Manufacturing memory · Manchester, UK
The shop floor already knows what is wrong with the design.
It has no way to tell the person drawing the next revision. Sclptr is the record that does that. It is captured at the bench, checked by a human, tied to the exact design feature, and waiting at the drawing when that feature is changed.
Pre-launch · in development · not yet trading
- Part & feature
- Brake bracket · Hole 3
- Observed
- 12 occurrences
- Impact
- Rework · cycle time · ergonomics
- Current annual loss
- £4,100
- Change cost
- £1,200
- Expected saving
- £3,500 / yr
- Payback
- 4 months
Illustrative worked example. No customer data exists yet.
01. What it is
A persistent record of how a product is actually built.
Sclptr is a manufacturing-memory platform for mid-market UK engineered-product companies. It captures and validates shop-floor issues, links each one to the affected part, operation, design feature and revision, and builds a lasting record of how products are really manufactured.
Each record carries its operational and financial context, how often the problem happens, what it costs, what a fix would cost, so design and manufacturing teams can judge a proposed change on evidence rather than recollection.
Using that accumulated evidence, Sclptr is being developed to assess how the next revision is likely to perform in assembly before it reaches production. That capability is staged research, not a shipped feature.
02. The problem
The factory forgets, on a predictable schedule.
The person who builds a product and the person who designed it sit in different rooms. What the builder learns decays before it reaches the drawing.
-
07:14
A worker cannot get the screwdriver square to a hole. It cams out and leaves swarf. He works around it. There is a shift to finish.
-
15:00
Eight hours later it is “a bit of an issue on the bracket” in an end-of-shift meeting. Nobody records which hole, or why.
-
2 years on
Revision 2 is drawn with the same hole in the same place. The worker could have said exactly what to change. Nobody asked him.
“Reduce this process from a number of iterations to three, two or even lesser — that would be incredibly valuable. That's something manufacturing companies would want.”
A process engineer at a UK automotive manufacturer, reviewing the concept in a personal capacity. Not an employer endorsement.
03. How it works
Six steps, and the sixth is why the other five keep working.
Reporting systems die when nothing visibly happens to what people report. The answer returning to the operator is architectural here, not a courtesy.
Part, job and drawing version come from the scanned code, not from the voice recording, so a misheard word cannot file a problem against the wrong part.
04. Trust by design
Four rules, held by architecture rather than by policy.
Products in this category usually die because the people on the floor refuse to use them. Each of these costs Sclptr something real, and each is held anyway.
-
Capture only when the operator chooses
Triggered by a foot switch. Never ambient, never always-on. The system structurally under-captures, and that ceiling is accepted.
-
No individual scoring, ever
The record is about the product, never the person. That forgoes an entire product category, including per-operator analytics, ranking and coaching triggers.
-
A human validates every record
Nothing enters the permanent record until a supervisor confirms it. Audio and transcripts are held briefly for that check, then deleted if the record is rejected.
-
Each customer's data stays sealed
Schema-per-tenant isolation. No model is ever trained across customers on raw data, which forgoes the fastest-compounding form of improvement.
05. What it costs
Priced per production line and per capture point. Never per user.
Charging per seat would tax the exact behaviour the product depends on. So it does not. Adding operators is free.
£5,000
90-day paid pilot
One workflow, up to three capture stations. Credited in full against the first year on conversion. Sclptr funds the trial hardware.
£9,000/year
Platform, per production line
Dashboard, design-history view, validation workflow, onboarding, support, and building the customer's part-and-feature list.
£3,000/year
Per capture station
Software licence and support at one capture point: tablet, close-talk microphone, foot switch, QR context, mount. Depth capability added from Year 2 at no extra licence fee.
What one production line costs each year
- Platform, per line: £9,000
- Capture stations: £3,000 each
£18,000 a year, for one line with three stations, is the configuration written into all three signed letters of intent. The platform fee is the same £9,000 in every column; only the station band grows.
06. The market, counted
Companies counted from public statistics, not estimated.
123,048
manufacturers across the markets counted so far
The UK, Ireland, the Netherlands, the Nordics, Germany and Austria. The count excludes the United States and Japan.
7,900
UK manufacturers the current product serves
Beachhead sectors at 10–249 staff (5,005) plus six growth sectors at 50+ staff (2,895).
39.3–39.8%
of UK manufacturers sit in Sclptr-relevant sectors
Stable across every year of ONS data from 2021 to 2024, a structural feature, not a point estimate.
Sources: ONS ad-hoc table 2750 (ref. AH1690), 2024 data, released 8 April 2025 · DBT Business Population Estimates 2025 · Eurostat SBS_SC_OVW, manufacturing, 2023.
07. Who is building it
Sheikh Daman Alam
Mechanical engineer, BEng (Hons) from the University of Manchester. The idea came from his own year as a manufacturing engineering intern at GE Vernova in Lisburn, working full time on high-voltage equipment assembly, where the hardest problem was not spotting assembly issues but getting validated evidence of them to the design engineer responsible for the next revision.
Before that: design engineer on inspection robotics at Orbix NDE; engine assembly at Honda in Gurgaon; signage design and CNC production at MCR Signs in Manchester. He has also built and shipped Mezbaan, a live multi-tenant property-management platform, single-handedly from concept to paying customers, which is the evidence that he can build and ship software, not only specify it.
Sole founder, working on Sclptr full time. An unpaid advisory group has signed letters of support.
- Arif Syed, Co-Founder and CTO, Orbix NDE LinkedIn ↗
- Professor Lee Margetts, UKAEA Chair of Digital Engineering for Fusion Energy, University of Manchester LinkedIn ↗
- Dr Akın Ataş, Lecturer in Engineering Design, University of Manchester LinkedIn ↗
- Dr Alessandro De Rosis, Senior Lecturer in Mechanical and Aerospace Engineering, University of Manchester LinkedIn ↗
- Andrew Hatcher, Managing Director, The Applied Knowledge Network LinkedIn ↗
08. Where this is today
Stated plainly, because the stage matters.
Not yet registered. Incorporation is deliberately held until the founder's immigration position is settled.
In development. The capture-to-validation workflow is being built; version 1 is targeted at six months from registration.
None yet. No paid pilot has started and no customer is paying a subscription.
Three signed, non-binding letters of intent from Manchester manufacturers, July 2026, each naming a specific workflow at the stated price.
Discovery conversations with manufacturers in the target segment, plus two documented independent reviews, one by a named practitioner working in the industry and one by a chartered engineer. Market size is counted from published national statistics rather than estimated.
None. Every forward figure in Sclptr's planning documents is a forecast or an assumption, and is labelled as one.
Self-funded. No external investment has been raised and none is assumed in the plan.
Detailed planning documents, including the business plan, financial model, and technology and research programme, are available to serious enquirers on request.
09. Contact
If you make something, and the floor keeps telling you the same thing twice.
Sclptr is looking for mid-market manufacturers willing to run a paid pilot on one real production workflow, and for people who want to argue with the idea. Both are useful.
Sheikh Daman Alam · Founder · Manchester, United Kingdom