Aspegio, home Write to an engineer

Engineering consultancy · Founded in London, 2022

Bespoke software, AI and models for problems that don’t fit the catalogue.

We build full-stack software, applied AI and simulation for engineering and commercial teams in enterprises and small companies, where getting it wrong is expensive.

Where we work

The gap between the catalogue and the consultancy

There is a gap between catalogue vendors and the big consultancies. On one side, products that answer the average question. On the other, slide decks and a bench. The problems our clients bring sit in the gap: too specific for a platform, too technical to delegate to a pyramid.

  • The people in the room do the work.

    One or two senior engineers, occasionally three, directing the coding agents that write much of the code. The engineers who scope your problem are the engineers who deliver it.

  • Scoping can end the project.

    Every engagement starts with a short, paid scoping window. If the data cannot carry the model, or a spreadsheet already answers the question, we say so and stop.

  • You own what we build.

    Source, documentation and the means to run it without us. Handover is part of the work, not an afterthought.

What we are working on

The work, by kind.

Most of it is under NDA, so we describe it by kind and never by client. This is what the practice has built, taught and advised on in 2026. Much of it serves technical sales in heavy industry: the teams who turn a customer’s enquiry into a priced proposal.

  1. 01

    Proposals for engineered equipment.

    Software that reads a customer’s enquiry documents, sizes and costs the equipment in code, and prepares the proposal for an engineer to review.

  2. 02

    Forecasting and planning.

    Tools that bring sales data and managers’ judgement into one forecast the business can sign off.

  3. 03

    Engineering calculations.

    Sizing, mass-balance and costing models moved out of spreadsheets and into tested code.

  4. 04

    Simulation.

    Discrete-event simulation of automated warehouses, and the studies run on it.

  5. 05

    Coaching technical teams.

    Engineers building their own tools and analyses with the AI assistant their organisation already has, with written guides and review from us.

  6. 06

    Advice on running it.

    Operating models, hosting, security review and build-or-buy assessments for organisations taking AI tools into daily use.

How we work

Frame, thread, iterate.

We do not spend months on a specification. We get a thin version working from end to end, quickly, and improve it while your team uses it.

  1. 01

    Frame

    Agree the problem, who will use the tool, and one real case to carry from end to end.

    What you get

    A written scoping note

  2. 02

    Thread

    Build the thinnest version that works from end to end. One real case goes in. One real answer comes out.

    What you get

    A first working version, end to end

  3. 03

    Iterate

    Put it in your team’s hands and improve it in short rounds, against their real work. We show you where it’s wrong, so you can trust where it’s right.

    What you get

    A tool your team has shaped, and full ownership at handover

The method in full

Why build a model

On average, nothing waits.

Factories, fleets, queues and teams rarely behave like the spreadsheet average. The cheapest place to discover that is inside a model: a small virtual world that can be tested, challenged and improved before money moves.

Here is the smallest one we can show you. One machine that finishes ten jobs an hour. Keep demand below that and the averages say the machine keeps up. Then watch what the jobs do.

A live model · one machine, one queue

Simulated hour 48

9.0 jobs an hour, against a capacity of 10

The spreadsheet

0

jobs waiting. Utilisation is 90%, so on average the machine keeps up.

The model

8.3

jobs waiting on average. A job waits 55 minutes. The longest queue was 82.

Jobs arrive at random and take a random time to finish. The model’s figures come from 200,000 simulated jobs (22,215 hours), run from seed 13792055. Queueing theory covers this simple case exactly and gives 8.1. Real systems are rarely this simple, which is why we simulate them.

Who we work with

Engagements since 2022.

Most of our work sits under client confidentiality. We are comfortable under NDA by default.

  • FLSmidth
  • Mytra
  • Rolls-Royce
  • University of Oxford
  • Anglo American
  • First Mode

A short film

Before money moves.

A short film by Aspegio. One enquiry: read by a language model, calculated by code, and signed by an engineer before any money moves.

Before Money Moves · 1920 × 1080 · 75 seconds · English captions
Harry Munro, CEO and Director of Aspegio Ltd and Chief Technology Officer of Aspegio (Bermuda) Ltd.
Harry Munro · UK CEO and Director · Bermuda CTO

Who you work with

Work directly with the engineer.

Aspegio is led by Harry Munro, CEO and Director of Aspegio Ltd in the UK and Chief Technology Officer of Aspegio (Bermuda) Ltd. He is a chartered mechanical engineer who has worked across AI, software and simulation since 2011.

Earlier work includes engineering and simulation programmes with Improbable, the UK Ministry of Defence and Transport for London. He founded Aspegio in 2022, after a decade inside larger organisations where the most interesting work never fit the catalogue.

The engineers who scope your problem are the engineers who deliver it.

Harry Munro CEng MSc BEng (Hons) MIMechE

About the practice

When we are not the right call.

  • Off-the-shelf software resale.
  • Very large, multi-year transformation programmes.
  • A standing team of twenty. We work in small senior teams.
  • Work that needs to be cheap above all else.

Bring us the problem that doesn’t fit.

No product to sell you. No bench to keep busy.

Write to an engineer harry@aspegio.com

First call 30 minutes, always free.