Thirteen solutions, every one shipped in production.
This isn't a wishlist. Each service maps to systems we've built and run — with the exact stacks behind them.
Causal Labs is a Malaysia-based software partnership serving Singapore, Malaysia and Southeast Asia, offering thirteen production-proven capabilities across four areas: full product platforms (multi-tenant SaaS, spreadsheet-to-web migration, 3D web), automation and messaging (workflow engines, WhatsApp, trading systems), AI and data engineering (LLM pipelines, causal inference, graph databases, document intelligence, media AI), and RegTech plus the cloud infrastructure underneath. Engagements typically run 4–16 weeks depending on scope.
Malaysia is our base, where our MyInvois/LHDN e-invoicing work is built specifically for Malaysian compliance requirements. Singapore is a core market we serve regularly, and we take on projects across Southeast Asia more broadly, plus further afield where the engagement is a genuinely good fit. Geography narrows the default starting conversation, not who we'll work with.
Platforms & Products
Full product builds — multi-tenant SaaS, spreadsheet replacements and premium 3D web experiences.
Most engagements in this cluster start from a specific business constraint rather than a blank page: a spreadsheet that's outgrown what one person can safely maintain, a product idea that needs multi-tenant isolation from day one, or a marketing team that wants a 3D-caliber site without shipping a WebGL bundle that tanks Core Web Vitals. We build the full platform — data model, auth, tenancy, deploy pipeline — not just the interface layer, so the system is something you can hand to another engineer without archaeology.
Multi-Tenant B2B SaaS Platforms
End-to-end product platforms with strict tenant isolation, real-time collaboration and audit-grade compliance.
Interactive 3D & WebGL Experiences
Premium, performant 3D on the web — the kind you're looking at right now.
Excel-to-Web Migration & Modernization
Retire fragile spreadsheets for fast, multi-user web apps — with proof the numbers still match exactly.
Automation & Messaging
Systems that remove manual steps from how a business actually runs — workflows, WhatsApp and trading.
The pattern across workflow, WhatsApp and trading automation is the same: a process that works fine by hand or in a no-code tool until volume, error-handling or auditability requirements outgrow it. We design for the failure case first — retries, idempotency, structured logs — because an automation that silently drops a message or double-fires a trade is worse than no automation at all. Visibility into what the system is doing matters as much as the automation itself.
WhatsApp & Conversational Automation
Shared team inboxes, broadcast campaigns and automated conversations on the WhatsApp Business stack.
Workflow Automation & Integrations
Visual, drag-and-drop workflow builders with durable execution that connect the tools a business already uses.
Algorithmic Trading & Quant Systems
Backtesting frameworks, exchange connectivity and autonomous bots — engineered for correctness under real money.
AI & Data Engineering
LLM pipelines, causal inference, graph databases, document intelligence and media AI.
LLM pipelines, causal inference, graph databases, document intelligence and media AI all face the same core engineering problem: turning messy, real-world input into something structured enough to build a product decision on. That means source traceability (so an extracted field can be checked against the original document), cost control (batch processing, caching, model selection by task), and — for the causal work specifically — validation gates that catch a confounded result before it reaches a dashboard.
LLM Pipelines & AI Data Extraction
Multi-stage extraction pipelines that turn unstructured documents into clean, structured, queryable data.
Causal Inference & Applied Data Science
Going beyond correlation — estimating real causal effects from messy, real-world financial data.
Graph Database Engineering
Knowledge graphs that model entities, relationships and time so you can ask questions tables can't answer.
Document Intelligence & OCR
Reliable extraction of structured data from PDFs, scans and spreadsheets at scale.
Media & Video AI
Transcription, diarization and AI generation pipelines for audio and video at production quality.
RegTech & Compliance
Regulated-domain delivery done by the book, starting with Malaysian e-invoicing.
Regulated-domain work — starting with Malaysian e-invoicing (MyInvois/LHDN) — is delivery done by the book: UBL 2.1 schema conformance, XAdES digital signatures, audit logging and encrypted credential storage aren't optional extras here, they're the spec. We treat compliance requirements as fixed constraints to design around from the start, not a checklist to satisfy after the system already works.
Cloud & Infrastructure
The deployment, observability and security layer underneath every system above.
Every system above needs somewhere to run, something watching it, and a way to ship a fix without downtime. This cluster is the deployment, observability and security layer underneath the other twelve services — containerized builds, CI/CD, secrets management and encrypted credential storage — so a production incident is diagnosable in minutes rather than a multi-hour SSH session.
Need the website itself, not just what runs behind it?
Custom website design and development, priced from S$2,500 — real published tiers, no quote-gate.
How an engagement runs
The same six-step process behind every service above, whether it's a four-week automation or a sixteen-week platform build.
Requirements gathering
We start with the process or system as it actually runs today, not as a hypothetical spec — the failure modes of the current spreadsheet, script or manual workflow usually define the real requirements better than a feature wishlist does.
Scoping & proposal
You get a written scope with a concrete delivery window and a stack recommendation, not a vague estimate range — the leadTime figures listed against each service below reflect what similar engagements have actually taken.
Design
For platforms this means a real data model and tenancy design before any UI; for automation and AI work it means mapping the failure cases (a failed API call, a malformed document, a duplicate webhook) before writing the happy path.
Build
Senior engineers build the system directly — this isn't a spec handed to a separate delivery team. The stack for each service (React Flow and n8n for workflow automation, PaddleOCR for document intelligence, and so on) is chosen for the specific problem, not defaulted to a house framework.
QA & review
Verification matches the stakes of the system: cell-for-cell golden tests for a spreadsheet migration, idempotency and retry tests for a workflow engine, schema conformance checks for e-invoicing. Where correctness is provable, we prove it before handover, not after.
Handover & support
You receive the system with documentation, not just a deploy link — and a support window afterward, because a workflow engine or SaaS platform that ships without anyone able to operate it isn't actually done.
Questions about working with us
Do you take on smaller or partial engagements, or only full builds?
Both. Some engagements are a full platform build; others are a single automation, a document-extraction pipeline, or an infrastructure cleanup bolted onto an existing system. Scope is set in the proposal stage against what the process actually needs, not a fixed package size.
Can one project combine multiple services from this list?
Yes — this is common. A RegTech e-invoicing engagement, for example, often needs document intelligence for the source data and cloud infrastructure for the deployment. Related services are cross-linked on each service page for exactly this reason.
Do you work with early-stage startups as well as established SMEs?
Yes, though the shape of the engagement differs — a startup's SaaS platform typically prioritises speed to a working product, while an established SME replacing a spreadsheet prioritises provable parity with the numbers already in use.
What happens after handover — is there ongoing support?
Every engagement includes a handover period covering documentation and knowledge transfer. Ongoing support and maintenance retainers are available separately and are scoped case by case, depending on the system's operational needs.
Do you sign NDAs and handle client data confidentially?
Yes. Regulated-domain and financial engagements in particular routinely involve NDAs, and we design access, credential storage and logging around whatever confidentiality requirements the engagement carries.
How is pricing structured?
Fixed-scope pricing tied to the proposal from the scoping stage, sized to the delivery window for that service. We don't run open-ended hourly billing — the goal is a number you can plan a budget around before work starts.
Do you only work with Malaysian companies, or do you take on custom software development in Singapore and beyond?
Malaysia is our base — our MyInvois/LHDN e-invoicing work is built specifically for Malaysian compliance requirements — but Singapore is a core market we serve regularly, and we take on projects across the wider Southeast Asia region. We'll consider engagements outside the region too; tell us what you need and we'll give you an honest answer on fit.