automated mining plans can now be generated algorithmically.
- In the past, industries like aerospace, energy, and agriculture required vast headcounts and multibillion-dollar budgets. Today, 20-person teams can prototype entire ecosystems.
- AI agents plan, budget, simulate, and optimize multi-phase projects like satellite constellations, chemical plants, and even lunar habitats.
- These tools transform entrepreneurs from builders to orchestrators—guiding adaptive systems that design, test, and iterate faster than any human team.
- As Friedberg notes, startups can now outmaneuver Fortune 500s in core markets by building AI-native infrastructure from day one.
Biology Meets Code: Digitizing Life for Better Health and Food
- DNA sequencing, which cost $100M in 2000, now costs under $100, allowing millions of organisms to be digitally modeled and modified.
- Oh uses genome-scale models (GLMs), akin to GPT-style architectures, to predict how DNA changes affect physical traits in plants.
- These models suggest precise edits to CRISPR sequences, shortening breeding cycles from decades to months.
- In one case, Oh used GLMs to produce potatoes with 90% reduced planting mass, which required less fertilizer, water, and generated less waste.
- Protein prediction models simulate 3D folding and functional binding in real time—enabling biotech labs to design new therapies before wet-lab testing.
- Friedberg likens this to programming biology: "Instead of compiling code, we culture cells."
- In human therapeutics, these same AI models help discover biologics that reset aging processes, enhance immune targeting, and suppress cancer growth by rewriting epigenetic markers.
- EVO2, Oh’s internal model, incorporates genomics from global agricultural datasets, guiding real-time edits and live simulation.
Rewriting Culture and Media Through AI
- AI enables dynamic media construction. Consumers could request "Star Wars from the villain's POV as a noir detective drama" and receive it instantly.
- Creators define rule sets—tone, ethics, character boundaries—and users explore infinite permutations within that structured sandbox.
- Friedberg envisions a future where culture becomes layered: a shared canon (like the Bible), and infinite derived stories (like interpretations).
- AI-generated music, films, and books will coexist with originals, offering immersive personalization while preserving collective touchstones.
- Far from replacing creators, AI tools will elevate them into world-builders, setting constraints and fostering fan-generated works at scale.
- This unlocks new economic models for creators, enabling monetization of derivatives, real-time collaborations, and niche fandoms.
Taming Hallucinations and Scaling Trust in AI Systems
- While hallucinations remain a concern, Friedberg sees structured solutions emerging fast: QA agents, model triangulation, and domain-specific validators.
- Ensemble outputs (where multiple models cross-check answers) are already common in sensitive domains like finance, law, and drug discovery.
- Chain-of-thought reasoning and modular output chains allow outputs to be traced, audited, and corrected in real time.
- Friedberg compares this to cloud computing: once messy, now streamlined via foundational architectural patterns.
- Crucially, users won’t see the verification backend—just better, more consistent, and trusted performance across all applications.
A startup in 2025 isn’t just faster or cheaper—it’s more audacious. AI reshapes not just how we build, but what we build. In biology, storytelling, and beyond, the rules are being rewritten.
We are entering an age of engineered abundance. Creativity, scaled by computation, is our new currency.
