Matterialis for suppliers — your datasheets, answering buyers on your website

What it is

Structured data and an AI layer over your own product catalogue.

We read the technical and safety datasheets you already publish and turn them into a structured database you own. On top of it sits an assistant that answers buyers on your website, qualifies them, and shows you what the market asked for. It sells what you sell and nothing else.

Your range 1,240Requests 412Install

Northfield Specialty Polymers1,240 datasheets · structured in 2 h 48 min

Property criteriaReset

Search 500+ properties

Viscosity

≤30 000 mPa·s

Glass transition

≥100 °C

Die shear strength

≥8 kg-f

Claims

Halogen-free

Your grades, rankedEvery row is a Northfield datasheet

Rank	Grade	Evidence	Criteria met	Viscosity	Tg	Die shear	Halogen-free
1	NF-6210 high-Tg die attachTDS rev. 4 · 2026-03	Strong	4 of 4	24 000	135	11.2	Claimed
2	NF-6202 B-stageable die attachTDS rev. 7 · 2025-11	Strong	3 of 4	28 000	40	10.4	Claimed
3	NF-4180 silver epoxyTDS rev. 2 · 2024-08	Limited	2 of 4	18 500	110^AI	—	—
4	NF-7110 low-stress underfillTDS rev. 5 · 2026-01	Strong	2 of 4	9 800	85	—	Claimed
5	NF-5500 potting compoundTDS rev. 3 · 2025-06	Limited	1 of 4	6 200	62^AI	—	Not claimed
6	NF-9020 thermal gap fillerTDS rev. 1 · 2025-09	Weak	0 of 4	41 000	—	—	—

GREEN MEETS THE CRITERION · RED MISSES IT · — NOT STATED ON THE SHEET · ^AI ESTIMATED, NOT MEASURED

01

Your catalogue, structured in seconds per datasheet

Every PDF you publish goes through the pipeline: each property with its unit, test method and conditions, each classification, each claim. A datasheet takes seconds. A range of a thousand grades is structured the same day and reviewed by our materials scientists. What comes out is a database you can search by requirement and connect to anything, including your own AI.

A searchable, AI-ready catalogue, from documents you already publish.

Swipe the panel to explore

northfield-polymers.com/adhesives · the assistant, open

Northfield Specialty Polymers

Product assistant

New ×

We use a competitor's silver-filled epoxy today, datasheet attached. Need an equivalent with moisture absorption under 0.5 wt%.incumbent-silver-epoxy-TDS.pdf

Reads as: a replacement for an electrically conductive silver epoxy, matching its resistivity and cure, with a moisture-absorption ceiling of 0.5 wt%.

NF-4180 silver epoxy

Electrically conductive · heat cure

2 met0 missed0 unclear1 not stated

Volume resistivity: 2 × 10⁻⁴ Ω·cmmet
Cure schedule: 150 °C · 30 minmet
Moisture absorption: —not stated

Meets volume resistivity and cure schedule. The datasheet does not state moisture absorption, so that requirement is reported as not stated.

NF-4180 matches the incumbent on conductivity and cure. Your 0.5 wt% ceiling cannot be confirmed from the sheet; Northfield can supply the measured value. Shall I pass this to the applications team?

Hand-off Your requirements, the attached sheet and this result go to Northfield's sales team with your reply.

Yes, please. Also, what is the shelf life?Send

Answers from Northfield's datasheetsTalk to a person

02

A sales engineer on every product page, all day

Buyers describe the application in their own words, or attach the datasheet of the product they use today. The assistant turns that into criteria, searches your range, and shows each requirement as met, missed, unclear or not stated, with the value from your sheet beside it. When the fit is there, it hands your team a qualified lead with the requirements already written down. When the sheet does not say, it says so and offers a person.

First response in seconds, the qualified lead in your inbox, and your team's time kept for the deals.

Swipe the panel to explore

Your range 1,240Requests 412Install

Last 30 daysExport CSV

Conversations

287

412 requests answered

Matched

61%

best product met every requirement

Leads handed over

58

with requirements attached

Unmet demand

23%

no match, no fit, or a product gap

Median first reply

6s

at any hour, in the buyer's language

Requests per day

Answered from the rangeUnmet

Requirement	Asked	Met	Missed	Not stated	Buyers wanted
Glass transition temperature°C	48	19	29	0	≥ 100–150 °C
ViscositymPa·s	41	36	5	0	≤ 15 000–30 000 mPa·s
Moisture absorptionwt%	22	4	6	12	≤ 0.3–0.5 wt%
Thermal conductivityW/m·K	17	11	6	0	≥ 2–5 W/m·K

Unmet demand

what buyers asked for that the range could not offer

Product gap

High-Tg die attach for automotive under-hood, Tg ≥ 150 °C

Glass transition ≥ 150 °C asked 29 times. Best offered: NF-6210 at 135 °C.

No fit

Low-outgassing epoxy for a satellite payload, ASTM E595

No grade in the range has an outgassing figure; buyer handed to a person.

Not documented

Silver epoxy with moisture absorption under 0.5 wt%

NF-4180 leads but the sheet does not state moisture absorption. Twelve buyers asked this month.

Product gap

Gap filler above 5 W/m·K for a power module

Thermal conductivity ≥ 5 W/m·K asked. Best offered: NF-9020 at 3.2 W/m·K.

03

See what the market asks you for, and what you could not offer

Every conversation is yours to read, and the dashboard counts them: requests per day, which properties buyers ask for and at what thresholds, how often your best product met them, and where it missed. Unmet demand is listed apart. A column of misses is a product to develop. A column of not stated is a datasheet to finish, which is cheaper.

Demand intelligence from your own website, counted from verdicts.

Swipe the panel to explore

Add to any page of your site

<script src="https://chat.matterialis.com/v1/embed.js" data-key="nf-8c21…" async></script>

Runs only on the domains you allow · takes your brand colour · changes nothing else on the page

04

Live on your website with one line of code

We structure your datasheets. Your web team adds one script tag to the product pages. The assistant appears in the corner, in your brand colour, isolated from the rest of your page so nothing else changes. It runs only on the domains you allow, so nobody can lift it onto another site.

No platform migration. Your website, your domain, your leads.

Where we differ

A listing shows your product beside everyone else's. This answers on your site.

Most suppliers reach buyers through a marketplace, on its terms, or through a PDF library and a contact form. Matterialis puts a sales engineer on your own website, over your catalogue alone.

Capability

On your site

Your website todayPDF library and a contact form

A marketplace listing

A general chatbot

Buyers ask by requirement, in their own words

Yes

No

Partly

Yes

Each requirement checked against your datasheet

Yes

No

Partly

No

Only your products are shown

Yes

No

Qualified lead handed to your team

Yes

Partly

No

You see every request, and what went unmet

Yes

No

Partly

No

From public product information, October 2026. We compare what each route is built to do. A marketplace listing and Matterialis on your site are not exclusive; many suppliers will want both. Tell us if we have anything wrong.

Trust

What your buyers hear is what your datasheets say.

Grounded

Grounded in your data.

Every answer comes from your datasheets. Each requirement is reported as met, missed, unclear or not stated, with the value from your sheet beside it. No value is invented, and no other supplier's product is ever shown.

Team

Built by materials scientists and chemists.

The team comes from IBM Research and the University of Cambridge, and the method is written up and patent pending. The people reviewing your structured catalogue read datasheets for a living.

IBM Research

University of Cambridge

Your data

Stored apart, owned by you, never used for training.

Your private catalogue lives in its own isolated store, owned and fully controlled by you. You can read, export and delete every conversation. Nothing you share is used to train shared models or shown on anyone else's site.

FAQ

The questions suppliers ask.

Will it ever recommend a competitor's product?

No. Every product the assistant shows, links or counts comes from your catalogue, by construction: the search it runs can only return your grades. If a buyer names a competitor's product or attaches its datasheet, the assistant looks for the products in your range that meet the same requirements and presents those on their own merits.

What do we need to provide?

Your datasheets. For most suppliers that means the technical and safety data sheets already on your website, which we read directly. If you have better data than you publish, such as measured values, approvals or application notes, you can share it and it will be used; nothing you share is used to train shared models. Then one script tag on your product pages, which your web team adds in minutes.

What happens when the datasheet does not answer the question?

The assistant says so. The requirement is shown as not stated, the buyer is told it is a gap in the document, and they are offered a person to ask. You see the question in your dashboard under documentation gaps, which is often the cheapest product improvement available: a value measured once and added to the sheet answers every future buyer who asks.

Can it make things up?

The parts that matter are not generated. Which products are shown, each requirement's verdict and the value beside it are computed from your datasheets, never written by the model. The model turns the buyer's words into criteria and explains the result in prose, and it is held to the values the search returned. It never quotes a price, lead time or approval that is not in your documents. Estimated values from similar products are always marked as estimates.

What do we learn from it?

Every conversation, in full, and the counts across them: which properties buyers ask for and at what thresholds, how often your best product met them, where it missed, which products lead and which get clicked, and the requests that found nothing at all. Product gaps and documentation gaps are listed separately, because one needs development and the other needs a datasheet update.

Does it work for distributors with several manufacturers' lines?

Yes. The catalogue is scoped to the products you sell, whoever makes them, and the assistant presents them as your range. If a line has to be kept out of the public assistant, it is left out of the scope.

Can our sales team use it internally?

Yes. Your sales and technical teams get the same search over your range, and can add internal test data, approvals and application notes to it. That workspace stays apart from the public assistant, so an inbound request gets a referenced answer in minutes and nothing internal reaches a buyer. An assistant that drafts those answers for your reps is the next piece of this product; ask us about timing.

Your datasheets answer your buyers, on your website.

Structured data and an AI layer over your own product catalogue.

Your catalogue, structured in seconds per datasheet

A sales engineer on every product page, all day

Requests per day

Unmet demand

See what the market asks you for, and what you could not offer

Live on your website with one line of code

A listing shows your product beside everyone else's. This answers on your site.

Buyers ask by requirement, in their own words

Each requirement checked against your datasheet

Only your products are shown

Qualified lead handed to your team

You see every request, and what went unmet

What your buyers hear is what your datasheets say.

Grounded in your data.

Built by materials scientists and chemists.

Stored apart, owned by you, never used for training.

The questions suppliers ask.

See it on your own catalogue.

A demo on your range

A pilot on one product line