One operational problem, one AI system, accepted against numbers you set
For chemical, manufacturing and construction companies in the UAE and Saudi Arabia. One team from diagnosis to acceptance.
84% of GCC companies surveyed use AI; 11% get 5% or more of earnings from it
use AI in at least one business function
have scaled AI across the company
get 5% or more of earnings from AI. McKinsey calls them "value realisers"
Source: McKinsey & Company with the GCC Board Directors Institute, The state of AI in GCC countries: In pursuit of scale and value, 2025. Online survey of 139 senior executives and board directors.
Industrial AI projects stall for three reasons: wrong problem, unready data, vendor-led design
The wrong first problem
The pilot is picked because it demos well, not because it cuts cost, frees cash or adds capacity. Symptom: the pilot never leaves the IT team.
Data that is not ready
Specifications, SOPs and price history sit in PDFs, email and the ERP, and nobody owns them. The model reads old or conflicting versions. Symptom: answers nobody trusts.
Design shaped by the vendor's product
The supplier fits your problem to its licence. You buy the licence, users do not open it, nobody measures the business case. Symptom: licences nobody uses.
Three of six use cases run on data you already hold: no sensors, no new platform
| Use case | Sectors | What it improves | Data it needs | Who builds |
|---|---|---|---|---|
| Technical document assistant | Chemicals, oil and gas services, manufacturing | Time to answer, fewer shift errors | SOPs, MSDS, manuals | We build |
| Tender and RFP assistant | Construction, EPC, engineering | Bid preparation time | Past bids, standard clauses | We build |
| Quotation and pricing assistant | Chemicals, lubricants, cables | Quote turnaround, margin | ERP prices, past quotes | We build, partner integrates |
| Procurement and contract review | All | Cycle time, missed clauses | Contracts, supplier master | We build, partner integrates |
| Quality inspection by camera | Manufacturing, cables | Defect rate, scrap | Existing line cameras | Partner builds, we lead |
| Predictive maintenance | Oil and gas, chemicals | Unplanned downtime | Historian, failure records | Platform partner, we lead |
Highlighted rows run on your documents and ERP data. A pilot takes 8 to 12 weeks.
Two systems Alexander built and runs, shown live in 20 minutes
Search and answer over a private archive
Runs daily over about 978 000 messages, voice notes and scanned images. The same design answers questions from your SOPs, MSDS sheets, manuals and contracts and cites the source page.
An agent that can act only through approved actions
Configures network equipment through 37 controlled tools. Every action is logged and checked by automated tests: 164 of 164 tool checks passed on 7 October 2026. Inside your systems it can do only what you approve.
Demonstrations, not client references.
A go or no-go in 3 weeks, then an accepted system 8 to 12 weeks later
Diagnose and design, 3 weeks
We rank your use cases, check the data, draw the architecture and get a fixed partner price. You decide go or no-go before any build money is spent.
Build, 8 to 12 weeks
A named partner builds. We run the design reviews and report weekly against plan and budget. Your team owns the data and the users.
Accept
Your process owner runs the acceptance test on real cases, against the baseline and target written in the proposal.
Review at 30 and 90 days
We measure the KPIs again after go-live and recommend scale or stop.
Andrei brings 12 years of Gulf industrial sales; Alexander designs and builds the AI
Alexander Kuznetsov
- Designs and builds AI assistants and agents himself; both demonstrations above are his
- In technology since 2008
- Co-founder and former CEO of an industrial packaging company: turnover 4x in four years, operating costs -32%
- In your project: owns the architecture, the partner choice and the acceptance test
Andrei Saenko
- 27 years in commercial leadership, including BASF and Henkel; 12 years in the Middle East
- BASF Middle East, Sales Director: EUR 50 million revenue, profitability up 12%
- Opened accounts at Saudi Aramco, Ma'aden, Alfanar and Alzamil Group
- In your project: owns the business case and takes the go or no-go decision with your sponsor
You see every partner, every price and every fee we receive before you sign
- Every delivery partner we propose is named in the proposal, with its price, before you sign.
- If a partner pays us a fee, the proposal shows who pays and how much. We take no other payments.
- You may choose a different partner. Our architecture and oversight fee does not change.
- We never audit a partner we brought. If you want an independent check, we name another firm.
- If you prefer, we select the partner for a fixed fee and take nothing from the vendor.
Three steps from briefing to accepted system, plus team training
Board AI briefing
Three candidate problems and a next-step note. Credited in full against the diagnostic.
AI diagnostic
Ranked use cases, data check, priced pilot, go or no-go.
Pilot: design, deliver, accept
One system in one site or function, accepted against the KPIs in the proposal.
Team enablement
15 to 25 people learn to use AI on their own tasks, with a 30-day practice plan.
Every offer has a fixed fee, written in the proposal before work starts.
Send one operational problem. Within five working days you get a fixed-price diagnostic proposal.
One paragraph: the problem, who owns it, and the limit (budget, date or system).