The AI brain
for businesses.
Every distributor is losing money in two directions at once — cash sitting in stock nobody ordered, and orders they could not fill. Purpigal learns how your business actually decides, then starts making those calls with you.
Built on real transaction history, running in real pilots:
- transactions analysed
- 82,336transactions analysed
- distinct SKUs
- 431distinct SKUs
- industries in pilot
- 2industries in pilot
The gap
Your systems store records. None of them know why.
- ERP
- CRM
- HRIS
- Accounting
- Slack
- BI dashboards
- Documents
- Spreadsheets
Every department already has software. None of it knows why the company does what it does.
So the same two mistakes repeat. Stock is bought because it was bought last year, and it sits. A customer asks for something that ran out three weeks ago, and the order goes elsewhere. Neither appears on a P&L as a mistake — one looks like inventory, the other never happened at all.
The judgment that would prevent both lives in people. When the procurement manager leaves, the purchasing strategy leaves with them. When the CFO retires, decades of it go too.
What it does
One brain. Every function.
Purpigal sits above the systems you already run and learns how the business actually decides. It starts where every decision has a measurable outcome — and expands from there.
Hover or tap a region. Arrow keys work too.
Live today
Procurement & Inventory
Cerebellum — where precise, repeated sequences are learned and timed
Knows future sales, warehouse space and supplier reliability before buying. This is what runs today: upload the files you already keep, ask in plain language, and get back what to restock, how much, and the rows that prove it.
82,336 transactions analysed · in pilot with a steel distributor and a coffee chain
How trust is earned
Companies don’t hand over decisions on day one.
So the system earns them in stages — observing before it advises, and advising before it acts.
- 01ObserveUnderstand the business.
- 02AdviseRecommend actions.
- 03AssistExecute with approval.
- 04AutomateHandle routine decisions independently.
- 05OptimizeContinuously improve based on outcomes.
Each stage has to deliver enough value on its own to earn the right to the next one.
What a pilot looks like
Four weeks from your files to your first answer.
No integration project, no data cleanup phase, and nothing to install before you find out whether it works.
Step 1
We connect to your data.
Wherever it already lives — sheets, documents, an ERP, or all three. Bring as much history as you can lay hands on; more of it is always better, and none of it needs cleaning up first.
Step 2
First answers in two to four weeks.
That is from receiving your data to something you can actually use, not to a demo.
Step 3
Then you can ask your business questions.
Demand forecasts that refresh on their own, restocking recommendations, expiry alerts, and custom alerts for whatever you need watching — answered against your own data and how your company works, not generic benchmarks.
Step 4
We meet every week.
With the people who will actually use it and the people who handle the data. That is how the system learns the way your business decides rather than the way software assumes it does.
A pilot is the beginning of something you keep using. The longer it runs, the more of your patterns it has learned, and the less of your day goes on decisions it can already make well.
Pricing depends on what your business actually needs, so we quote it after the first call rather than before.
Validation
Running on real operations.
Purpigal was built against a Lebanese steel distributor’s full transaction history — 82,336 transactions across 431 distinct SKUs, every one of them real.
It is now in pilot with a steel distributor and a coffee chain — two businesses whose stock behaves nothing alike, which is the point.
We are taking on a small number of additional pilots.
- transactions analysed
- 82,336transactions analysed
- distinct SKUs
- 431distinct SKUs
- industries in pilot
- 2industries in pilot
Pilot partners are described by industry and not named, at their request.
See it against your own numbers.
Bring the history you already keep. We will show you what it finds in it before you commit to anything.
We usually reply within a working day.
How it ships
A template per industry, then tuned to your company.
An operations problem is not generic. A steel distributor and a coffee chain both run out of stock, but almost nothing else about them is the same — the lead times, the shelf life, the vocabulary, the decisions that actually matter.
So Purpigal ships as a vertical template: the data shapes, the decision logic and the language of one industry, already in place. From there it is customised to how your company runs — your suppliers, your thresholds, your approvals. You are not configuring a blank tool.
Steel distribution
Thousands of SKUs, long supplier lead times, and working capital tied up in stock that moves slowly.
In pilot
Coffee retail
Perishable stock across multiple sites, daily reordering, and demand that swings with weather and footfall.
In pilot
Your industry
Distribution, retail, manufacturing, wholesale. Wherever your stock decisions live — spreadsheets, documents, an ERP — the template is the first thing we build with you.
Next
Why it holds up
Built to survive the follow-up question.
No integration project.
It reads the files a distributor already keeps. An ERP connector exists, but it is not a prerequisite.
Answers, not dashboards.
The output is a prescription with a number attached, not a chart to interpret.
It shows its work.
Every answer cites the underlying rows, and forecasts carry calibrated prediction intervals rather than bare point estimates.
Your raw data never leaves.
Figures are aggregated and PII-redacted locally before any model call.
Questions we get
The things people ask on the first call.
Our data is a mess. Is that a problem?
No. Messy is the normal starting point, and we can work with the messiest version of it. There is no cleanup project to complete before you get anything back — if there were, it would just be another reason the work never starts.
Do we need an ERP?
No. Purpigal reads what you already keep, whether that is loose spreadsheets and documents or the same records sitting inside an ERP. A connector exists for the companies that run one, but it has never been a prerequisite.
Where does our data actually live?
That is your choice. Cloud deployment keeps it in secured cloud storage. Local deployment puts it on a single machine in our own offices, and nothing leaves that machine — where that machine sits is something we agree with you before any data moves. Either way, figures are aggregated and PII-redacted before any model call.
What does it actually tell us?
Most of what people ask for day to day is demand forecasting, restocking recommendations and expiry alerts. Beyond those you can ask it about the business in plain language and get an answer grounded in your own records rather than a general one.
What does it cost?
It depends on what your business needs, so we quote after the first call rather than publishing a number that would be wrong for most companies. The call costs nothing.
Contact
Talk to us.
We are onboarding design partners for paid pilots. If you run distribution and keep your history in spreadsheets, documents, an ERP — anywhere at all — we would like to see it.
We usually reply within a working day.
Or email contact@purpigal.com