# AI’s New Moat Is the Work Loop

*By VC Tech Radar • August 13, 2026*

Tracks the strongest signals around self-improving domain labs, verifiable agent environments, robotics reliability, post-training economics, and outcome-based vertical AI.

## 1. Funding & Deals

**Event Horizon Labs is an early-stage bet on self-improving domain expertise.** Dalton Caldwell says he led the Series A in EHL, a new quantitative research lab founded by Owen. Its thesis is to automate the type of work Owen previously did at Citadel with self-improving AI; Caldwell calls Owen one of YC’s most talented AI researchers and says several teams are pursuing the same direction. [^1]

## 2. Emerging Teams

**Suhail’s autonomous-AI-scientist project is moving from experiment to compute-backed team.** The build log says the project has a validated basic RLVR post-training stack, made its first hire while recruiting for post-training or low-level optimization, and grew from one person to three. After a key research component worked but needed scaling—and GPUs were delayed by networking issues—Suhail reported that much greater quantities of compute were locked down and ready. [^2][^3][^4][^5][^6][^7]

The diligence question is whether the validated post-training loop compounds once the new infrastructure is online; the founder’s own updates frame scaling research and compute access as the immediate bottleneck. [^6][^7]

## 3. AI & Tech Breakthroughs

**Agent training is becoming an environment-and-verifier business.** A current industry talk describes the shift from low-skilled crowdsourced behavior-cloning data toward expert-built environments containing realistic “worlds,” high-fidelity app clones, and tasks with rubric or unit-test verifiers. The speaker reports 2.5 million expert hours in the second quarter; in one 1,800-task post-training run using about $500,000 of compute, the reported score rose from 4.7% to 26% and generalized to other benchmarks. [^8]

That points to a valuable infrastructure layer beyond raw model supply: humans still have to measure performance beyond a model’s frontier, while the next data wave is moving toward 100–1,000-hour tasks and social interaction—an area where the speaker says only about 1% of evaluations currently measure performance. [^8]

**Robotics reliability is being attacked with reinforcement learning and memory, not demos alone.** The robotics talk reports a general-purpose value function trained on robot experience, with human intervention used to avoid dead-end trajectories; the resulting policy ran for 13 hours, exceeded 90% espresso success, and gained roughly 2x box-building throughput from the RL stage. [^9] Multi-timescale memory—short video memory plus compressed text for longer history—enabled a non-repetitive kitchen-cleaning task lasting 10–15 minutes, while a single Pi07 model reportedly matched or outperformed fine-tuned specialists and generalized to scarcely represented appliances and a new robot platform. [^9]

## 4. Market Signals

**The near-term moat is shifting from the base model to post-training, domain data, and cost control.** The post-training thesis is to use frontier models to reach product-market fit, then use product data to encode a company’s distinctive taste and expertise. Because applications can be cloned from screenshots, the tuned model becomes the proposed defensible asset; the speaker says post-training can also reduce serving cost by 5–10x. [^10] The same talk warns that reward hacking and training-to-serving drift can make apparent progress useless, making repeatable evals and production A/B tests part of the product rather than a launch afterthought. [^10]

**Vertical agents are being framed around cash outcomes rather than task volume.** Stuut’s collections analysis covers billions of dollars of receivables and finds that the largest 10% of past-due invoices hold 65.7% of overdue dollars. In its data, 81.7% of outbound collection emails require no human involvement, while escalations, disputes, manual calls, and broken-promise follow-ups remain human-resolved; agent-run teams average 1.35 outbound asks per $1,000 collected, with three in five completed tasks resolving fully automatically. [^11] The product pattern is clear: automate the search and repetitive asking, while reserving people for relationships and exceptions.

**AI is changing founder leverage faster than it is changing the fundamentals of company building.** Garry Tan argues that agentic and vibe coding can make one person “400 of that person” and expects a wave of experienced 35–45-year-old technical founders. Paul Graham’s counterweight is that almost all standard startup advice still holds: the core remains building what users need and finding growth. [^12][^13][^14][^15] The investment implication is to widen founder sourcing toward experienced operators without relaxing scrutiny on distribution and durable demand.

**Local and open inference is becoming a meaningful deployment channel.** Hugging Face says Transformers.js crossed 10 million monthly downloads—nearly 10x its level six months earlier—and attributes local adoption to free, private execution amid compute shortages and cyber-attack risk. Cohere’s North Micro Vision adds to the supply side: its smallest vision-language model is open-source under Apache 2.0 for document understanding. [^16][^17]

## 5. Worth Your Time

- **Watch [RL Environments Explained: How AI Agents Learn Real-World Work](https://www.youtube.com/watch?v=a00xIn5kwhM).** The clearest explanation in the corpus of why agent progress depends on realistic apps, expert-built tasks, and verifiers—not just more model-generated trajectories. [^8]

[![RL Environments Explained: How AI Agents Learn Real-World Work | Brendan Foody, Mercor](https://img.youtube.com/vi/a00xIn5kwhM/hqdefault.jpg)](https://youtube.com/watch?v=a00xIn5kwhM&t=179)
*RL Environments Explained: How AI Agents Learn Real-World Work | Brendan Foody, Mercor (2:59)*


- **Watch [Chelsea Finn: This is the State of the Art in Robotics](https://www.youtube.com/watch?v=cRZNwgvcWUg).** Focus on the memory and generalist-policy sections for a concrete view of how robotics is moving from isolated skills toward longer, compositional tasks. [^9]

[![Chelsea Finn: This is the State of the Art in Robotics](https://img.youtube.com/vi/cRZNwgvcWUg/hqdefault.jpg)](https://youtube.com/watch?v=cRZNwgvcWUg&t=1068)
*Chelsea Finn: This is the State of the Art in Robotics (17:48)*


- **Watch [Post-Training Is How You Keep Your Taste](https://www.youtube.com/watch?v=yAvJ7b_FxUA).** Useful for deciding when a startup should rent frontier intelligence and when its product data, evaluation loop, and economics justify owning a specialized model. [^10]

[![Post-Training Is How You Keep Your Taste | Fireworks CEO Lin Qiao](https://img.youtube.com/vi/yAvJ7b_FxUA/hqdefault.jpg)](https://youtube.com/watch?v=yAvJ7b_FxUA&t=1545)
*Post-Training Is How You Keep Your Taste | Fireworks CEO Lin Qiao (25:45)*


- **Read [Paul Graham on startup advice in the AI era](https://x.com/paulg/status/2087601208123126228).** A short corrective to claims that AI has already overturned the basic work of finding a product users want. [^14][^15]

---

### Sources

[^1]: [𝕏 post by @daltonc](https://x.com/daltonc/status/2087595898998288681)
[^2]: [𝕏 post by @Suhail](https://x.com/Suhail/status/2064418847428608493)
[^3]: [𝕏 post by @Suhail](https://x.com/Suhail/status/2071246378504998916)
[^4]: [𝕏 post by @Suhail](https://x.com/Suhail/status/2075596761511702823)
[^5]: [𝕏 post by @Suhail](https://x.com/Suhail/status/2084905990596776415)
[^6]: [𝕏 post by @Suhail](https://x.com/Suhail/status/2083762613893353504)
[^7]: [𝕏 post by @Suhail](https://x.com/Suhail/status/2087772563816763700)
[^8]: [RL Environments Explained: How AI Agents Learn Real-World Work | Brendan Foody, Mercor](https://www.youtube.com/watch?v=a00xIn5kwhM)
[^9]: [Chelsea Finn: This is the State of the Art in Robotics](https://www.youtube.com/watch?v=cRZNwgvcWUg)
[^10]: [Post-Training Is How You Keep Your Taste | Fireworks CEO Lin Qiao](https://www.youtube.com/watch?v=yAvJ7b_FxUA)
[^11]: [𝕏 article by @realtarek](https://x.com/i/article/2087329858305044480)
[^12]: [𝕏 post by @a16z](https://x.com/a16z/status/2087676551735418967)
[^13]: [𝕏 post by @a16z](https://x.com/a16z/status/2087694916768194868)
[^14]: [𝕏 post by @paulg](https://x.com/paulg/status/2087601208123126228)
[^15]: [𝕏 post by @paulg](https://x.com/paulg/status/2087602421791105033)
[^16]: [𝕏 post by @ClementDelangue](https://x.com/ClementDelangue/status/2087518483718545533)
[^17]: [𝕏 post by @cohere](https://x.com/cohere/status/2087571573947392419)