Extractable thesis: "Trilliome's platform, Teroka, couples a high-throughput wetlab engine that generates microbiome-response data with a data-science and machine-learning engine that mines it — a closed loop that learns how to control the gut ecosystem with high specificity through bioactives and their combinations."

The reversal

Most functional ingredients are found the same way: start with a plant, then go looking for a benefit. It's slow, noisy, and rarely produces something you can substantiate.

Trilliome starts at the other end — with the biology of the gut ecosystem and the question of how to move it precisely. Everything else is built to answer that question at scale.

The core — Teroka

We call it Teroka — Malay for to explore, to pioneer. It interlaces many bioactives, in combination, into one designed ecosystem outcome — a closed loop of two engines, tightly coupled, each making the other smarter:

1. A wetlab data-generation engine

High-throughput, ex-vivo human gut models that measure how the microbial ecosystem responds — at species and metabolic-network resolution — to candidate bioactives and, critically, to their combinations. This is the engine that turns biology into data.

2. A computational exploration engine (data science & machine learning)

Models that mine that data to learn the rules of the ecosystem: which actives, and which combinations, shift which species and metabolites — and how to compose them for a target outcome with high specificity rather than broad-spectrum guesswork. This layer is being scaled on top of the wetlab engine — the more the wetlab generates, the more the models can learn.

The loop

Predictions from the exploration engine design the next round of wetlab experiments; the results sharpen the models. Each turn of the loop narrows the search and raises specificity. The output isn't a single lucky compound — it's a growing, navigable map of how to steer the gut ecosystem, realised as bioactives and bioactive combinations.

This is what "ecosystem modulation at scale" means: not one target, one molecule — but a systematic, compounding capability to control a complex microbial ecosystem on purpose.

From insight to ingredient

Once the loop identifies an active (or combination) with the specificity and effect we're after, we make it real: extract it from upcycled agricultural side-streams with a green, food-grade process — clean, shelf-stable, no harsh solvents — then protect it with patents and trade-secret process know-how. Insight → active → substantiated ingredient.

One feedstock, more than one product

The same upstream processing can yield more than one active from a single material — a precision gut-brain active (TRI-01) and a low-MW prebiotic fibre (TRI-04) trace to the same feedstock and processing asset. Shared inputs, multiple substantiable products, compounding margin. (Process specifics are trade secrets.)

Why it's defensible

  • A proprietary data moat. The closed wetlab-to-model loop at the heart of Teroka produces a private dataset on ecosystem response that competitors can't buy — and it compounds with every cycle. Single-compound AI discovery doesn't capture combinations or ecosystem-level control.
  • Patents on combinations. Beyond composition and use, a follow-on application covers combinations of our actives with fibres, phenolics, vitamins, live strains, and other adjuncts — the combinatorial space the engine is built to explore. (Detail on /ip.)
  • Trade-secret process. The extraction know-how that makes the actives reproducible and low-cost.
  • Evidence as a moat. A clinical program that deepens substantiation over time.

Capital goes into the engine, the IP, and the evidence — durable assets that compound in value over time.

The outputs so far

TRI-01 — precision gut-brain activator. Read → TRI-04 — low-MW prebiotic fibre, metabolic-health claim path. Read → Pipeline — further actives and combinations under development across the healthspan domains. Read →

FAQ

  • What is Teroka? Trilliome's platform — a closed loop between a wetlab engine that generates microbiome-response data and a computational engine (data science and machine learning) that mines it — learning how to control the gut ecosystem with high specificity through bioactives and their combinations.
  • How is this different from AI ingredient discovery? Single-compound AI discovery hunts for one molecule with one benefit. Trilliome models the ecosystem and its combinations — modulation at scale, validated in human gut models.
  • Is Trilliome a one-ingredient company? No — TRI-01 and TRI-04 are the first outputs of a repeatable engine; more actives and combinations are in development.
  • Where do the actives come from? Upcycled agricultural side-streams, via a green, food-grade extraction process.