that would worktw2

Systems for companies that make, move and sell real things.

We build and run software for small and mid-sized operators: planning engines, data platforms, integrations, and the unglamorous glue between them. What we ship is ordinary, dependable software with no hidden AI cost. It goes live in weeks, and it's priced so you can leave it switched on.

Bring us a problem How we work
Manufacturing · Logistics · Distribution · Trading
What we do

The jobs we take

If your business runs on physical goods and your problem involves a computer, it probably fits. The domain varies. The shape rarely does: useful data trapped in disconnected systems, and decisions made from spreadsheets and memory.

Two systems that won't talk
ERP to warehouse. Production floor to finance. The quoting tool to the thing that actually schedules the work. We build the joins and keep them maintained.
A planning problem
Production scheduling, procurement, allocation, replenishment. Real optimisation: mathematical programming against your actual constraints. Not a dashboard with opinions.
Data you can't get answers from
Years of operational data across five systems and no way to ask it a question. We build the platform that makes it answerable, and the pipelines that keep it current.
A manual process eating someone's week
Re-keying orders. Chasing documents. Compiling the same report every Monday. We automate it properly. Deterministic where it can be, AI where the input is genuinely messy.
Something that doesn't fit a category
Half-finished software from a previous project, or a vibe-coded brittle mess. A pricing curve that is too complex for Excel. We like the odd-shaped and difficult projects.
Opinions we build by
01

Deterministic by default. No token meter.

Routine questions shouldn't burn AI tokens or give a different answer every time. Deterministic engines do the daily work. We use AI where it makes most sense: reading messy documents, turning a question into a query.

02

The pilot proves it. Test agreed up front.

Weeks, not months. Before we build anything, we agree in writing what "working" means for your operation. If it doesn't pass, you've spent little and learned a lot. If it does, you already trust the next step.

03

No handoffs. The full chain, one team.

Data engineering, the model, the software, the hosting, the maintenance: the same people carry the problem end to end. Nothing gets lost between a strategy deck and an implementation partner, because there isn't one.

04

Built fast, because building got fast

Modern AI has collapsed the cost of building software. We pass that on as speed and scope, not as a chatbot bolted to your invoices.

Proof of method

The toolset

These aren't off-the-shelf products you configure yourself. They're our tools: proven components we shape to your operation, so you get the speed of a product with the fit of something built for you. Together, there are no seams.

Data foundation

Plumb

Gets your data out of the ERP and spreadsheets it's trapped in, and into a model you can plan on.


  • A domain model and a database of record
  • The foundation everything else stands on
  • In most projects, the bulk of the work, and we say so
Planning engine

Lathe

Schedules, allocations, procurement plans, price curves, all computed by real optimisation against your actual constraints.


  • Same inputs, same answer
  • You can see why it chose what it chose
  • Open-source solvers by default
Working record

Seam

The record of the work itself: jobs, decisions, and the systems they touch, tied together so nothing lives only in one person's head.


  • Jobs, decisions, and their trace
  • One shared document of work and decisions
  • The connective tissue for the other two
Reference

Recent work

The domains vary. The shape of the problem, and the shape of the fix, stay recognisably the same.

Maritime · Commodities

Trading intelligence for marine fuel procurement

A full platform for a joint venture between two of the world's largest shipping and commodities businesses: fixture ingestion, a governed data warehouse, pricing analytics, and forward-curve construction for a market with no clean benchmark.

Financial services

National-scale data platform architecture

Cloud data platform design for a central bank's development programme: lakehouse architecture, governance, and a security model for institution-wide analytics.

Manufacturing · Supply chain

Advanced planning and scheduling

Optimisation-driven production planning for manufacturers, delivered directly or as the engine inside a systems integrator's implementation. If you deliver ERP or MES projects and need the hard mathematical core, talk to us.

Small business · Applied AI

Practical AI for operators

Hands-on workshops and automations for small businesses that move real goods. We cover cases where AI helps, where it doesn't, and how to tell the difference before spending money.

Who we are

About us

There's a sea of AI startups out there with sleek one-syllable names and not much underneath. We went the other way: a name with a few more words in it, in the spirit of Iain M. Banks's ship Minds. That Would Work is also the standard. Not "that would demo well". That would work.

tw2 is two practitioners with over forty years of AI, data science and operations research between them. They cover every part of bringing these systems to life, from strategy to hardened production. AI is easy to use and can do remarkable things now. Figuring out how to extract value is hard work.

JC

John Curry

Twenty years in tech and AI across financial services, public sector, telecoms, manufacturing, mining and retail. John shapes AI strategy, finds the ideas actually worth doing, and builds them: data science, old-school AI, GenAI, and getting systems into production. The business problem comes first; the tech comes next. Based in Ireland, working internationally.

AB

Anindya Banerjee

Ani learned supply chain mathematics at Apple, at a scale where the numbers really have to hold up, then built Callisto Analytics: forecasting, optimisation and planning systems for manufacturing, retail and logistics. If your problem has moving parts, constraints and consequences, it's Ani's kind of problem. Based in Perth after years in Singapore, working across APAC, Europe and the US.

Commercial

How engagement works

01

Bring the problem

A conversation, on us. You describe the problem in your language; we tell you honestly whether it's worth solving with software, including when the answer is no.

02

Pilot

A fixed-price pilot in weeks, with the success test agreed in writing before we start. Real data, real users, a real go/no‑go.

03

Subscribe

One monthly number: the platform licence plus continuous development. We host it, maintain it, and keep building. Your standing technical team, without the headcount.

Why a subscription

Software becomes an unmanageable black box without continuous attention. The subscription means your system improves every month it runs. It also means we're accountable for outcomes over years, not invoices. Most clients pay less than the cost of a single technical hire for a team that already knows their operation.

that would work

We want to hear it from the most sceptical person in your operation. If a proposal can't get there in plain language, it doesn't get built.