- AngelsRound
- Posts
- Synthetic Sciences 🔬
Synthetic Sciences 🔬
AI co-scientists for research
Spotlight
What if AI becomes the world's most productive researcher?
Quick Pitch: Synthetic Sciences builds AI co-scientists that help researchers automate scientific research from hypothesis to publication.
The Problem
Research Is Fragmented: Scientists still juggle separate tools for literature review, coding, experiments, compute, and publishing.
Infrastructure Slows Discovery: Managing GPUs, environments, and model training often takes as much effort as the research itself.
AI Stops Too Early: Most AI tools help answer questions. Few can carry context across an entire research project.
Why It Matters
AI has become remarkably good at generating information. The next frontier is generating discoveries.
Scientific R&D is a $2 trillion industry, yet much of the work remains manual. Open weight models and cheaper compute are making specialized research agents practical. If they can accelerate discovery, the companies building research infrastructure could become as valuable as the companies building foundation models.
Snapshot
Industry: AI for Scientific Research
Headquarters: San Francisco, CA
Founded: 2025 (YC W26)
Backed By: Y Combinator, a16z, Pioneer Fund
Funding: Raised $1M (Pre-Seed) • Raising $2.5M Seed
Traction: Early revenue • Institutional interest secured
Target Customers: Researchers, research labs, and R&D teams
Founder’s Edge
Ishaan Gangwani: #1 in India’s AI Olympiad selection, has reached the highest level in competitive programming and has published research at top global AI conferences.
Aayam Bansal: Built a COVID-19 helpline serving 50K+ users and sold a patented AI orthopaedic device at 17.
Playing Field
General AI Models: Powerful assistants but not designed for end to end scientific research.
AI Research Platforms: FutureHouse and Sakana AI are building systems that automate parts of scientific discovery.
Big Tech: Google DeepMind and Microsoft Research validate the opportunity but are not broadly commercialized.
Synthetic Sciences' Edge: Connects the full research workflow while continuously learning from researcher interactions.
Analysis
Bulls Case 📈
Benchmark leading performance shows early technical differentiation.
Early institutional demand validates adoption beyond individual researchers.
Research workflows generate proprietary data that compounds over time.
Bottom up adoption creates a path into research institutions.
Bears Case 📉
Frontier AI labs could expand into scientific research quickly.
Researchers may be slow to trust AI with critical scientific work.
The path from research software to valuable IP remains unproven.
Scientific validation takes time, slowing commercial adoption.
Verdict
The near term business is selling productivity to researchers. The bigger opportunity is owning the feedback loop between hypothesis, experiment, and result.
If Synthetic Sciences captures that loop, it builds a proprietary dataset that becomes more valuable with every experiment. That's the foundation for a much larger business than AI software alone.
Who's Hiring — Startups That Just Raised $100M+
While many companies are slowing hiring, some of the fastest growing AI and infrastructure startups are expanding their teams after major funding rounds.
Baseten — AI inference platform — hiring across engineering, infrastructure, product, and GTM.
Sierra — Enterprise AI agents — hiring across engineering, product, design, and GTM.
Together AI — Open source AI infrastructure — hiring across engineering, research, and developer relations.
Ramp — Finance automation platform — hiring across engineering, product, design, and operations.
Groq — AI inference chips and cloud — hiring across hardware, engineering, AI, and operations.
The Startup Pulse
Another happening week in startup funding. Three signals from this week:
AI is becoming increasingly industry specific
Enterprise AI adoption continues accelerating
Infrastructure remains the foundation, but applications are where capital is spreading
Prime Intellect — Secured a $130M Series A to build distributed AI infrastructure, betting the future of AI training won't be controlled by a handful of hyperscalers.
Norm AI — Closed a $120M Series C as enterprises look for AI to automate compliance alongside the AI systems they're deploying.
Ollama — Raised a $65M Series B as more companies choose to run AI models locally to improve privacy, security, and control.
Written by Ashher
