Drinkwell

Proposed for the World Economic Forum Annual Meeting 2027 · Schwab Foundation for Social Entrepreneurship

The Bengal Basin Blueprint

AI for Climate Impact

The first water system in South Asia where local people help build and audit the AI that runs it. Public money builds, companies co-finance, communities grade the water, and AI keeps the record straight.

It starts in Narayanganj, on the Shitalakshya and Meghna rivers, and is designed to repeat across the Bengal Basin, the delta aquifer that Bangladesh and West Bengal share and home to an estimated 278 million people, where Drinkwell already works on both sides of the border.

Progress page for Schwab Foundation staff · updated 24 September 2026

Where we stand

700+treatment plants powered by Drinkwell in Bangladesh and India
2.5B+litres of safe water processed
~280Water ATMs run in partnership with Dhaka WASA, renewed to February 2028
US$115.8MADB loan to Narayanganj City Corporation, with a Water ATM planned for every ward
4 yearsGap Inc. partnership for Water ATMs in Bangladesh, approved April 2026

Drinkwell is a materials science company. Our proprietary resin is regenerated instead of thrown away, and it runs inside connected machines in public spaces. That is physical AI, adopted by the public sector.

The chain that runs back to Davos

At Davos in 2026 Minhaj Chowdhury argued that South Asia's water problem is not a shortage of money but a utilization problem: the region budgets about US$20 billion a year for water and spends roughly half, and what gets built often runs below capacity. The answer is to raise utilization by pairing utilities with the companies that operate in their cities. Here is what followed.

  1. US Secretary of State's Award for Corporate Excellence. Drinkwell and Gap Inc. are fellow awardees, and the first conversation about Water ATMs near Gap's suppliers begins.
  2. ADB pilot. With support from ADB, Drinkwell installs two pilot Water ATMs in Narayanganj under the project's readiness financing.
  3. ECNEC approves the Narayanganj Green and Resilient Urban Development Project, which scales the Water ATM model into a public investment.
  4. Schwab Foundation Social Entrepreneur award, Davos. Gap's leadership attends the ceremony. Minhaj meets Gap's CEO the next day.
  5. Dhaka WASA renews the Water ATM partnership for two more years, to February 2028.
  6. Gap Inc. approves a four-year partnership for Water ATMs in Bangladesh under its 2030 water goals, with year-one rollout from June 2026.
  7. ADB approves the US$115.8 million loan to Narayanganj City Corporation. The city builds a shed in each ward; Drinkwell supplies and runs the Water ATM, and Gap's partnership is the private co-financing that makes Drinkwell's side affordable.
  8. Drinkwell on the cover of the Forum's report AI for Impact: The Role of Artificial Intelligence in Social Innovation in Asia, with EY, Microsoft and SAP.
  9. AI and Social Innovation workshop, Geneva, with the Schwab Foundation, and WaterGrade launches: households grade their water from A to F.
  10. Sustainable Development Impact Meetings, New York. Collective Action to Shape the Future of Social Innovation and AI. Preliminary conversations with SAP and EY.

Where the Blueprint starts

Every active Drinkwell site in Bangladesh and India, and the Narayanganj layer: the pilot Water ATM, the sites shortlisted for the ward rollout, and the low-income communities they are meant to serve.

  • Active site, Bangladesh
  • Active site, India
  • Narayanganj pilot Water ATM
  • Shortlisted ward site
  • Low-income community

The Blueprint, part by part

In development

1. An open-weight, sovereign Bangla water model

A small AI model trained by the people closest to the water: community-based organisations with decades of field experience, households, young Watergraders and Water ATM operators, in Bangla and its dialects. How they train it. The weights stay in Bangladesh with the public partner, so the city owns its water intelligence instead of renting it. It builds on the national Bangla language model called for in Bangladesh's draft National AI Policy.

Building

2. A public ledger anyone can audit

Every Water ATM in the Narayanganj programme and the Gap partnership traced in one open record: what it cost, the litres it delivered, when its filter was regenerated, and an A to F grade set by the people who drink the water. WaterGrade, the grading layer, launched in July 2026 and is in private beta.

Live

3. Gap Inc. as champion

A four-year partnership, with year-one Water ATMs rolling out in Dhaka from June 2026 and sited by need rather than by factory. At Davos, Gap would convene apparel brands and suppliers to fund water where garment workers live and to adopt the same open grade.

In conversation

4. Founding partners from the Corporate Leadership Council

SAP has supported Drinkwell's leadership team pro bono since 2022, and EY has backed Drinkwell through EY Ripples, its TRANSFORM upskilling workshops and World Entrepreneur Of The Year 2026. Subject to the Schwab Foundation enabling this announcement to take place at Davos, SAP and EY would be keen to support, based on preliminary conversations held during the Sustainable Development Impact Meetings 2026.

How local people train the model

Foundation models learn from text on the internet. The people who know most about water in the Bengal Basin rarely write there: they speak Bangla and its dialects, they keep records on paper, and many of them are young. The Blueprint's model is trained by them, on data they consent to share, and judged by them before it is used.

Who trains it

Community-based organisationsGroups rooted in their own villages and neighbourhoods that have worked on arsenic, salinity and safe water in the region for decades bring what no dataset holds: well-testing records, field reports and hard-won judgement about what fails and why. Their members also check every label before it reaches the model.
HouseholdsFamilies who drink the water grade it from A to F and leave voice notes in their own words and dialect about taste, colour, queues, breakdowns and cost. These are the voices the grade is built on.
YouthYoung people aged 18 to 35, who are rarely represented in foundation model training data, train as certified Watergraders. They collect reports, transcribe and translate dialect into standard Bangla, label the data, and are paid for it.
OperatorsThe women and men who run Water ATMs every day log faults, filter regenerations and water tests. That turns daily operating experience into training data.

Six steps from voice to model

  1. Consent first. Every contributor agrees to how their data is used and can withdraw it. The data sits in Bangladesh under a community data trust, governed by the city, community-based organisations, youth representatives, and Drinkwell. The data trust follows principles drawn from Indigenous data sovereignty, the CARE Principles: collective benefit, authority to control, responsibility and ethics.
  2. Collect in local languages. Households and operators contribute voice notes, photos and grades through WaterGrade and the Water ATMs, in Bangla and its dialects. Community-based organisations contribute their records, with permission.
  3. Label with two keys. Youth Watergraders transcribe, translate and label. Members of community-based organisations validate. A record only enters the training set when both have signed off.
  4. Fine-tune a small open model. We start from an open model that already reads Bengali, then fine-tune it on this data. It is small enough to run on modest hardware in Bangladesh. The weights are released under an open licence, with a model card that credits every contributing group.
  5. Community evaluation. Community-based organisations, youth and households write the test questions and score the answers. The model is only used once it passes their benchmark, not just ours.
  6. Trace every decision. Each training record carries its provenance: who, when and under what consent. Every model output that triggers an action, such as a repair, a filter regeneration or a water test, is logged to the public ledger, where households' grades show whether it was right.

Guardrails: the model advises and people decide. It never declares a source safe to drink without a laboratory test, and it holds no names or personal details. Contributors are credited, and part of the Blueprint's funding pays for their time.

Why Davos, and why now

The first Annual Meeting after the UN Water Conference. Blue Davos opened the Forum's Year of Water in 2026 and called for fit-for-purpose finance and basin partnerships, noting that the private sector provides only 2 to 3 percent of water investment. The Blueprint is a basin partnership with private co-financing built in.

Local voices in the model. The Forum's own AI for Impact work argues that social innovators should be frontline architects of AI. Here, the people who drink the water write the grades that train and audit the system.

A partnership Davos can claim. The award put Drinkwell in the room with Gap's leadership; Gap's partnership co-finances Drinkwell's side of ADB's Narayanganj programme. Davos 2027 is the moment to multiply it by ten.

Partners

Gap Inc.Four-year partner and proposed champion
Asian Development BankNarayanganj Green and Resilient Urban Development Project
Narayanganj City CorporationPublic owner of the ward Water ATMs
Dhaka WASAWater ATM partner, approximately 280 booths across Dhaka
SAP and EYCorporate Leadership Council members; would be keen to support, subject to the Schwab Foundation enabling the announcement at Davos