Case studies
Projects running today in real companies. What was there before, what we built and what changed.
- 5×
- faster generating tender documents
- 4×
- more queries handled than before
- 30% → 1%
- fewer order errors in their ERP
- Optimal
- production planning
- Hours
- instead of weeks per manual
Why you won't see names
Our clients let us tell you what we achieved, not who they are. The sectors, sizes and processes are the real ones; the identity is anonymised by agreement with each of them. On a call we can go into whatever detail you need.
5×
faster generating tender documents
- Sector
- Electrical installations · public tenders
- Tamaño
- Industrial SME · small technical team
From three days to one morning per tender document
An electrical installation firm that lives off public tenders now produces its technical documents 5× faster and submits more bids with the same team.
The situation
- Every tender required a long technical document, assembled by copy-pasting the previous project.
- The work always fell on the same senior engineers — the ones with the least spare time.
- Finding the tenders worth bidding for was another manual job: checking public portals one by one.
- When two tenders landed in the same week, they had to choose which one to bid for.
What we did
- We built a crawler over the public procurement portals that only flags what matches their line of work.
- We loaded their back catalogue of tender documents: their writing style and technical judgement come from there.
- We locked in their templates and the sections the tender always demands, so none get missed.
- The engineer reviews, corrects and signs. What gets submitted is still their call.
The result
- 5×
- faster generating tender documents
- Same team
- more tenders submitted
- 1 format
- everything consistent and complete
Drafting stopped being the bottleneck. The question is now which tenders are worth bidding for, not how many they have time for.
4×
more queries handled than before
- Sector
- Vehicle dismantling & parts · WhatsApp
- Tamaño
- SME · consumer enquiries all day
4× more queries handled, without hiring
A parts dismantler whose enquiries all arrived by WhatsApp now answers instantly, at any hour, with the real prices from its catalogue.
The situation
- Almost every enquiry came in on WhatsApp: «do you have this part?», «how much?», «do you ship?».
- They were answered between customers, and evening or weekend messages waited until the next day.
- Every answer meant looking up the part and its price by hand.
- Sales were lost for not replying in time, not on price.
What we did
- We connected WhatsApp through Meta's official API, not a hacked-in number that drops out.
- We wired the pricing intelligence to their real catalogue, so what it quotes is what they charge.
- We wrote down their rules: what it can confirm alone, what it must never say, when to call in a human.
- We tested it on their hardest conversations before letting it talk to a customer.
The result
- 4×
- more queries handled than before
- 24/7
- answers in seconds, out of hours too
- Real price
- looked up in their catalogue, never improvised
Three weeks after the first delivery the client extended the scope. That's the best metric we have.
30% → 1%
fewer order errors in their ERP
- Sector
- Automotive parts · orders into the ERP
- Tamaño
- SME · orders by email, WhatsApp and PDF
Order errors dropped from 30% to 1%
A parts distributor that received free-text orders and typed them into the ERP by hand went from 30% of orders carrying an error to 1%.
The situation
- Orders arrived however the customer liked: email, WhatsApp, an attached PDF or a photo of a delivery note.
- Someone typed them into the ERP by hand, one by one, every morning.
- Almost a third went in with an error: wrong reference, wrong quantity, an old delivery address.
- Every error was paid for twice: in returns and credit notes, and in calls to fix it.
What we did
- We built automatic reading of the order, whatever channel and format it arrives in.
- Before it reaches the ERP, every line is validated against the catalogue, stock and that customer's terms.
- Anything that doesn't add up doesn't slip through: it goes to a review queue for a person to decide.
- Anything that does add up enters the ERP verified, with nobody typing a thing.
The result
- 30% → 1%
- fewer order errors in their ERP
- Minutes
- from order received to order recorded
- Fewer
- returns, credit notes and incident calls
The team stopped typing orders and now reviews only the exceptions. The error stopped being part of the process.
Optimal
production planning
- Sector
- Manufacturing · production planning
- Tamaño
- Industrial SME · several lines and shifts
Production planning, solved in minutes
A manufacturer that planned production by hand in a spreadsheet now gets the optimal plan in minutes and re-plans the same day when something breaks.
The situation
- The production plan was made by hand in a spreadsheet, and depended on one person.
- Any change — a rush order, a breakdown, material that didn't arrive — meant redoing the whole thing.
- Since redoing it cost hours, they often ran on a plan that was no longer the right one.
- The result: idle machines on one side and overtime on the other.
What we did
- We wrote down the plant's real rules: capacity, shifts, changeovers, priorities.
- The system cross-checks work orders, bills of materials, per-operation times and stock.
- When something unexpected happens it recomputes on the spot instead of forcing a restart.
- The planner adjusts what they want and publishes. They decide — with the work already done.
The result
- Optimal
- production planning
- Minutes
- to re-plan when something breaks
- Fewer
- idle machines and overtime hours
Planning stopped being someone's whole afternoon and became a decision taken with data, the same day.
Hours
instead of weeks per manual
- Sector
- Machinery manufacturer · technical manuals
- Tamaño
- Industrial SME · broad machine catalogue
A technical manual per machine, without writing it
A machinery manufacturer that wrote every machine's manual by hand now goes from the specification straight to a Word document in its corporate format.
The situation
- Every machine that ships needs its manual, and it was written from scratch each time.
- The information already existed — specs, drawings, photos — but scattered across departments.
- Each manual came out with a different structure depending on who wrote it.
- Translating it for each market multiplied the work by the number of languages.
What we did
- We loaded their previous manuals: structure, technical vocabulary and tone come from there.
- The system starts from the machine's specification and images and builds the document chapter by chapter.
- It comes out in Word in their corporate format, ready to review and publish — not loose text.
- The engineer reviews and corrects. What reaches the end customer still goes through a person.
The result
- Hours
- instead of weeks per manual
- 1 structure
- the same skeleton across the catalogue
- Corporate Word
- ready to publish, not loose text
Documentation stopped being what held up the launch of a new machine.
What's eating most of your time?
Tell us. On a short call we'll tell you whether we can help — and how.