- AngelsRound
- Posts
- Core Automation
Core Automation
Frontier AI lab building continual learning systems
Spotlight
What if every frontier AI lab is scaling the wrong architecture?
Quick Pitch: Core Automation is building AI that learns continuously instead of through repeated retraining, while using AI agents to automate its own research lab.
The Problem
Scaling Costs More: Each new generation of frontier AI requires significantly more compute, data, and capital than the last.
Incumbents Are Locked In: The largest AI labs are heavily invested in today's architecture, making fundamental change difficult.
Models Stop Learning: Most AI systems stop improving once deployed, requiring entirely new training cycles to get better.
Why It Matters
Frontier AI has become one of the largest infrastructure races in history.
The biggest labs are investing billions in compute, data, and talent to train ever larger models.
According to Stanford's 2025 AI Index, GPT-4 required an estimated $78 million in training compute, while Gemini Ultra reached $191 million. Meanwhile, training compute for frontier models continue to double every five months.
As costs continue to rise, the search for more efficient ways to build AI is becoming increasingly important.
Snapshot
Industry: Frontier AI Research
Headquarters: San Francisco, CA
Founded: 2026
Backed By: Spark Capital, Nvidia, Sequoia Capital, Accel, and others
Funding: Raised $100M (Pre-Seed), Raising $400M (Seed)
Traction: Founding team from OpenAI, Google DeepMind, Adept AI, and PyTorch with significant GPU commitments secured
Target Customers: AI infrastructure providers, enterprises, and eventually autonomous AI deployments
Founder’s Edge
Jerry Tworek: Former VP of Reinforcement Learning Research at OpenAI, where he led research behind the o1 through o4 reasoning models.
Julia Villagra: Former Chief People Officer at OpenAI.
Playing Field
Frontier AI Labs: OpenAI, Anthropic, Google DeepMind, and xAI continue scaling transformer-based models.
Alternative Research: Most startups focus on improving today's AI systems.
Core Automation's Edge: Pursuing continual learning beyond repeated retraining while using AI agents to accelerate research.
Analysis
Bulls Case 📈
Founding team helped build today's frontier AI models.
A new architecture could outperform incremental scaling.
Continual learning could reduce AI training costs.
Unconstrained by legacy AI architectures.
Bears Case 📉
Replacing transformers is one of AI's hardest technical challenges.
Large AI labs can rapidly adopt promising architectural advances.
Scientific breakthroughs can take years to validate and commercialize.
Converting frontier research into a repeatable business remains unproven.
Verdict
Every frontier AI lab today is investing in larger models, more compute, and bigger training runs. Core Automation is betting the industry's next breakthrough won't come from scaling today's architecture, but from replacing it. If they're right, the next frontier AI leader may not be the company with the most compute. It may be the one with the better architecture.
Who's Hiring — Austin Tech
More than $2.4B has been invested into Austin startups this year. That capital is now translating into hiring across engineering, product, AI, and GTM.
Apptronik — Humanoid robotics — hiring across robotics, AI, software, and hardware.
NinjaOne — IT management platform — hiring across engineering, product, sales, and customer success.
Function Health — AI-powered health platform — hiring across engineering, product, clinical, and operations.
Halcyon — Anti-ransomware platform — hiring across engineering, security, product, and GTM.
SpyCloud — Identity threat protection — hiring across engineering, product, security, and sales.
The Startup Pulse
Another happening week in startup funding. Three signals from this week:
Vertical AI is attracting increasingly larger rounds
AI applications are beginning to rival infrastructure in investor interest
Investors are backing businesses that solve industry specific problems, not just better models
Fireworks AI — Secured a $1.505B Series D. As enterprises adopt more open source AI models, demand is growing for infrastructure that makes them production ready.
Neko Health — Closed a $700M funding round. Investors are betting AI will shift healthcare toward earlier detection and preventive care instead of reactive treatment.
Wonder — Announced a $650M Series D. The company is building an AI powered platform that brings together food delivery, logistics, and restaurant operations ahead of a planned public offering.
Want to get the most out of ChatGPT?
ChatGPT is a superpower if you know how to use it correctly.
Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.
Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.
Written by Ashher
