Engineering & Products

Software & Platform

Beyond consulting, UTS builds its own software — it doesn't buy it. A complete business platform, a remote accounting platform and AI solutions — all on engines and building blocks whose source we own and maintain ourselves.

What you can run in your business

Mature products working today at real customers in Qatar — each with its own site, trial and pricing.

Jadwal

Jadwal

One engine, professional business software for any industry.

Available

A complete ERP — accounting, inventory, point of sale, invoicing, HR and operations — on one core extended by ready industry packs, with an AI assistant that works inside the system, not beside it.

25modules on the core
8industry packs
5years in production
3deployment modes
  • Double-entry accounting built in — every sale and purchase posts automatically
  • Cloud, desktop or a custom build on the same core
  • Fully Arabic and English, multi-branch, multi-user, role-based
  • An AI assistant that drafts quotes and contracts — a human always approves
Tadween

Tadween

Your accounting, handled remotely — by a specialist team with AI ahead of it.

Available

Qatar's remote accounting platform: photograph your invoices and documents from your phone, AI extracts and classifies them, a qualified accountant reviews and posts them — and your financial statements arrive ready.

8steps in the document cycle
24/7document upload
  • Photograph a document in the mobile app and upload it in seconds
  • Automatic data extraction with a proposed classification and entry
  • Human review and approval before posting — always
  • Financial statements and reports, ready whenever you want them
AI Solutions

AI agents, automation & intelligent documents

Our flagship AI platform — private assistants, WhatsApp & voice agents, automated workflows, and professional document production for Qatar businesses. Explore it on its own dedicated site.

Visit ai.uts.qa

What we build them on

We don't assemble our products from rented parts. This is the engineering layer we own, maintain and mostly publish as open source — and it is why we can build what we build at the speed and cost we do.

AI agent building blocks

The layer under every agent we run: an open standard and published packages that make Arabic AI cheap, fast, auditable and self-hostable.

MSM — Multi Small Models

An open standard that replaces one giant model with a pipeline of small, specialised ones.

Every AI system built on a giant language model pays a hidden tax: a 200-billion-parameter model trained on the accumulated knowledge of civilisation, deployed to answer "I want to order a pizza". MSM breaks the task into small layers, each mastering exactly one job.

Open standardArabic-firstSelf-hostablePer-layer trace
Built with TypeScript · Ollama · YAML manifests

nemo — the intent router

Sub-millisecond intent classification — no neural model, no GPU.

It sits in front of your language model and answers one question fast: do I know what this means, and where should it route? Built on hyperdimensional computing in pure TypeScript, and it learns with every interaction.

Sub-millisecondHDC (MAP)Arabic & EnglishNo GPU
Built with TypeScript · zero heavy deps

CST — Contextual Semantic Tokenizer

Turns raw text into structured semantic tokens — zero dependencies, reversible.

It maps every word to a field, role, relation or structural marker, turning text into a semantic stream ready for HDC, context injection, intent routing or semantic search — in Arabic and English alike.

Zero dependenciesMultilingualReversibleArabic morphology
Built with TypeScript

brain — the perception gate

Fast path first — most requests never touch an LLM at all.

The agent brain for Arabic and English: a mathematical perception layer classifies the domain in under a millisecond, and the language model runs only when it is uncertain. In a well-seeded domain, 60–70% of requests skip the model entirely.

60–70% LLM skipArabic-native reasoningSub-1ms gateSelf-hostable
Built with TypeScript · Ollama · Qwen

@msm-core SDK — the agent runtime

The execution loop, memory, guards and output — thirteen published packages you wire once and ship.

The application layer above MSM: an agent loop with guards and tool dispatch, context assembly from parallel sources within a token budget, composable output validation, five-tier memory, and durable long-running missions with human approval.

Agent loopTiered RAG5-tier memoryDurable missions
Built with TypeScript · published on npm

Developer engines

Core infrastructure we build and maintain — each a small core that replaces a whole stack of moving parts.

monlite

One file. Your entire backend.

An embedded database giving you documents, full-text and vector search, a durable job queue, cache, locks and cron — all in one file. When you outgrow it, the same code runs on Postgres.

DocumentsVector / RAGFull-textQueue
Built with TypeScript · Python · Go · SQLite

Locus

The reactive, geo-first datastore that speaks Redis.

Point any Redis client at it and get what vanilla Redis can't: a gap-free ordered changefeed, geo-native objects with live geofencing, and mergeable sketches — in one static binary with zero third-party dependencies.

Redis-compatibleChangefeedGeo-nativeGeofencing
Built with Rust

Motus

The live order & location engine.

The map-side core behind any delivery or field-service product: take an order, attach exactly one driver, stream live GPS only to the parties allowed to see it via a share link, record the track, then destroy everything on schedule.

Live trackingMulti-tenantWebSocket / SSEShare links
Built with Go · Locus

Chat Engine

A real-time messaging core in Elixir.

Connections, ordering, presence, receipts, catch-up and clustering behind clean interfaces — database-agnostic and unopinionated. Run a chat app, a live dashboard or an IoT command bus on it.

Fan-outPresenceReceiptsCatch-up
Built with Elixir · BEAM

Koine

Embeddable chat for your product.

Add real chat to any product in a day: one-to-one and group conversations, receipts, typing indicators, presence and catch-up. Your backend issues a token, your app opens one connection — identity, history and moderation stay yours.

Embeddable chatJWT identityMulti-tenantModeration
Built with Elixir · TypeScript

Pulsar

The live pulse of any topic — right now, and made sense of.

Type a subject and instantly see what the world is saying about it right now — with AI synthesising, clustering and answering over it: trends, semantic search, cross-source clusters and entity extraction.

Live ingestSemantic searchClusteringRAG
Built with Elixir · Chat Engine

Want one of these in your business?

Whether it's an off-the-shelf product or a custom build on our engines, we'll help you deploy it.

Get in Touch