Mendee AI Station turns Kazakhstan's 24 km³/year of untapped brackish groundwater into clean drinking water using AI-optimized batch reverse osmosis. Solar-powered, off-grid, 5–7× less energy than seawater desalination.
until Kazakhstan's projected 50% water deficit — UNDP
Rivers
Kazakhstan's three major transboundary rivers are losing flow — and the country controls none of them upstream.
Kyrgyzstan → Uzbekistan → Kazakhstan
27 km³/yr → 17 km³/yr
Feeds the Aral Sea basin. 1 million people and 700,000 hectares of irrigation depend on this river.
World Bank 2024
China (Xinjiang) → Kazakhstan
17.8 km³/yr → 12.7 km³/yr
Lake Balkhash receives 70% of its water from the Ili. China plans to irrigate 600,000 ha by 2050, further reducing flow.
CA Waters · Astana Times 2025
China → Kazakhstan → Russia
Upstream diversions growing → Astana water supply threatened
Supplies Astana via the Irtysh–Karaganda canal. 3 hydroelectric plants at risk. The capital's drinking water depends on this flow.
World Bank · ScienceDirect 2022
Problem
Surface flows declining. Glaciers melting. Transboundary rivers diverted upstream. By 2040, the country faces a 50% water deficit.
50%
Projected water deficit by 2040 (UNDP)
−38%
Decline in Syr Darya transboundary inflow
~0%
Current use of brackish groundwater reserves
Transboundary river inflow dropped from 44 to 34 km³/year. Upstream neighbours are taking more. Glaciers losing 1% volume annually.
Only 343 of 494 fresh aquifers are commercially exploited; 50% concentrated in the south. The Syr Darya basin faces 90–130 m drawdown by 2040.
By 2035: 24.5 km³ demand vs declining supply. 6.8 km³/year deficit projected. Current infrastructure covers only 42% of technical capacity.
Solution
24 km³/year of brackish groundwater — virtually unused. Mendee AI Station turns this "useless" water into drinking water at 5–7× less energy than seawater desalination.
Machine learning models continuously optimise membrane pressure, flow rates, and energy consumption based on real-time water chemistry — 30%+ energy savings vs static configuration.
Each 50 m³/day system runs on solar PV with LiFePO₄ battery storage. No diesel dependency. ~20–30 tonnes CO₂ saved per unit per year.
1–10 g/L salinity vs 35 g/L for seawater means 5–7× less energy, simpler membranes, and dramatically lower cost per cubic metre.
Brine is not waste — it's raw material. Lithium ($15–20K/tonne), magnesium, potassium and salt extracted as additional revenue streams.
Brackish vs Fresh vs Seawater
| Fresh GW | Brackish GW | Seawater | |
|---|---|---|---|
| Reserves | 34 km³/yr | 24 km³/yr | Unlimited |
| Current usage | 17% | ~0% | N/A inland |
| Energy to treat | None | 5–7× less | Baseline |
| Available where | South & mountains | Everywhere (steppe) | Coastlines only |
| Competition | High (drinking) | Zero | High |
Technology
AI-optimized batch reverse osmosis with real-time water chemistry sensing, predictive maintenance, and autonomous operation — engineered for the Central Asian steppe.
Roadmap
B2B first. Prove on industrial clients, then expand to farms and government contracts.
Business model
Hardware sales bootstrap revenue. SaaS, services, minerals, carbon credits and data turn it into a platform business. By Year 5, target £21–23M revenue at ~72% gross margin.
| Revenue stream | Year 1 | Year 5 | Margin |
|---|---|---|---|
| Hardware (desalination) | £1.0M | £10.0M | 60% |
| SaaS subscription | £120K | £4.8M | 90% |
| Service & Maintenance | £50K | £2.0M | 70% |
| Smart Irrigation | — | £2.5M | 65% |
| Brine minerals | — | £0.5–1M | 80% |
| Carbon credits | — | £0.2–0.3M | 95% |
| Data sales | — | £0.3–0.5M | 90% |
| Total | £1.17M | £21–23M | ~72% |
Hardware vs recurring revenue mix
Scale check
Even covering Kazakhstan's entire 2035 water deficit would consume only 28% of its brackish reserves — well within sustainable extraction rates.
0.008%
Of brackish reserves used at Year 1 pilot scale
28%
Of reserves needed to cover 2035 deficit entirely
80
Systems targeted by Year 5
£150–250M
Target company valuation at Year 5
Team
Mendee AI Station is at the earliest stage — one founder, a clear plan, and the search for the right people to execute it.
Alex Choysen
Founder & CEO
Business strategy, investor relations, go-to-market. Background in digital product development. UK-based, Kazakhstan roots.
CTO — recruiting
Chief Technology Officer
ML/AI optimization, embedded systems, water treatment engineering. Part-time initially (£3K/mo), full-time from Year 1.
Actively recruitingResearch advisors
Advisory & Science
Building the advisory board through research collaborations with Nazarbayev University (KZ) and UK water-engineering departments.
Building partnershipsTarget advisory institutions
We're forming our advisory board through research collaborations. Scientists who co-publish with us become natural advisors.
Nazarbayev University
Hydrogeology, Environmental Engineering
Cranfield University
Water Science Institute
Imperial College London
Civil & Environmental Engineering
Al-Farabi Kazakh National University
Geography & Environmental Sciences
UCL
Bartlett School (Environment, Energy & Resources)
KBTU
School of Chemical Engineering
Pilot programme
Mendee AI Station is looking for an industrial facility near Astana for a 3-month pilot deployment. We provide the system — you provide the site and water access.
Selection Q2 2026 · [email protected]
Researchers
Collaborate with Mendee AI Station on field data, fieldwork access, and joint publications
Get involved
Mendee AI Station is the water-technology arm of Mendee Digital Ltd (Co. 14071822) — same legal entity behind PRVD.LDN translation services and svoi.uk. We talk to impact investors, water-tech VCs, academic departments, hardware partners, and Kazakhstan-based potential pilot hosts. Operations split between London (IP & HQ) and Astana (production / pilots).