YC Demo Day is in 2 days, June 16
Group 3 of 4 is the hardest one to score from the outside: AI Meets the Real World. 39 startups building atoms, not bits, robots, drones, satellites, reactors, defense. Group average LRS: 46.3, the lowest of the four groups so far.
Same method: 6 public signals, a 0 to 100 Launch Readiness Score (LRS). Search demand, social pain, competition, monetization, funding, urgency. Public data only, the outside view every investor can run before the meeting.
Highest scored:
1. Eden Robotics, 61.8, robots for physical work
2. qomplement, 60.1, agentic ERP for supply chain
3. Dayjob, 58.9, AI dispatch for short-haul trucking
4. Andustry, 57.5, supplier discovery for manufacturers
5. Hexa, 54.8, autonomous ops for industrial distributors
Lowest on the outside view:
35. matforge, 36.8, AI scientists for new materials
36. Saudara AI, 34.1, AI sourcing for Indonesian factories
37. Apollo Atomics, 33.8, ultra-compact nuclear reactors
38. Dispatch, 32.9, reusable heat shields for cargo return
39. Ornadyne, 31.2, bird-like surveillance drones
When the score reads a team low here, it usually means the opposite of a weak company. Atoms do not generate search, and a low public signal often sits on top of signed pilots, defense contracts, or a category that buys through procurement, not Google. Closing that gap is the Demo Day job.
3 things the data says about this group:
1. The demand is invisible, and it is brand-shaped. We score these from the outside on public search, and the median is 100 searches a month, 30 of 39 under 500. But the category keyword is near-dead even for the giants: "microreactor" gets 450 searches a month while "Oklo" gets 66,000. Across this group the incumbent brand outsearches the problem term by a median of 17 times. Demand in atoms hides on brand names and inside procurement, not in keywords, so an outside-in search signal under-rates the whole category. The teams that look weak here are the ones whose buyers were never going to be found through a keyword.
2. The buyers split in two, with no middle. Purchase intent is bimodal: 14 of 39 sit under 20% transactional, researchers still asking whether the thing is possible, and 14 sit over 80%, operators ready to sign. Almost nothing in between. Half this group is selling to curiosity and half to budget, and which side you are on changes the entire go-to-market.
3. The capital already chose sides. Every company here gets the same "under-served" label, but the funding records tell a different story. 24 of 39 sit in a category where a competitor already raised 100 million dollars or more in the past year: Apollo Atomics against X-energy's 700 million, warehouse robots against Pudu's 150 million, humanoid eldercare against Apptronik's 350 million. Only 6 are genuinely open lanes with no nine-figure incumbent, data-center cooling, subsea, supplier discovery, short-haul dispatch. In atoms the moat is capital and contracts, so the real question is not market size, it is whether you can out-deploy a competitor who already has the war chest.
How to use it:
- Founders in the batch: find your card, look at your lowest of the 6 signals. If it is demand or competition, that is probably the public-data lens missing your buyer, and the room will ask you to prove the private traction it cannot see. If it is monetization, bring real per-unit economics.
- VCs and angels: want the full breakdown on any project, the named competitors, and the real funding history of the category? The board is free at
fluenta.space. DM us for the full per-project file, same-day, no charge. Or pull every report over the Fluenta MCP and run your own diligence in Claude or Cursor.
- Everyone watching from outside the room: this is the group that gets cloned slowest and copied hardest once it works. The highest scorers are the teams that found a buyer in a market with no search term. Scroll it if you want to see where the next decade of hard things gets built.
Browse all 194 scored:
fluenta.space/app
Tomorrow, Group 4 of 4: Care & Capital.
Tagging the teams above in case you want to grab your report. No hidden subscription, no signup gate - the analysis is already done, the file is yours to pick up. DM here or email hello@fluenta.space and we'll send it same-day.
@StamatisTWIY
@andresfgg_ @kerimtaray @qomplementai @hetdv @JStuerken @andustry_hq @advaith_sridhar @saudaraAI @edwardharyono_ @jenandtech_ @ApolloAtomics @dispatchspace @geourgkivijian
If the score reads you wrong, that gap is the thing to close on stage.