Deterministic by default
Routine work uses deterministic software, so the same inputs produce the same answer. We use AI for documents and natural-language questions.
Manufacturing · Logistics · Distribution · Trading
We build systems for companies that make, move and sell real things.
We build software for small and mid-sized operators. Goes live in weeks, has no hidden AI costs and is priced to stay in use.
What we do
The toolset
Most projects start with systems that do not connect, planning stuck in spreadsheets or data you cannot get answers from. We use three products to connect and optimise your operation.
Data foundation
Plumb
Plumb brings ERP and spreadsheet data into one model that is ready for planning.
- A domain model and planning database of record.
- Keeps source systems synchronised.
- Usually the largest part of the work, and we say so.
Planning engine
Lathe
Lathe calculates schedules, allocations, procurement plans and price curves against your real constraints.
- Same inputs produce the same answer.
- You can see why it chose what it chose.
- Open-source solvers by default.
Working record
Seam
Seam records jobs, decisions and affected systems, so knowledge does not live in one person's head.
- Jobs, decisions, and their trace.
- One shared document of work.
- The connective tissue for the other two.
The morning run
- 06:00ERP and warehouse synchronisedDone
- 06:1514 supplier documents added to the modelAI
- 07:30Monday margin report compiled and sentDone
- 07:41Week 29 production plan solvedOptimal
The routine work ran without anyone watching a screen.
Week 29 production plan
| Line | Mon | Tue | Wed | Thu | Fri |
|---|---|---|---|---|---|
| Line 1 | #4211 · 12t | #4302 | |||
| Line 2 | #4187 · 8t | ||||
| Line 3 | #4198 · 6t | Maintenance | #4211 | ||
Why is #4187 on line 2?Line 1 is committed to #4211 until Wednesday. Moving it would break the six-hour cure-time constraint and add €4,120.
How we work
One team from first conversation to production
The same people handle data, models, software, hosting and maintenance. Nothing gets lost in a handoff.
Reference
Recent work
Different sectors produce similar problems: scattered data, difficult decisions and systems that need to hold up in daily use.
Maritime · Commodities Trading intelligence for marine fuel procurement
We built a procurement intelligence platform for a joint venture between two of the world's largest shipping and commodities businesses. It covers fixture ingestion, governed data, pricing analytics and forward curves.
Financial services National-scale data platform architecture
We designed lakehouse architecture, governance and institution-wide security for a central bank development programme.
Manufacturing · Supply chain Advanced planning and scheduling
We built optimisation-driven production planning for manufacturers, both directly and as the planning engine inside ERP and MES implementations.
Who we are
About us
That Would Work is a team of two practitioners with more than 40 years of experience across AI, Data Science and Operations Research.
The name is in the spirit of Iain M. Banks’s ship Minds, and it doubles as our standard. The software has to work in the operation, not just in a demo.
2Practitioners
40+Years
3Regions
John Curry
John has spent 20 years in tech and AI across financial services, the public sector, telecoms, manufacturing, mining and retail. He turns useful ideas into production systems. He is based in Ireland and works internationally.
Anindya Banerjee
Anindya learned supply chain mathematics at Apple, then built Callisto Analytics to deliver forecasting, optimisation and planning systems. He is based in Perth and works across APAC, Europe and the United States.
Commercial
How engagement works
Bring the problem
Tell us how the operation works and where it fails. We will say whether software is worth the effort, including when the answer is no.
Run a pilot
We run a fixed-price pilot in weeks with real data and real users. We agree the success test in writing before we start.
Subscribe
One monthly number: the platform licence plus continuous development. We host it, maintain it, and keep building.
Why a subscription: The system improves every month, and we remain accountable for the result. Most clients pay less than the cost of a single technical hire, for a team that already knows their operation.