Tires are the last craft-bound mass-produced safety-critical product
Treadum is an independent startup building original AI-native software for tire and rubber manufacturing. Not a consultancy, not an integrator, not a division of an equipment maker. We build the autonomy layer, and we are judged on whether the plant’s numbers move.
The processes we care about
Why an autonomy layer, and why now
Plants have automation. They do not have autonomy.
Tire plants are full of sophisticated equipment that executes recipes precisely. What they lack is a layer that perceives what is actually happening, decides what should change, and acts inside safe limits — continuously, at every station, with the evidence recorded. That gap is where scrap, energy and escapes live.
- Automation executes a recipe; autonomy defends an outcome
- Sampling finds drift hours late; sensing finds it at the station
- Fixed cure recipes carry a safety pad that costs energy every tire
- Craft knowledge is retiring faster than it is being transferred
How we build
Earn autonomy
Shadow, then assist, then bounded. No plant should hand over a press because a vendor is confident. Gates are measured, signed and reversible.
Evidence over assertion
Every recommendation cites what justified it. Every action is logged immutably. If we cannot show the evidence, we do not show the recommendation.
The engineer is the authority
Disagreement is signal, not friction. Overrides are captured as structured data and reviewed weekly with the process team.
Your IP is yours
Compound and construction data stays in your tenant. No cross-tenant training on your recipes, ever, under any commercial pressure.
A system of action
We do not sell dashboards. If a capability cannot change what the line does or what the plant can prove, it does not ship.
Say the uncomfortable thing early
If a site is not a fit, or a pilot missed its metric, we say so. Long-term credibility in a safety-critical industry is the only asset that compounds.
What compounds over time
Software is copyable. Linked plant data, earned trust and signed envelopes are not.
Linked genealogy is the hard asset
Defect imagery is common. Defect imagery linked to the compound batch, component roll, builder, drum, press, mold, cure trace and final grade is rare — and it is what makes prediction possible rather than merely classification.
Signed envelopes
Every graduated autonomy gate is a relationship a competitor has to re-earn from scratch, station by station.
Captured judgement
Per-builder and per-engineer memory encodes decisions that would otherwise retire with the people who made them.
Corrections improve the model that makes corrections
Every override, hold and outcome is training signal from a plant nobody else can observe. The loop gets tighter the longer it runs, per site.
Depth beats breadth
Connectors into legacy inspection machines and press controls are unglamorous, slow to build and difficult to displace once proven.
Evidence as a by-product
Once audit evidence is produced by the runtime, replacing the runtime means rebuilding the evidence chain.
Where we are going
Sequenced by trust earned, not by feature count.
Wedge agents in production
Defect and inspect, build and splice, cure and mold running from shadow to bounded autonomy at design-partner sites. [ASPIRATIONAL]
Full plant loop
All nine agents under one orchestrator with the as-built twin gating production changes at a reference site.
Group scale
Multi-site fleet management, cross-site model portfolios and group benchmarking with strict isolation controls.
Adjacent scopes
Retreading, technical rubber goods and upstream compound development, using the same loop and the same evidence model.
The operations layer
Every tire mixed, built, cured and graded on Treadum — the autonomous operations layer for the world’s tire and rubber manufacturing.
How we differ from the alternatives
| Alternative | What it does well | Where it stops |
|---|---|---|
| Equipment OEM software | Deep control of its own machines | Locked to one vendor’s assets; no cross-process loop or genealogy |
| Machine-vision point tools | Strong single-modality detection | No fusion, no localization to a station, no control action |
| MES and quality platforms | Records and workflow | Reports what happened; does not change what the line does |
| Generic industrial-AI platforms | Flexible modeling toolkits | No tire-specific process models, envelopes or evidence chain |
| Internal data-science teams | Deep plant knowledge | Rarely resourced for edge runtime, governance and 24/7 operation |
How we work with plants
- Walk first: We spend time on the floor before proposing anything. If the wedge is not obvious after two days, we say so.
- Write it down: Baseline, metric and success criteria are agreed in writing before any work starts.
- Ship weekly: Model reviews, override analysis and gated releases run on a weekly rhythm with your engineers in the room.
- Report honestly: Both sides read the same immutable log. There is no version of the numbers that only we can see.
- Expand on merit: The next cell is sold by the last cell’s results, not by a renewal conversation.
The people running Treadum
A small team in Amsterdam, backed by a $1.5M seed round from 99SA Ventures and working on real tire lines with design partners.
Nadeem leads Treadum’s company, commercial and product direction: which cell goes first, which number it is judged on, and who at the plant signs off on the baseline. He works with tire makers to move mixing, building, curing and inspection onto closed-loop AI, one measured cell at a time.
Rema leads engineering: the factory-edge runtime, the vision, X-ray and shearography models that decide at the building drum, the as-built tire twin, and the bounded write-back into mixer, press and line controls. She is why every agent earns trust in shadow mode first and carries a genealogy record behind every action.
How partners describe working with us
“We had three people watching X-ray images and still shipped uniformity rejects. Treadum flagged the cord shift at the building drum, not four hours later at final inspection.”
“The cure agent found 40 seconds of margin on a construction we had run the same way for eleven years. It proved it in shadow mode before it touched a press.”
“Genealogy is the part I did not expect to care about. Every tire now has a linked record from batch to grade, and audit prep went from weeks to an afternoon.”
[PLACEHOLDER] Design-partner quotes are illustrative until pilot references are published.
Treadum at a glance
Customer counts and pilot results are [ASPIRATIONAL] until design-partner references are published.
Straight answers
No. Treadum builds original product. We integrate deeply with plant systems because that is required to close the loop, but we do not sell integration services or resell anyone’s hardware.
Because the value is cross-process. A loop that only sees one vendor’s machines cannot link a cure escape to a mixing drift, and that link is where the yield is.
The shortage is the constraint, not the surplus. Plants cannot hire builders as fast as they retire. Autonomy targets the rate and consistency gap and moves people toward supervision and exception handling.
By marking aspirational claims plainly, agreeing baselines in writing, and letting design-partner results speak before we do.
Put one cell on autonomy in 90 days
Pick one wedge — splice inspection, X-ray defect detection, cure-state optimization or uniformity prediction. We baseline it, run shadow mode, then graduate to bounded autonomy under a signed control envelope.
Pilots start in shadow mode. No line changes until accuracy and safety gates pass.