Matterialis for agentic labs and AI-scientist platforms

Your agent can reason about chemistry. It cannot look up a material.

Matterialis plugs real, purchasable materials into your agent the way search plugs into a language model: 200,000+ supplier grades, 500+ properties, substitutes ranked against constraints, and every value marked measured or predicted. Your agent calls it over an MCP server.

Request beta access Book a technical call MCP server: beta, in development. Onboarding partner platforms now.
ClaudeCowork matterialis · MCP
You

We need to drop AMP-9140 from the barrier film. Has to hold 60 MPa tensile and 120 °C, clear RoHS and REACH, and be sourceable in the UK. What are my options?

Claude

Checking the library rather than guessing. Calling the Matterialis server.

matterialisfind_substitutes200
IncumbentAMP-9140
Tensile, min60 MPa
Service temp, min120 °C
ComplianceRoHS, REACH
3 candidates returnedeach with its evidence
GradeEvidenceCriteriaTensileT maxMatch
Polyamide 6, 30% GFAMP-4412Strong3 of 316515094 %
PBT, 20% GFAMP-3118Strong3 of 311014088 %
PC / ABS, FRAMP-2207Limited2 of 362125AI86 %

Two clear it outright. AMP-4412 has the most headroom on both. On AMP-2207 the 125 °C is model-generated, not measured, so I would not commit to it without a lab check.

Every value carries its status. Predictions stay marked AI.

Ask a follow-up, or tell it which one to take to the lab
Library200,000+supplier grades, indexed from technical and safety datasheets.
Property space500+properties per grade, each value marked measured or predicted.
InterfaceMCPone server that Claude and any other MCP client can call. In beta.
ResearchNeurIPSthe method behind it, accepted at the AI4Mat workshop, NeurIPS 2026.
How it works

A materials library your agent can call, with the evidence attached.

Grade search, property lookup, substitute ranking and nearest-neighbour queries, all behind one MCP server. Every result comes back structured, so your agent can reason over it and cite it.

One callMCP server · beta, in development
Asks for
A grade to replace, a grade to find neighbours of, or the properties of a named grade
Constraints
Property floors and ceilings, restricted-substance lists, end use and industry, all in the same call
Returns
Real supplier grades, ranked by criteria met and then by similarity
Per candidate
Its supplier, the criteria it meets and misses, and the trade-off stated
Per value
Marked measured or predicted, every time
Called from
Claude, or any other MCP client
01

One tool call returns real grades your agent can cite

Industrial chemistry is not in any general model's training set. Ask an agent for a material and it guesses: no library behind it, no values it can cite, and a different answer each run. Matterialis exposes grade search, property lookup and substitute ranking over an MCP server. Claude and any other MCP client call it like any other tool, and the results come back structured: each candidate with its supplier, the criteria it meets, and the status of every value.

Your agent stops guessing about materials and starts looking them up.
SearchResults 10Library 200,000+
Under-bonnet bracket
Property criteriaReset
Search 500+ properties
Tensile strength
≥60 MPa
Max service temp.
≥120 °C
Water vapour trans.
≤5.0 g/m²·d
Compliance
RoHS · REACH
Ranked resultsRanked against your criteria
RankMaterialEvidenceCriteria metTensileT maxWVTRMatch
1Polyamide 6, 30% glass fibreAMP-4412 · 25038-54-4Strong3 of 31651503.2AI94 %
2Polyamide 66, 35% glass fibreAMP-4680 · 32131-17-2Strong3 of 31801603.691 %
3Polybutylene terephthalate, 20% GFAMP-3118 · 24968-12-5Strong3 of 31101402.488 %
4Polyphenylene sulphide, 40% GFAMP-5902 · 26125-40-6Strong3 of 31902200.983 %
5Copolyester, high-clarity extrusionAMP-1355 · 25038-59-9Strong3 of 3721201.877 %
6Polycarbonate / ABS, flame-retardantAMP-2207 · 25037-45-0Limited2 of 3621256.8AI86 %
7Polypropylene TPO impact copolymerAMP-0914 · 9010-79-1Limited1 of 3281051.174 %
8High-density polyethylene, injectionAMP-0271 · 9002-88-4Strong1 of 331950.468 %
9Thermoplastic polyurethane, aliphatic 95AAMP-7741 · 9009-54-5Limited0 of 34811012.0AI79 %
10Polylactide / PBAT compoundAMP-6620 · 9051-89-2Weak0 of 3456522.0AI71 %
GREEN MEETS THE CRITERION · RED MISSES IT · AI MODEL-GENERATED, NOT MEASURED
02

Constraints in, a ranked shortlist out

A real materials task carries several constraints at once: property floors, restricted-substance lists, end use and industry. An agent that searches a catalogue one keyword at a time cannot hold them together. Your agent passes them all in one call. Candidates come back ranked by criteria met and then by similarity, with the trade-off stated, so the agent reasons over a short, structured list. Filtering by region is in development.

A shortlist your agent can act on, in one round trip.
Swipe the panel to explore
SearchResults 10Library 200,000+
Under-bonnet bracket
Ranked results / Polybutylene terephthalate, 20% glass fibre

Polybutylene terephthalate, 20% glass fibre

AMP-3118 · CAS 24968-12-5 · Chang Chun
88%
Match
Strong evidence Polyester (PBT) 6 AI-generated values

Lower moisture pickup than the incumbent and a tighter shrinkage window, but density sits above the stated ceiling.

Properties · mechanical
Tensile strength110 MPaScraped
Flexural modulus5940 MPaAI-generated
Notched impact, 23 °C8.6 kJ/m²Scraped
Elongation at break2.9 %Scraped
Thermal
Heat deflection, 1.8 MPa140 °CScraped
Glass transition48 °CAI-generated
Barrier & transport
Water vapour transmission rate2.4 g/m²·dayAI-generated
Moisture absorption0.28 %AI-generated
03

Every value carries its provenance, so your agent knows what to measure

The property that decides the experiment is often missing from the datasheet. An agent that cannot tell a measured value from a guess either stalls or commits to a number nobody stands behind. Matterialis predicts the unpublished properties from everything the sheet does say, and marks each one as predicted beside the measured values. Your agent gets both, labelled, and can route the uncertain ones to the lab.

It narrows the list to the experiments worth running.
Swipe the panel to explore
NeighbourhoodResults 12Library 200,000+
Sector filter · off
Search / Nearest materials by behaviour

Sorbitan monostearate, technical grade

AMP-8802 · CAS 1338-41-6 · query material
12
Near neighbours
Nonionic surfactantHLB 4.7Food & beverage

Five of the eight nearest grades are sold into a different industry. Distance from the centre is similarity, so a hollow dot close in is a cross-sector match worth reading.

Your sector Another sector Query grade RINGS ARE SIMILARITY, NOT DISTANCE IN ANY ONE PROPERTY
04

Nearest materials by behaviour, from any industry

An agent exploring a design space needs to ask what else behaves like this grade. Catalogues are organised by supplier and sector, so nothing indexes that question. Our own encoders place every grade in a space of what it is and how it behaves, and the nearest neighbours of a query grade come back ranked by similarity wherever they are sold: a food-grade ester for a cosmetics problem, a packaging polymer for a membrane.

Options a keyword search would never return.
Swipe the panel to explore

Where it fits

You own the agent and the lab. We supply the materials layer.

What you build

The agent, the workflow, the hardware

Planning, design of experiments, lab orchestration and the user's interface. Everything the scientist sees and everything that touches the bench stays your product.

What Matterialis supplies

Real, purchasable materials as a tool

Grade search, property lookup, substitute ranking and nearest-neighbour queries over 200,000+ supplier grades, with the evidence attached. One MCP server, which your agent calls when a task touches a real material.

What stays yours

Your users' data and results

Standard runs on the public index alone, and nothing of your users' is required or stored. Enterprise adds their own data in an isolated workspace that is never used to train shared models.

Trust

What stands behind each value your agent receives.

Research

Accepted at AI4Mat, NeurIPS 2026.

Our work on how to represent a material is accepted at the AI4Mat workshop at NeurIPS 2026. The paper will be published shortly, so your engineers can read how the method works before they wire it in.

Team

IBM Research and the University of Cambridge.

A team from IBM Research and the University of Cambridge, inside Amphico: machine learning and materials science in one room.

Origin

Built for our own product first.

Amphico makes PFAS-free waterproof membranes. Matterialis exists because we had a reformulation problem the available tools could not help with. We still use it for that.

Your data

Your users upload nothing.

Standard runs on the public index alone, and none of your users' data is stored. Enterprise adds their own data, in an isolated workspace never used to train shared models.


FAQ

The questions platform teams ask.

Which clients can call it?

Claude, and any other client that speaks the Model Context Protocol. The screens on this page show Claude Cowork calling it; an agent you built yourself calls it the same way, provided it speaks MCP. An MCP server is the only interface we offer today.

How does access work?

Each beta partner gets its own access, issued when we set you up, and we agree on the technical call how your agent holds it. The tools look things up and return results. None of them changes anything on your side.

What latency and call volume should we expect?

We do not publish figures during the beta. Tell us the workload, roughly how many calls a task makes and how many tasks run a day, and we will size your access to it and tell you plainly what to expect.

Where do our users' queries go, and what is kept?

On Standard, the server needs only the query your agent sends. Nothing of your users' is required, and none of their data is stored. On Enterprise, their own data sits in an isolated workspace that is never used to train shared models. If your data has to stay in a particular jurisdiction, raise it on the technical call and we will tell you directly whether we can meet it.

What will it not do?

It will not run experiments or drive lab hardware; that stays in your product. It works at grade level, so it does not predict how a finished formulation performs, and it does not resolve interaction order. Filtering by region is in development and is not yet something your agent can ask for.

It also never passes a prediction off as a measurement. Every predicted value arrives marked as predicted, so your agent can decide what to send to the lab.

How is it priced?

There are two subscriptions, Standard and Enterprise, and they differ on one thing. Standard runs on the public index alone. Enterprise adds your users' own data, in an isolated workspace never used to train shared models. During the beta we agree terms with each partner platform directly.

Next step

Plug it into one workflow.

Tell us which of your agent's tasks touches a real material. We set you up on the MCP server beta and you run it on your own prompts. Three ways platform teams start:

01

Beta access

Email us the task. We connect your agent to the MCP server and you test it on your own prompts.

02

A technical call

Thirty minutes with the team that built it: what each tool takes, what it returns, and how your agent should use it.

03

The waitlist

Not ready to wire it in yet? Leave your work email and we will tell you when you can get in.