One operating system.For textile and apparel factories.

AI that spots fabric and garment defects, and confirms each fix.Quality, traceability and maintenance, all on one record.

Validated and supported by

The defect-to-fix loop

See it. Solve it.
Know it’s fixed.

Finding a fault matters. Knowing it was fixed matters just as much. We’re building YarnOS to connect everything in between.

See

Decide

Alert

ConfirmThe key step

Learn

Reading fabricCamera + encoder · Machine 07

01 / 05 See

Every metre tells a story.

A camera and encoder read the fabric as it is made. Each frame has a position, so an issue can always be traced back to its metre.

A scripted illustration of the proposed workflow.

Alert-only first. Your factory decides if, and when, a machine can be stopped.

One record. Across the floor.

A whole factory.
One connected story.

Start with quality. Add modules as your factory is ready. Every part connects to the same record of rolls, machines and events.

01 / Quality · Planned first build

An alert is a starting point.

Inspect fabric as it is made and measure finished garments against the buyer’s spec, even 3D pieces like contour bras. Every defect stays linked to its fix and the proof that it held.

Fabric inspection & buyer grading
Garment measurement, flat and 3D
Cause capture & camera-verified fixes

Concept illustration

Room to growPlanned for later

Production & stock

Planning links

Group view

Connect first. Finance, payroll and tax can stay in the ERP you already run.

Made for Sri Lankan floors

Local by nature.
Made for your floor.

Made to fit the floor you already have, in the languages your people speak.

01 / A familiar voice

Good work starts
with understanding.

Spoken alerts in English, Sinhala and Tamil. With the image, machine and metre attached, every message arrives in context.

ආයුබෝවන්

02 / A natural fit

New intelligence.
Familiar systems.

CSV and API first. Alongside the ERP and machines you already run.

ERPMachinesCSV

YarnOSOpen API

03 / At your pace

Start with alerts.
Stay in control.

Nothing stops a machine until your factory chooses to trust it.

What YarnOS may do

Send spoken alertsFrom the first pilot

Draft work ordersA person approvesAsks first

Built here. Ready to work alongside you.

Let’s talk

AI that does the everyday work

Let the work flow.
Leave the busywork.

We’re designing AI workflows for the routine work around quality. You help us decide what matters first.

Planned, not built. We’ll start with the ones factories ask for first.

AI drafts. A person approves. Every action is logged.

Work order WO-0187Example

Replace needle, Machine 07

Drafted by YarnOSFrom the needle-line alert at metre 41214:21
Routed to the mechanicLine 2 · Machine 0714:24
Waiting for the shift supervisorNothing goes ahead until then

Your floor. Your rules.

Your factory.
Your ground rules.

Local-first by design. People in control, from day one.

Sandakada pahana, the moonstone carved at a threshold

On your network. On your terms.

Designed to work offline, with factory data kept on site.

Dumbara weaving, lozenges built from woven steps

Open from the start.

CSV import and export, plus an open API. Your records remain yours.

Nil manel, Sri Lanka’s blue water lily

Built around people.

No worker tracking or scoring. People review alerts, approve actions and set the rules.

Works offline
Data kept on site
CSV and open API
No worker scoring
People approve actions
Every action logged

Proof before promises

Earn trust.
One machine at a time.

Research and first concepts are complete. We’ve visited MAS Unichela and put our questions to technical people in the industry. Now we’re validating the idea while we build and test a first prototype on public fabric datasets, aiming for deployable pilots by April 2027.

Done

Listen

Visited MAS Unichela and asked technical people in the industry how the floor really works.

Coming up

Visit

A fabric mill visit, to see where defects happen as the fabric is made.

Under way

Validate & Build

Confirm with the industry that the problem is real, while we build and test a first prototype on public fabric datasets.

The go / no-go gate

After the gate

Test

Only then: agree a footage source and test the idea on real footage.

By April 2027

Pilot

One site, one machine, alert-only, reviewed together.

Pilot target≤1false alert per 100 metres

Measured on real footage before any pilot.

To start1site, one machine

Small enough to review together, closely.

Machine stops0until you choose

Alert-only first. Your factory decides when to trust it.

Our plan to April 2027. The details will shift as we learn from factories.

The people building YarnOS

Built by Opti5 Labs.
Grounded in Colombo.

A team working across engineering, design and operations. Close enough to visit your factory. Curious enough to listen first.

Chanitha Abeygunawardena

Product direction & engineering

Ilakian Puvanendra

Core engineering & integrations

Sasiru Virajith

Design, engineering & operations

Rajindu Amithirigala

Project delivery, design & engineering

Dinith Ranasinghe

Marketing & go-to-market

Developed through the Agile Team ProjectInformatics Institute of Technology / University of Westminster

Let’s start with a conversation

The next chapter of
your factory starts here.

Tell us where this is wrong.
Help us build something that belongs on your floor.

yarnos@opti5labs.com

Colombo, Sri Lanka

A small first step

A 30-minute conversation. Optionally, one to two hours of footage from one machine, under an NDA.

Something back

A research summary, first sight of results and priority for an alert-only pilot. No obligation.

One operating system.For textile and apparel factories.

Be first on the floor.
Join the YarnOS waitlist.

Thirty minutes.
On your schedule.