Tursi | Trusted Supply for Autonomous Science

01 / The problem

Autonomous labs run at machine speed. Their supply chain doesn't.

AI can propose thousands of experiments a day. But materials are still bought through catalogs, purchase orders and PDFs designed for humans. So today's self-driving labs quietly shrink their search to whatever is already on the shelf.

/01

Built for human buyers

No public ordering APIs. Contract pricing behind logins. Pack sizes made for a bench chemist, not a robot dispensing milligrams.

/02

Quality you can't compute

Lot-to-lot variation silently corrupts results. Certificates of analysis live in PDFs, invisible to the models that need them most.

/03

Humans in the loop

Receiving, weighing, labeling and restocking are still manual, and every hour a robot waits for a person is an hour of science lost.

02 / Platform

One supply layer. From agent request to robot deck.

Tursi is the materials backbone for AI-run labs: machine-callable, quality-verified, and secure by design.

Agent-native ordering

A clean API and MCP server your agents and schedulers call directly. Live availability, structured specs, and orders triggered by your own inventory signals.

REST API
MCP
Webhooks
Inventory sync

Robot-ready formats

Pre-weighed, pre-dissolved and barcoded. Packed in the vials, racks and SBS-format plates your platform already handles. Delivered ready to load.

Pre-weighed
Barcoded
SBS plates
Deck-ready

Verified lots

Every lot independently characterized. Purity, impurities and test results delivered as signed, machine-readable data your optimizer can actually use.

XRD
ICP-MS
Machine-readable CoA
Lot lineage

Trust & compliance layer

When machines buy matter, every order must be authorized. Policy rules, compliance screening, human approval for flagged items, and a tamper-evident audit trail.

Policy engine
KYC
Screening
Audit log

03 / How it works

Closed-loop supply. No humans required.

Supply becomes part of the experiment loop, not a ticket in someone's inbox.

Step 01

Request

Your agent asks for materials via API or MCP.

Step 02

Authorize

Policies and compliance checks run instantly.

Step 03

Source

Best qualified lot sourced, or made to order.

Step 04

Verify

Characterized and signed before it ships.

Step 05

Deliver

Robot-ready formats, straight to the deck.

Step 06

Learn

Lot data flows back into your model.

Developer preview

Your lab's supply chain is one call away.

Order like you'd call any other tool. Get back a tracked, verified shipment and the data to go with it.

order.pyIllustrative

from tursi import Tursi

client = Tursi(api_key="...")

order = client.orders.create(
    items=[
        {"material": "Li2CO3", "purity": ">=99.99%",
         "mass_mg": 250, "aliquots": 96},
        {"material": "NiO", "purity": ">=99.9%",
         "mass_mg": 500, "aliquots": 48},
    ],
    format="barcoded-vial-rack",
    require=["xrd", "icp-ms"],
    deliver_to="lab-sf-01/deck-3",
)

# signed, machine-readable lot data
for lot in order.lots():
    model.ingest(lot.certificate)

04 / Materials

Starting with the building blocks of materials discovery.

High-purity inorganic precursors for self-driving materials labs: the inputs AI-designed batteries, catalysts and semiconductors are made from.

MO_x

Oxides

Transition-metal and rare-earth oxides for solid-state synthesis.

CO₃

Carbonates

Alkali and alkaline-earth carbonates, battery-grade purity.

NO₃

Salts & nitrates

Soluble precursors, pre-dissolved to spec for liquid handling.

M⁰

Metal powders

Elemental powders with controlled particle size and inert packing.

05 / Vision

Soon, rooms full of robots will invent new medicines, materials and machines around the clock. Someone has to supply them.

Early access

Building an autonomous lab?

We're working with a small number of self-driving labs on early pilots. Tell us what you're building and what you can't get.

hello@tursi.ai →