Who We Are
Univa is a Malaysian owned software company that designs, builds and operates AI driven production platforms. Founded and led by Muhamad Ahnaf Thaqeef, Founder and Managing Director, Univa has been operating for [AHNAF: confirm operating-since year], generating early revenue with repeat customers, and is registered as [AHNAF: SSM legal name and number]. The team is small by design: fewer than five full time staff, each shipping end to end production systems rather than slideware.
We are not a consultancy that hands over a plan and walks away. Every product in our portfolio is built, tested and, in most cases, already running with real users and, where payments are involved, real transactions.
What We Build
Univa works across four capability areas.
- Web and mobile production platforms. Full stack applications built and operated end to end, from the first line of code through live deployment and ongoing maintenance.
- Payments infrastructure. Integrated payment rails, including live DuitNow QR processing, built into production platforms that handle real transactions today.
- Business operations and AI document tooling. AI assisted document intelligence, e-Invoice compliance formatting, and operations suites that automate the paperwork Malaysian SMEs still handle by hand.
- AI training and enablement. Corporate AI training and masterclasses delivered directly to Malaysian teams, building practical AI literacy alongside the tools we ship.
How We Work
Univa ships production systems, not prototypes. Every platform we build starts with deterministic engineering discipline before any AI layer touches it: rule based cores for anything financial, unit-tested and auditable, ahead of automated test suites and a security review before launch. Where we do use AI, for document parsing, conversational coaching or content generation, it runs behind guardrails: prompt injection defense, output validation, and human in the loop oversight for every consequential decision. We treat this discipline as a product requirement, not overhead, because the businesses and institutions we work with cannot afford AI that hallucinates.
Flagship Product: Keystone
Keystone is a financing recovery engine for Malaysian home buyers. When a bank declines a buyer's financing application, Keystone does not let the sale die at that "no": the developer invites the rejected buyer back, the buyer consents and uploads supporting documents, and the engine re-underwrites the file and routes it up a three-rung ladder, lightest intervention first. Rung one screens the file against SJKP, the national housing loan guarantee, and assembles a bankability dossier for a participating bank. Rung two offers a rent-to-own bridge on completed unsold stock for buyers a guarantee cannot catch. Rung three coaches overcommitted or arrears-affected buyers back to readiness over WhatsApp, in Bahasa Malaysia first, until they re-enter the ladder without re-uploading a single document.
Every DSR calculation, matching decision, ledger schedule, readiness score and graduation gate runs on deterministic, unit-tested TypeScript, never on an AI model doing arithmetic. The AI lane is confined to document parsing, forced tool use, and every extracted figure is cross-checked against the deterministic assessment before it can appear in a dossier. The platform runs prompt-injection defense on all document input, output scrubbing, PDPA consent-first data handling, and a full audit trail per buyer, with a human-in-the-loop operator console for every consequential action.
Keystone is live and public today, not a slide. The product runs at a working demo URL with seeded personas covering all three rungs, an operator console, and a WhatsApp concierge simulator, backed by 722 automated tests passing, clean type checking and linting, and a security checklist cleared in full.
Why Univa Fits S P Setia
AI native with a deterministic core. Every product we build starts with deterministic logic and adds AI only where it earns its place. For a developer's financing and sales pipeline, that discipline matters more than a demo that looks impressive but cannot be audited: no number reaches a bank dossier that was not first produced by the deterministic engine.
Built for Malaysian home financing, SJKP-aware. Keystone is engineered around SJKP, the national housing loan guarantee, and its own eligibility rules, including its treatment of gig and platform income, so it recognises buyers a standard bank haircut would otherwise reject. The buyer journey and the operator console run in English by default with a Bahasa Malaysia toggle, and the WhatsApp concierge coaches in Bahasa Malaysia first, because the buyers who most need a second chance are not always most comfortable in English.
A working product, not a proposal. Keystone is built, tested and running today, demonstrated end to end on real seeded scenarios rather than described on a slide. We are ready to bring a working platform into a pilot from day one of the programme.
Zero balance-sheet design for the developer. Rung one recovers a sale with no Setia capital involved: the buyer is approved by a participating bank on the standard guarantee. Where a rent-to-own bridge is used, capital sits in a dedicated special purpose vehicle, not on Setia's balance sheet, so the pipeline recovery does not require the developer to carry new financing risk.
Team
Muhamad Ahnaf Thaqeef, Founder and Managing Director. [AHNAF: add a one to two line bio here, background and what led to building Univa. No mention of any employer or salary.]
[AHNAF: confirm listing Shafiq as second team member, role and one-line bio. A two person team reads stronger to reviewers than a solo founder listing.]
Contact
Muhamad Ahnaf Thaqeef, Founder and Managing Director
Email: ahnaf@univa.my
Website: univa.my
Phone: +6017-474-7960