Engagements
We start inside your business, not in a brief.
A foundry answers whether a thing can be built well. The harder question is whether this business should want it built at all — and you can only answer that from inside the business.
Platforms
5in catalog
Verticals
6
Time to an instance
2–4weeks
New engagements
3a quarter
Companies cast
11same engine
Two motions. One of them is where the revenue is.
Most firms hold one of these and call it a business. Held together, each one makes the other cheaper.
Deployed
A senior engineer works inside your business.
Not a discovery workshop and not a requirements document. Someone who can ship sits with the people doing the work, learns the operation well enough to be useful in it, and comes back with the one workflow that is actually worth building. This is the forward-deployed model, and it is where the judgement lives.
Answers: should this account want it built?
Cast
The factory builds it, on a stack that already exists.
What comes back goes to an agent workforce running against a platform we have already built and hardened elsewhere. Because the spine is reused, the work is measured in weeks rather than quarters — and it arrives deployed, observable and handover-ready rather than as a prototype.
Answers: can we build it well?
Every deployment makes the next build cheaper.
The deployed engineer finds what is worth building; the factory builds it; whatever generalises goes back into the catalog. That loop is the asset — not any one platform in it.
01
Deploy
An engineer goes into the account — your systems, your constraints, your regulator. Paid from day one; no speculative discovery phase.
02
Find the workflow
The output is not a backlog. It is the single operation where software changes the economics, argued from evidence inside the business.
03
Build on a platform
The factory instantiates an existing stack and customises it. Weeks, not quarters, because the compliance spine and infrastructure are already paid for.
04
Generalise
Whatever proved reusable returns to the catalog. The next account starts from that platform instead of from zero — and occasionally a platform earns its own company.
and the account that comes after starts from step three
What the loop has already produced.
Each of these began as a workflow found inside an account. The platform stays ours; the instance built for the partner is theirs. Partners are described by sector — we do not publish client names.
LegalOS
A legal operations stack for in-house counsel — contract lifecycle, obligation tracking, and an auditable trail of who agreed to what, when.
What we found
Counsel were not short of contract software. They were short of an answer to which obligations across a decade of executed agreements were about to bind a business unit that had never read them.
The legal function of a large diversified Indian conglomerate
The compliance and engagement operating system for accountancy practices — tenant-isolated Postgres with row-level security and document trails built in.
What we found
Compliance calendars, client engagement, and working papers lived in three systems and a spreadsheet. The partner-level question was never "can this be one system?" but "which of these does a regulator actually hold us to?"
An international audit and advisory network
Foreign Remittance Withholding
Built on LedgerVault
A case workbench for payments to non-residents — take in the document pack, resolve the rules in force on the payment date, characterise the supply, compute the withholding, and prepare the forms the team files by hand. It prepares; it does not file.
What we found
Every cross-border payment out of a capability centre carried a withholding question the team answered from memory and a spreadsheet. The work worth building was not another tax tool; it was a versioned registry of rules, treaties and precedents, and an auditable determination for every payment.
A firm that builds and operates GCCs for US clients
Keel
The operating layer for standing up a Global Capability Centre — talent mapping, workspace, finance and HR run as one instrumented workflow rather than six vendors, so an office park can hand over a running centre instead of a floor.
What we found
Every new centre repeated the same ninety-day scramble from scratch. The work worth building was not another HR tool; it was the repeatable spine that made centre number six cost a fraction of centre number one.
In build for commercial real-estate developers whose office parks host capability centres — not yet in an account
Mould
A logistics operations platform for asset-heavy networks — movement, utilisation and exception handling against the unit economics of each lane.
What we found
The dashboards were already good. What no one could answer was which lanes were quietly unprofitable once you charged them honestly for the assets they tied up.
An industrial logistics group
What an engagement actually looks like.
- Who shows up
- Senior, full-stack
- People who ship across the stack, in your context. We do not hand off between a strategy team and a delivery team — the person who found the problem is on the build.
- Sprint length
- 2–4 weeks
- Each sprint ends in a deployed environment someone can actually use. The cadence forces decisions early and exposes integration risk before it compounds.
- What you own
- Your instance
- The customised instance built for your business is yours, on terms set per deal. The underlying platform stays ours — which is precisely why you are not paying to build it from scratch.
- Where it runs
- Vercel · AWS Mumbai · on-prem
- Cloud-native by default, on-prem when your regulator demands it. Same manifests either way.
- How many
- Three a quarter
- We take a small number of new engagements at a time so that deployment means something. Most are referred.