Matsi / Formulation Intelligence Formulation Intelligence A formulation is a recipe for a material: which polymers, which fillers, in what proportions. Getting one right normally means months of trial-and-error in a lab. Matsi searches that space computationally: it proposes candidate recipes and predicts how each will behave before anyone makes it.

Intelligence for the new materials economy

Intelligent discovery and design of material formulations that meet performance, cost, compliance and supply chain resilience constraints simultaneously.

1.64M formulations evaluated · 444 material–property models · 36,917 measurements measured 2026-08-21

Every number below comes from a file with a date on it. We don't estimate. These numbers 36,917 measurements are structured values pulled from the literature we have processed, not estimates. Of 444 material–property models, 178 are tuned against those measurements; the rest run on published values until a measurement tests them. A single search evaluates candidate recipes and predicts a full property profile for each, plus cost and carbon impact.

Corpus · 2026-08-22
polymers modelled24
individual ingredients modelled103
— of a catalogue of117
ingredient types27
properties predicted per material19
material–property models444
— backed by measured data178
research documents held13,545
— read and mined10,692
measurements collected36,917
— tied to a known material27,726
formulations evaluated1.64M
recipes live on the dashboard9,394
— carrying model version and evidence100%
Batch run · stf-run-20260528T033824
candidates evaluated150
properties per candidate16
plus cost & carbon2
batch wall clock (n=8)43.5 min
throughput, parallelised3.4 / min

Run stf-run-20260528T033824 · figures measured 2026-08-21

Matsi answers here, and your brief reaches the team at the same time. We train on our own corpus, not on customer data.

Performance The four requirements A candidate formulation has to clear all four of these, not score well on average. They are equal quarters because they are pass/fail, not weighted priorities. A recipe that comes in cheap but fails the performance test is no use to anyone. 01
Profitability 02
Resilience 03
Sustainability 04

§ 02 · Is this you?

You need recycled content without losing performance.

Every trip through recycling shortens the polymer chains, and past a point the material just won't hold the spec any more. What you really want to know is how much recyclate you can carry, and what you'd have to change to carry it.

How the search works →

You have a compliance deadline.

EU PPWR, California SB 54, a compostability certification, a food-contact rule. The deadline won't move, so the material has to, and it has to do it on the equipment you already own.

What the constraints look like →

You need cost parity with a bio-based alternative.

Matching the petrochemical incumbent on performance is one problem. Beating PLA, PBAT and PHA on cost is a completely different one. Both have to land, or nobody switches.

Bring us the numbers →

§ 03 · What a candidate has to clear The constraint column The four rows are what we optimise for. The right-hand column turns each one into something you could actually test. “Performance” is a word; perf ≥ perf_legacy is a check: the new material must be at least as strong as the one you use today. Note the two benchmarks differ on purpose: performance is measured against the oil-based incumbent, cost against the bio-based alternatives.

Four requirements. A formulation has to clear all four, or we don't call it a result.

PillarClaim Constraint
01 Performance Ensure finished product performance delivers on customers needs better than that of legacy product material formulations. perf ≥ perf_legacy
02 Profitability Deliver stronger value proposition on materials costs with 10X the productivity of R&D resources’ conventional approach. Δcost < 0 vs PLA · PBAT · PHA
03 Resilience Navigate market volatility, trade policy implications and supply chain risks and trade-offs with transparency and agility. risk → bounded
04 Sustainability Leverage existing resources, more intelligently designed, to ensure compliance and move towards greater circularity without compromise on functional and business needs. bio ↑ · compliance = pass

§ 04 · Specimens · Plate 01–07 Specimen plates Seven everyday plastic objects. For each we show what it is really made of and the rule it has to satisfy: a food-contact regulation, a flammability rating, a compostability standard. These are the constraints a formulation has to clear before it can replace anything. Where we do not yet have a sourced constraint, the plate says so rather than guessing.

Each object is drawn with two inks. Where they cross, a third colour appears that neither ink contained. Formulation works the same way — combine inputs, and you get properties none of them had on their own.

Why rigid container is hard Every pass through recycling shortens PET's polymer chains, which drops intrinsic viscosity. Below a threshold the material will no longer stretch-blow-mould into a bottle that holds its shape under top load and internal pressure. Recycled content and mechanical performance therefore pull directly against each other, and food-contact rules limit what you are allowed to add to compensate. The closure is a different polymer again, which complicates the stream the whole part eventually enters.
Plate 01 · Rigid container
rPET · PP closure
Constraint
IV ≥ 0.78 dL/g · food contact
EU 10/2011
Why produce packaging is hard Barrier and compostability are close to opposites. A film open enough to break down in an industrial composter is also open enough to let oxygen through, which is what shortens shelf life. PLA contributes stiffness and clarity but is brittle; PBAT contributes toughness but is soft. The blend ratio trades one against the other, and the processing window for the blend is narrower than for either component on its own.
Plate 02 · Produce packaging
PBAT/PLA blend film
Constraint
OTR ≤ 1000 cc/m²/day · compostable
EN 13432
Why injection moulded is hard The part has to survive years in a drawer and then break down in months in a composter: a contradiction the material itself has to hold. PHA degrades thermally close to its own melt temperature, so the processing window is tight, and it crystallises slowly, which lengthens the moulding cycle. Cycle time is cost, and a pen is a commodity with almost no cost headroom.
Plate 03 · Injection moulded
PHA barrel · PHA cap · brass tip
Constraint
MFI 15 g/10 min · compostable
EN 13432
More specimens e-reader · food packaging · diaper · dunnage
Why durable housing is hard Flame retardancy without halogens generally means loading the polymer with phosphorus compounds or mineral fillers. Those loadings embrittle the part, which is a direct problem for a housing whose job is surviving being dropped. The assembly is also bonded to glass, so the polymer has to hold its dimensions across a temperature range or the screen debonds.
Plate 04 · Durable housing
PC/ABS shell · TPU bumper · PMMA lens
Constraint
UL94 V-0 · halogen-free
IEC 61249-2-21
Why food packaging is hard A tray recycles cleanly only if it is essentially one polymer. But a single polymer rarely provides the oxygen barrier a food shelf life requires, so the conventional answer is a thin EVOH layer - which above a few percent is precisely what disqualifies the tray from its own recycling stream. The problem is to obtain barrier from something that does not break the recyclability you were trying to protect.
Plate 05 · Food packaging
Mono-PET rigid tray · PET lidding film
Constraint
EVOH barrier ≤ 5 % w/w · food contact
RecyClass DfR · EU 10/2011
Why absorbent core is hard Multi-layer, multi-polymer, single-use, and contaminated at end of life, close to the hardest case in packaging. Absorbency is measured as centrifuge retention capacity - grams of saline held per gram of polymer after spinning. That capacity comes from acrylic-based superabsorbents, and bio-derived alternatives hold substantially less per gram. So raising bio-content works directly against the one number the product is bought on. The nonwoven layer carries its own fibre requirements that most candidate polymers cannot meet.
Plate 06 · Absorbent core
PP nonwoven · SAP · LDPE backsheet
Constraint
CRC ≥ 30 g/g · saline retention
NWSP 241.2 / EDANA
Why structural dunnage is hard Protective foam has to be light, carry a compressive load at 10% strain, and recover so it can absorb a second impact. Those pull against each other: strength usually comes from density, and density is shipping cost. Bio-based foams also tend to crush once and stay crushed. EPS is both very good and very cheap at this, which sets a hard bar: a replacement has to match resilience and density at a price that survives a shipping budget.
Plate 07 · Structural dunnage
PLA foam · EPS alternative
Constraint
σ₁₀ ≥ 100 kPa @ 10% strain · density ≤ 25 kg/m³
ISO 844 / ASTM D1621