Company
EigenForge AI Labs.
A Singapore data and AI capability partner, working on the gap between what an organisation knows and what it can act on.
What we are
Three things, and we name them in this order.
A delivery partner on two production platforms. A professional services business that places named specialists with client teams — on site or remote. And a product company whose own line is in development, and marked as such.
We work with organisations that have two things in common. Their most important questions span several systems. And their data cannot go wherever it likes — because of a regulator, a residency rule, a classification, or simply a board that will not have it. We bring the engineering, the platforms and the governance, and we stay accountable for the outcome rather than for handing over a licence.
What we do
Three moves, in order — and the order matters.
Most AI programmes that stall attempted the third before the first was done.
Make your data answerable
We read your systems of record where they are — read-only, no migration, no change to how they run — and join them under definitions your business has agreed, with lineage on every figure. A question that spans five systems gets one answer.
Make AI usable on it
Answers grounded in those governed records rather than a copy of them, with policy applied per user when the query runs, and inference inside your own boundary. Answers people will act on without checking.
Make the work get done
Agents that read, reconcile, prepare, route and act — each with a named owner, a budget, an approval boundary and a signed record. Routine work leaves the queue, accountably.
How we hold ourselves to it
Four commitments everything else has to keep.
Read-only
We never write back to a system of record.
No migration
Nothing moves before value is visible.
One counterpart
Platform, engineering and accountability in one place.
Evidence by default
Every answer and every action traces to its source.
The test to apply to us, and to everyone else: switch it off — does everything carry on exactly as before? If the answer is no, you have not added a layer. You have added a dependency.
How we talk about our work
A claims discipline, published.
- No client, deployment or site named without written consent.
- Use cases described as envisioned applications of system capability, not as work delivered for a named body.
- No published improvement percentages. Value is estimated from your own records before anything is committed.
- Never blurring what is in production with what is in development — including inside a proposal.
- Hard limits stated unprompted: no agent grants, denies or alters an entitlement, eligibility or penalty.
If a sentence in any of our material leaves you unclear whether something is production or development, tell us and we will rewrite it.
What we deliver on
Two production platforms, and our own platform in development.
What we deliver commercially today runs on two platforms that are in production. Everything of our own above them is in development and is marked as such wherever it appears — including inside a proposal.
The founder
Vivek Chadha.
Twenty-eight years in technology, spent on both sides of the table — the person who architects the system, and the person who has to sell it.
Vivek has worked in the technology and architecture space for 28 years, across a range of technologies — enterprise systems and data platforms, and then the emerging stack as it arrived: AI, machine learning and deep learning, DevOps, blockchain and data science. In his previous roles he led Asia-Pacific operations across 23 countries as managing director and global CTO of a multinational technology services group — the CTO who also owned the P&L, running digital inside sales and demand generation alongside the delivery of emerging technology programmes.
That combination is the point. He has architected the systems and carried the revenue number, and sitting on the vendor side of some of the region's largest transformation programmes he watched the same pattern repeat: the model was rarely the problem. The data was — held in systems that were each correct in isolation and never designed to be queried together.
He went on to serve as CTO and director of a technology services firm, and then as an independent advisor on AI strategy, technology integration, agentic AI and AI-native solutions, forward deployed engineering, and operational scaling — the five areas EigenForge is now built around.
He founded EigenForge AI Labs in Singapore in 2026 to do the work properly: read systems in place rather than migrate them, ground AI in governed records rather than copies of them, and put agents to work inside limits a named person set.
Business advisor
Ravi Nair.
A technology entrepreneur, data architect and AI innovator who has spent his career turning fragmented data estates into intelligent, AI-driven platforms for large enterprises.
Ravi founded PassionBytes and has worked across the full technology value chain — modern data architecture, cloud-native platforms, distributed computing, big data engineering, machine learning, agentic AI and enterprise analytics — with proprietary and patented work in data access, integration and analytics. As our business advisor he works with us on technology investment, data and AI roadmaps, governance frameworks, operating models and organisational readiness for AI-driven transformation — and our clients get the benefit of that perspective in every engagement.
The company
Registered in Singapore, on purpose.
We chose a jurisdiction that treats trusted, governed AI as a national position rather than a compliance afterthought — and we publish our particulars rather than leaving you to look them up.
| Legal name | EigenForge AI Labs Pte. Ltd. |
| Registered office | 160 Robinson Road, #14-04, Singapore Business Federation Center, Singapore 068914 |
We work with clients across the globe — in all key regions — through our sister companies and partners, so a conversation that starts in Singapore can be delivered wherever your systems sit.
Questions we are asked
Answered plainly.
- What does EigenForge actually sell?
- Outcomes, delivered as a service. You buy an answerable business question — framed, built, validated against figures you already trust, and measured against a reference point. Underneath, we deliver on two production platforms: a patented real-time data platform for the governed data layer, and a perimeter-native agentic platform for governed agents. You are not buying a licence and assembling it yourself.
- Do you have to migrate our data?
- No. The data platform reads your systems of record read-only. No migration, no write-back, no schema change, no effect on transaction processing. Switch it off and everything carries on exactly as before, which is usually what gets it through security review.
- What if a system has no connector?
- An extensive connector library already spans ERP, databases, applications, streams, files and object stores. A new connector is typically days of routine work rather than an integration project. Where no connector is possible at all — a mainframe routine, a proprietary store, a bespoke access program — a patented method exposes that program's output as a standard table. Nothing in an estate is unreachable.
- What is a forward deployed engineer?
- An engineer of ours who sits with your team, in your context, on your problem — with the platform in their hands and the authority to build. Not a consultant producing recommendations and not a support contact answering tickets. The first week is spent understanding the business instead of configuring software, and the engagement is designed from the start to transfer capability to your people.
- How is AI kept from inventing things?
- Grounded retrieval over an ontology built on live data, rather than chunking tables into a vector copy. Relationships between entities are preserved, joins stay correct, nothing is duplicated so nothing drifts, and every answer traces back to records. Policy is applied at query time per user. Inference can run entirely on local models inside your boundary.
- How do agents stay under control?
- Autonomy is granted by action class, not by agent. Each class carries a gate agreed before anything is built: act automatically within limits, prepare the action and hold it for a named approver, or simply advise and let a person decide. The gate is enforced on the decision path. No agent grants, denies or alters an individual's entitlement, eligibility or liability.
- Can this run without any connection to the internet?
- Yes. Five deployment topologies come from one codebase: on premises, private cloud, public cloud, air-gapped and federated. Air-gapped includes local model inference, so the capability does not depend on reaching an external service.
- How is value measured?
- From your own records, before anything is committed. Each lever — recovery, claim evidence, containment precision, working capital, spend consolidation, effort returned — is estimated against a measure your finance function already reports, then measured against that estimate afterwards. We do not publish generic percentage improvements.
- What do we have to provide?
- Read-only access to the systems in scope for the agreed objects, an accountable sponsor who can confirm the results are correct, and a few short working sessions to agree what the key measures mean. What is not required: replacing any system, migrating data, changing transaction processing, restructuring the organisation, or committing to a multi-year programme in order to begin.
- Where are you based?
- Singapore. Our registered office is 160 Robinson Road, #14-04, Singapore Business Federation Center, Singapore 068914 — a jurisdiction that treats trusted, governed AI as a national position rather than a compliance afterthought. We work across the globe in all key regions through our sister companies and partners.
- How do we start?
- A discovery conversation about the questions you are asked most often and answer least well. If there is something worth doing, we write up the most promising one — the question, the sponsor, the reference measure and the data readiness — before anyone talks about scope or price.
- Is this a platform sale or a services engagement?
- A service, delivered on platforms. You buy an outcome — a question answered, a process carried, a control evidenced — and we deliver it on our own production platforms, which we operate with you. You are not handed a licence and left to assemble the value yourself. Where an integrator leads the engagement, the same platforms arrive back to back behind them.
- Do agents act on their own?
- Only where you have decided they may. Every action class carries a gate — Auto, Approve or Advise — enforced in policy on the decision path, not in a document. Agents prepare, evidence and route; a named person commits anything that matters. In public-sector deployments no agent decides a citizen's entitlement, eligibility or penalty, at any setting.
- Does our data leave our perimeter?
- No. The platforms deploy inside your estate — on premises, in your cloud tenancy, or fully air-gapped — and read your systems of record read-only. Data is not copied out to us, models can be self-hosted, and zero-egress operation is an ordinary configuration, not a special edition.
- How do you price what is not built yet?
- Honestly. Everything we propose is labelled with its real status — live, component or design — in the same breath as its name. Live patterns are priced as delivery; component assemblies are scoped builds on proven parts; designs are priced as builds, never presented as available. If the honest answer is that something does not exist yet, that is what the proposal says.
Contact us
Write to us — a person answers.
Every message reaches a person and we answer all of them, including the ones where the honest answer is that we are not the right fit.
General and new work
Questions, first conversations, and anything that does not fit the other two.
Research and grants
Open problems, collaborations, and grant conversations.
Or write to us at the registered office: 160 Robinson Road, #14-04, Singapore Business Federation Center, Singapore 068914.
Talk to us
Tell us the question you are asked most often and answer least well. We reply in person.