Kuala Lumpur is where Malaysia is building its startup and AI policy infrastructure — the inaugural KL20 Summit in 2024 set a public target of pushing Malaysia into the world's top-20 startup ecosystems by 2030, a second summit follows in June 2026, and the National AI Office has been steering strategy since its 2024 launch. But the AI opportunity that matters most in KL isn't a startup pitch — it's the enterprises and government agencies already running the approval chains, procurement cycles, and compliance obligations that keep the country moving.
That's the problem IRIS was built for: AI orchestration for workflows where a missed approval or a misrouted document has real cost — not a chatbot bolted onto a website. Our focus here is Kuala Lumpur itself, where the enterprise and public-sector buyers we serve are concentrated.
For enterprise and GLC buyers, the practical read is that AI procurement is increasingly expected, not experimental, at the policy level.
Three products, one shared discipline: map the real approval chain before proposing any automation, not the other way around. IRIS exists because a missed procurement deadline or a misrouted contract has a real cost attached to it — not a hypothetical UX complaint. JomWork exists because the person who knows how your organisation actually works tends to leave, and their knowledge leaves with them unless it's searchable. The document processing work extends the same discipline to the specific parsing problem Gov/GLC tenders create.
Live in production, built around Malaysian public-sector procurement patterns — audit-ready logging, BM+English processing, human sign-off checkpoints modeled on real ministry and GLC approval chains. Deployable on-premise or in Malaysian cloud regions. See IRIS →
RAG-based company knowledge, built for GLCs and enterprise teams where institutional knowledge is scattered and turnover creates costly onboarding delays. See JomWork →
Tender and compliance document automation — a real project build for Gov/GLC document parsing. Read the project log →
Why Teh Tarik Digital, not a generic enterprise vendor: most enterprise AI vendors sell a platform and leave the workflow redesign to you. IRIS was built the other way around — mapping the approval chain first, then fitting AI orchestration into where it actually breaks. Human sign-off stays in the loop at every stage that matters; IRIS surfaces information, it doesn't make the call.
In procurement and approvals, the cost of a miss isn't hypothetical — it's a missed deadline, a compliance gap, a document that sat in someone's inbox for three weeks while a vendor waited on a signature. The left column is what changes for the people running that process day to day. The right column is what's actually live and measured — not a projection, not a demo environment, a platform handling real procurement requests today.
We map before we build — the approval chain comes first, the tooling second. Most enterprise AI vendors do this backwards: they sell you a platform, then leave you to figure out where it fits your actual process. That ordering is why so many enterprise AI projects stall after the pilot — the tooling was never scoped to where the workflow actually breaks.
Map your approval chain, procurement cycle, or knowledge-access bottlenecks — the mapping happens before we talk tooling, because a workflow diagram tells us more than a feature list ever could.
Scope IRIS/JomWork to your specific workflow, with human sign-off points defined upfront — decided before a line of integration code gets written, not negotiated after the fact.
Connects to existing systems, BM + English document processing where needed — your approval chain doesn't change shape to fit our tooling, our tooling fits the chain you already have.
Live orchestration with audit-ready logging — and the team that mapped your approval chain in stage one is the same team you call when something needs adjusting later.
For GLC and government procurement specifically, the standard due-diligence questions apply here too — and we'd encourage asking them of any vendor, not just us. Most of the AI vendors GLC procurement teams meet with can describe a category of service; fewer can name the actual product, walk through where the data lives, or point to an operating history that predates the current AI hiring boom. That gap is usually where the real risk sits.
Teh Tarik Digital is a registered Malaysian entity with a decade of operating history, one accountable team from scoping through ongoing support, and a founding background in regulated banking and government technology work that shaped how IRIS handles approval chains and audit trails from day one — not a feature added after the fact.
IRIS's document-processing core, tested against 977 real government tenders.
Format detection, JSON schema output, and the ministry-name normalisation fix that took accuracy from 71% to 94%. Full case study →
The questions procurement and compliance teams actually ask before they'll take a second meeting.