Launching soon — join the waitlist

Command a team of
AI research agents.

Ypotheto turns a materials-research brief into ranked, source-backed candidates — generated, challenged, and tournament-tested by a coordinated team of AI agents. You stay in command of the why behind every hypothesis.

ABSTRACT · NOMENCLATURE

ὑποτίθημι (hypotithēmi)

Ancient Greek: “to place under; to propose.” The root of hypothesis — and of Ypotheto. Because every discovery starts with someone daring to propose an idea, and every real result comes from trying as hard as possible to break it.

FIG. 01 The problem

Screening candidate materials by hand doesn’t scale.

01

Scattered literature

The evidence you need is spread across thousands of papers, patents, and datasheets. Nobody can read them all — so shortlists get built on the fraction one person happened to find.

02

One brain, one guess

A single researcher — or a single chat with an LLM — produces one linear line of reasoning, with nothing forcing it to check its own work against the evidence.

03

Untraceable claims

By the time a candidate reaches a review committee, the “why” behind it is a memory. Reconstructing the sources costs weeks — if it’s possible at all.

FIG. 02 System architecture

Four agent roles. One command structure. Yours.

Define the research goal — a drop-in substitute for a restricted polymer, a lower-cost alternative to an incumbent material — and direct the agents through it.

Stage 01

Generate

A team of generation agents proposes distinct candidate structures against your target criteria — a slate of ideas, not a single guess.

MODULE: STRUC-GENACTIVE
Stage 02

Reflect

Reflection agents adversarially challenge every claim against literature they retrieve. Weak reasoning doesn’t survive contact with the evidence.

MODULE: LIT-VERIFYACTIVE
Stage 03

Rank

Survivors face each other in tournament rounds. The strongest contenders rise across iterations — scored on the science and your commercial constraints.

MODULE: RANK-TOURNACTIVE
Stage 04

Evolve

Top hypotheses are mutated, recombined, and sent back into the arena. Each round starts from a stronger population than the last.

MODULE: EVO-LOOPACTIVE

FIG. 03 Live inside every run

Watch hypotheses fight for survival.

This is the loop your agents run — propose, verify, cull, rank — over and over, while you steer.

ROUND 1 — GENERATION SIMULATION

FIG. 04 No black boxes

Every claim traces back to a source you can check yourself.

The output isn’t a transcript you have to reconstruct and fact-check. It’s a structured, source-linked shortlist you can hand straight to a review committee — every hypothesis carrying the literature that survived its interrogation.

  • Retrieved literature attached to every hypothesis
  • Full reasoning chain from brief to ranking
  • Commercial feasibility scored alongside the science

Be there when the agents
go to work.

Ypotheto is launching soon. Leave your email and you’ll be the first to know — nothing else, no spam, ever.