AI4CR - Progress REPORT
As part of Klarna's AI for Climate Resilience Program, six organizations are building AI tools to prepare for, withstand and recover from our changing climate. Here is where they are after their first months in the field.
Milkywire x Klarna, OCT 15, 2026
AI4CR - Progress REPORT
Six months of climate action
As part of Klarna's AI for Climate Resilience Program, six organizations are building AI tools to prepare for, withstand and recover from our changing climate. Here is where they are after their first months in the field.
Launched by Klarna and Milkywire in January 2026, the AI for Climate Resilience Program supports local innovators working with communities affected by climate change.
As an AI-powered company, Klarna is backing local innovators leveraging AI to boost communities on the frontlines of climate change. Through the AI for Climate Resilience Program (AI4CR), we provide funding, mentoring, and practical support to help organizations and projects turn their ideas into useful tools that can help people prepare for climate risks, protect livelihoods, and manage resources more effectively.
We are only a few months in, but we are already seeing signs of progress. This report looks at what has changed, what is beginning to work, and what we have so far learned about using AI for climate resilience.
Sholto West
Program Manager at Milkywire
Launched by Klarna and Milkywire in January 2026, the AI for Climate Resilience Program supports local innovators working with communities affected by climate change.
As an AI-powered company, Klarna is backing local innovators leveraging AI to boost communities on the frontlines of climate change. Through the AI for Climate Resilience Program (AI4CR), we provide funding, mentoring, and practical support to help organizations and projects turn their ideas into useful tools that can help people prepare for climate risks, protect livelihoods, and manage resources more effectively.
We are only a few months in, but we are already seeing signs of progress. This report looks at what has changed, what is beginning to work, and what we have so far learned about using AI for climate resilience.
Sholto West
Program Manager at Milkywire
Launched by Klarna and Milkywire in January 2026, the AI for Climate Resilience Program supports local innovators working with communities affected by climate change.
As an AI-powered company, Klarna is backing local innovators leveraging AI to boost communities on the frontlines of climate change. Through the AI for Climate Resilience Program (AI4CR), we provide funding, mentoring, and practical support to help organizations and projects turn their ideas into useful tools that can help people prepare for climate risks, protect livelihoods, and manage resources more effectively.
We are only a few months in, but we are already seeing signs of progress. This report looks at what has changed, what is beginning to work, and what we have so far learned about using AI for climate resilience.
Sholto West
Program Manager at Milkywire
Chapter I: THE program at a glance
About the program
Real impact happens when ideas meet action
The AI for Climate Resilience program promotes pioneering solutions that empower communities with technology to lead the acceleration of system-level climate resilience. The selection process is based on a rigorous multi-stage evaluation, developed in partnership with Milkywire’s independent advisory board, to identify where AI-powered solutions can drive the greatest real-world impact. In our first cohort alone, we received 1,200+ applications from for- and non-profits across every continent. Read more about the program here.
Finally, six projects were selected: Acres of Ice (India), GainForest (Brazil), Geotek Water Solutions (Nigeria), Saka Warga (Indonesia), SEEDS (India) and SEWA / IFPRI / Gram Vaani (India). These projects harness the power of AI for climate resilience and adaptation, turning great ideas into real impact.
Real impact happens when ideas meet action
The AI4CR program promotes pioneering solutions that empower communities with technology to lead the acceleration of system-level climate resilience. The selection process is based on a rigorous multi-stage evaluation, developed in partnership with Milkywire’s independent advisory board, to identify where AI-powered solutions can drive the greatest real-world impact. In our first cohort alone, we received +1,200 applications from for- and non-profits across every continent. Read more about the program here.
Finally, six projects were selected: Acres of Ice (India), GainForest (Brazil), Geotek Water Solutions (Nigeria), Saka Warga Foundation (Indonesia), SEEDS (India) and SEWA / IFPRI / Gram Vaani (India). These projects harness the power of AI for climate resilience and adaptation, turning great ideas into real impact.
Content
~
$
Million in total funding
+
hours of mentoring
Countries
~
$
Million in total funding
+
hours of mentoring
Countries
Clear objectives to measure what matters
We have defined five objectives to track how the AI tools are building toward lasting climate resilience: Adoption, Empowerment, AI-added value, Impact, and System capacity and integration. These are important indicators of whether the work can lead to lasting impact over time.
01.
Adoption
Highlight
10,000+ biodiversity observations logged through GainForest’s tools. This is an early sign that the technology is moving beyond development into real-world use by local participants. It also creates the data foundation the wider conservation model depends on.
02.
Empowerment
Highlight
Through four months of community workshops and testing, Saka Warga has evolved its AI Resilience Coach into a more interactive preparedness platform designed to better support behaviour change.
This includes adding AI-assisted task breakdowns, reminders, village progress tracking, peer learning and regular activation sessions to help communities organize, practise and sustain preparedness between disasters.
03.
AI-added value
Highlight
Acres of Ice has moved its camera-based AI volume-estimation model into active validation against drone surveys.
If validated, a roughly $400 camera set-up could replace drone monitoring costing around $200 per visit, while shifting monitoring from a few seasonal measurements to continuous daily data.
04.
Impact
Highlight
GeoTek has completed and commissioned three new groundwater systems, each capable of serving up to 15,000 people, and fitted them with remote monitoring to track performance and help maintain reliable water access over time.
05.
System capacity & integration
Highlight
SEEDS is designing its AI-enabled platform for integration into existing government systems, while embedding implementation in National and District Field Committees with government participation.
The intervention is being designed around the institutions that would ultimately need to use and maintain it, creating a more credible pathway from pilot to sustained public-service delivery.
Clear objectives to measure what matters
We have defined five objectives to track how the AI tools are building toward lasting climate resilience: Adoption, Empowerment, AI-added value, Impact, and System capacity and integration. These are important indicators of whether the work can lead to lasting impact over time.
01.
Adoption
A technically strong tool only creates value if people actually use it. Early adoption tells us whether the solution fits real needs, behaviours and operating conditions.
Highlight
1,500+ biodiversity observations logged through GainForest’s tools. This is an early sign that the technology is moving beyond development into real-world use by local participants. It also creates the data foundation the wider conservation model depends on.
02.
Empowerment
Resilience does not come from the tool itself, but from strengthening communities’ own ability to organise, make decisions and act for themselves.
Highlight
Through four months of community workshops and testing, Saka Warga has evolved its AI Resilience Coach into a more interactive preparedness platform designed to better support behaviour change.
This includes adding AI-assisted task breakdowns, reminders, village progress tracking, peer learning and regular activation sessions to help communities organize, practise and sustain preparedness between disasters.
03.
AI-added value
AI should enable something materially better than the existing approach — greater scale, lower cost, better decisions or capabilities that were previously impractical.
Highlight
Acres of Ice has moved its camera-based AI volume-estimation model into active validation against drone surveys.
If validated, a roughly $400 camera set-up could replace drone monitoring costing around $200 per visit, while shifting monitoring from a few seasonal measurements to continuous daily data.
04.
Impact
Ultimately, the technology has to translate into improved resilience for people facing climate risks.
Highlight
Geotek has completed and commissioned three new groundwater systems, each capable of serving up to 15,000 people, and fitted them with remote monitoring to track performance and help maintain reliable water access over time.
05.
System capacity & integration
A pilot is more likely to last and scale when it strengthens existing institutions rather than creating a parallel system that disappears when project funding ends.
Highlight
SEEDS is designing its AI-enabled platform for integration into existing government systems, while embedding implementation in National and District Field Committees with government participation.
The intervention is being designed around the institutions that would ultimately need to use and maintain it, creating a more credible pathway from pilot to sustained public-service delivery.
Specialist support to tackle complex challenges
Alongside funding, the projects receive specialist mentoring, practical training, and opportunities to learn from one another. This helps them work through technical and real-world challenges as they develop their tools, and make stronger decisions about what comes next.
01.
One-to-one mentoring
Connecting teams with expertise they may not have in-house, from AI architecture and data engineering to climate science, community governance, and fundraising.
02.
Seminars & masterclasses
Addressing challenges shared across the cohort
03.
Community of practice
Offering teams a place to exchange ideas and learn from each other’s experience.
Specialist support to tackle complex challenges
Alongside funding, the projects receive specialist mentoring, practical training, and opportunities to learn from one another. This helps them work through technical and real-world challenges as they develop their tools, and make stronger decisions about what comes next.
01.
One-to-one mentoring
Connecting teams with expertise they may not have in-house, from AI architecture and data engineering to climate science, community governance, and fundraising.
02.
Seminars & masterclasses
Addressing challenges shared across the cohort
03.
Community of practice
Offering teams a place to exchange ideas and learn from each other’s experience.
Oliver Paul
AI & systems-design mentor
Oliver Paul brings experience in machine learning research, data science leadership, and building AI startups as a co-founder and founding engineer. Within the program, he has supported two cohort organizations with AI architecture and technical strategy. He helps teams work through technical challenges and identify where stronger data and systems are needed before more advanced AI can add value.
“Technical guidance from our mentor strengthened our thinking around AI architecture, data engineering, system design, responsible AI, and long-term technology development.”
Geotek Water Solutions Ltd.
CHAPTER II: From ideas to implementation
Progress & learnings
Early signs of real progress
During the initial months, the program has mainly focused on building the foundations for success: reliable data, trusted relationships, working products and pathways into institutions and communities. These foundations are now taking shape, with progress across all five objectives: AI tools moving into use, creating new value, and becoming embedded in the systems and communities around them.
India
Acres of Ice
Project goal
Expand the model into two new countries, Nepal and Kyrgyzstan, providing more than 200 households with more reliable spring irrigation water and testing how the approach can work in different mountain environments.
Progress
•
Field surveys completed, sites selected, and local agreements secured for the first international deployments in Nepal and Kyrgyzstan.
•
Construction completed at one reservoir site in Nepal and two in Kyrgyzstan; operations expected to begin in the coming months.
•
Camera-based AI model for measuring stored ice volumes undergoing validation against drone surveys to reduce monitoring costs.
India
SEWA / IFPRI / Gram Vaani
Project goal
Reach more than 50,000 women across 22 districts with locally relevant climate advice, helping them prepare for climate shocks, protect their health and livelihoods, and improve access to finance.
Progress
•
Knowledge base for the AI advisory service nearly complete, covering climate, health, finance, and livelihoods.
•
Needs and working conditions mapped across four groups: construction workers, farmers, vendors, and home-based workers.
•
More than 80 SEWA members participated in the initial needs assessment.
Indonesia
Saka Warga
Project goal
Test the coach in seven pilot villages and train 150 volunteers. Work toward integrating it into disaster-preparedness systems such as the Indonesian Red Cross, while using AtmaGO, a platform with over 16 million users, to encourage wider community participation and share preparedness knowledge.
Progress
•
Community outreach and feedback expanded the original AI coach concept into a six-component preparedness platform.
•
Platform development completed to the point that it is now ready for phased rollout across seven pilot villages.
Brazil
GainForest
Project goal
Support at least 10 local communities to collect, store, and own more than 10,000 biodiversity observations and audio recordings. Use AI to classify and map this data, supporting impact certifications that can help attract conservation funding.
Progress
•
App developed to support data collection without an internet connection.
•
+10,000 biodiversity observations recorded.
•
Local community members now paid to collect data.
Nigeria
GeoTek Water Solutions Ltd
Project goal
Deploy 50 sensors across existing and newly constructed drinking-water facilities. Use the resulting data to improve monitoring, automate aspects of water provision, and guide well-drilling decisions for communities vulnerable to climate impacts.
Progress
•
Three of five planned water systems completed and brought into operation.
•
25 monitoring devices installed and transmitting live data.
India
SEEDS
Project goal
Demonstrate a model for disaster support that can be expanded to other communities by enrolling 50,000 households across five districts in three Indian states. Each household will have an e-wallet linked to eligible benefits.
Progress
•
Five pilot districts selected.
•
Local risks identified and prioritized in three of the five districts.
•
More than 27,000 households surveyed to identify vulnerable families and guide support before and after disasters.
Early signs of real progress
During the initial months, the program has mainly focused on building foundations for success: securing reliable data, building trusted relationships, products and pathways into institutions and communities. Less than a year in, those foundations are taking shape and we are excited to see progress being made across all five objectives: AI tools moving into use, creating new value, and becoming embedded in the systems and communities around them.
India
Acres of Ice
Acres of Ice helps mountain communities adapt to shrinking glaciers by storing excess winter water in artificial ice reservoirs for spring irrigation. AI is used to estimate stored ice volumes from low-cost camera images, reducing manual monitoring. AI is also used in the project’s planned development to identify suitable sites and automate reservoir operations.
Project goal
Expand the model into two new countries, Nepal and Kyrgyzstan, providing more than 200 households with more reliable spring irrigation water and testing how the approach can work in different mountain environments.
Progress
•
Field surveys completed, sites selected, and local agreements secured for the first international deployments in Nepal and Kyrgyzstan.
•
Construction completed at one reservoir site in Nepal and two in Kyrgyzstan; operations expected to begin in the coming months.
•
Camera-based AI model for measuring stored ice volumes undergoing validation against drone surveys to reduce monitoring costs.
India
SEWA / IFPRI / Gram Vaani
SEWA, IFPRI, and Gram Vaani are helping rural women workers respond to extreme heat, floods, and unpredictable weather. AI is used to combine weather data with local feedback and deliver practical advice in local languages through WhatsApp. By explaining what forecasts mean for everyday work and health, the service aims to help women protect their livelihoods and prepare for climate shocks.
Project goal
Reach more than 50,000 women across 22 districts with locally relevant climate advice, helping them prepare for climate shocks, protect their health and livelihoods, and improve access to finance.
Progress
•
Knowledge base for the AI advisory service nearly complete, covering climate, health, finance, and livelihoods.
•
Needs and working conditions mapped across four groups: construction workers, farmers, vendors, and home-based workers.
•
More than 80 SEWA members participated in the initial needs assessment.
Indonesia
Saka Warga
Saka Warga is helping rural communities prepare for floods, landslides, and storms. AI is used to provide a conversational coach that helps local volunteers understand hazards, map resources, and prepare disaster drills through voice or text in their own language. The tool offers practical guidance between formal training sessions, helping communities organize before a crisis.
Project goal
Test the coach in seven pilot villages and train 150 volunteers. Work toward integrating it into disaster-preparedness systems such as the Indonesian Red Cross, while using AtmaGO, a platform with over 16 million users, to encourage wider community participation and share preparedness knowledge.
Progress
•
Community outreach and feedback expanded the original AI coach concept into a six-component preparedness platform.
•
Platform development completed to the point that it is now ready for phased rollout across seven pilot villages.
Brazil
GainForest
GainForest helps Indigenous peoples and local communities document their conservation work and access funding while retaining ownership of their environmental data. AI is used within its Tainá platform to organize images, audio recordings, and traditional knowledge, classify biodiversity observations, and link them to locations. This evidence supports impact reports and certifications that can make community-led conservation more visible to funders.
Project goal
Support at least 10 local communities to collect, store, and own more than 10,000 biodiversity observations and audio recordings. Use AI to classify and map this data, supporting impact certifications that can help attract conservation funding.
Progress
•
App developed to support data collection without an internet connection.
•
+10,000 biodiversity observations recorded.
•
Local community members now paid to collect data.
Nigeria
Geotek Water Solutions Ltd
Geotek works to improve drinking-water reliability for rural and peri-urban communities facing climate variability. AI is used to combine satellite imagery, geological maps, community reports, and sensor readings to identify water-quality risks and anticipate pump failures. These insights help guide monitoring, maintenance, and decisions about where to drill new wells for communities vulnerable to climate impacts.
Project goal
Deploy 50 sensors across existing and newly constructed drinking-water facilities. Use the resulting data to improve monitoring, automate aspects of water provision, and guide well-drilling decisions for communities vulnerable to climate impacts.
Progress
•
Three of five planned water systems completed and brought into operation.
•
25 monitoring devices installed and transmitting live data.
India
SEEDS
SEEDS is working to help disaster-affected families access financial support more quickly and fairly. AI is used to check household damage reports and photographs against satellite and weather data to help verify losses. Verified cases are linked to eligible government and donor support, aiming to reduce delays in compensation and enable payments directly to families’ accounts.
Project goal
Demonstrate a model for disaster support that can be expanded to other communities by enrolling 50,000 households across five districts in three Indian states. Each household will have an e-wallet linked to eligible benefits.
Progress
•
Five pilot districts selected.
•
Local risks identified and prioritized in three of the five districts.
•
More than 27,000 households surveyed to identify vulnerable families and guide support before and after disasters.
Three key lessons
Across very different projects and contexts, the first six months are surfacing recurring lessons about what it actually takes to build useful AI for climate resilience.
Three key lessons
Across very different projects and contexts, the first six months are surfacing recurring lessons about what it actually takes to build useful AI for climate resilience.
01.
Hardware and data quality come before AI.
Across the cohort, much of the first six months has been spent building the physical and data foundations AI depends on: deploying and improving sensors, establishing data streams, preparing reliable databases, adapting technology for low-connectivity settings, and collecting large volumes of household and climate data.
This work is labour-intensive and depends on organizations with deep local relationships, able to generate data that reflects people’s real circumstances and vulnerabilities. Without reliable hardware and high-quality, locally relevant data, the AI layer has little to work with.
01.
Hardware and data quality come before AI.
Across the cohort, much of the first six months has been spent building the physical and data foundations AI depends on: deploying and improving sensors, establishing data streams, preparing reliable databases, adapting technology for low-connectivity settings, and collecting large volumes of household and climate data.
This work is labour-intensive and depends on organizations with deep local relationships, able to generate data that reflects people’s real circumstances and vulnerabilities. Without reliable hardware and high-quality, locally relevant data, the AI layer has little to work with.
“The most significant learning was that the primary challenge was not developing an AI assistant capable of answering preparedness questions, but determining how AI could meaningfully support long-term preparedness behaviours.”
Saka warga
02.
Information alone doesn’t change behaviour.
AI can make climate information more accessible and relevant, but lasting behaviour change depends on more than information alone. Across the cohort, projects are combining guidance with practical tasks, reminders, opportunities to practise, and social reinforcement to help people build new routines and act more consistently in response to climate risks.
02.
Information alone doesn’t change behaviour.
AI can make climate information more accessible and relevant, but lasting behaviour change depends on more than information alone. Across the cohort, projects are combining guidance with practical tasks, reminders and reinforcement, opportunities to practise, and social reinforcement to help people build new routines and act more consistently in response to climate risks.
“The key shift in our thinking was realising that communities do not have to choose between control of their data and ease of use. They can decide who has access to their data, revoke access when they choose, export it, or store it locally.”
Gainforest
03.
Product development and usability are becoming impact capabilities.
For many public-benefit organizations, building AI means developing a new kind of capability: product development. Teams are having to learn how to design, test and iterate digital tools around real user needs — navigating connectivity, language, digital literacy, incentives and local context — while still delivering on their core climate or social mission.
03.
Product development and usability are becoming impact capabilities.
For many public-benefit organizations, building AI means developing a new kind of capability: product development. Teams are having to learn how to design, test and iterate digital tools around real user needs — navigating connectivity, language, digital literacy, incentives and local context — while still delivering on their core climate or social mission.
CHAPTER III: LOOKING AHEAD
What comes next
The next phase will test whether these foundations translate into wider adoption and measurable resilience outcomes as tools move further into deployment, validation and use.
We will continue to share what we learn, engage in conversations that move the field forward, and contribute to a better understanding of how AI can be used responsibly and effectively for climate resilience.
With a new group of projects already selected for this year, we will also continue supporting promising innovators where funding, technical support, and the right partnerships can unlock greater impact.
Read more about the AI for Climate Resilience Program here.
The next phase will test whether these foundations translate into wider adoption and measurable resilience outcomes as tools move further into deployment, validation and use.
We will continue to share what we learn, engage in conversations that move the field forward, and contribute to a better understanding of how AI can be used responsibly and effectively for climate resilience.
With a new group of projects already selected for this year, we will also continue supporting promising innovators where funding, technical support, and the right partnerships can unlock greater impact.
Read more about the AI for Climate Resilience Program here.
© Milkywire AB, 2026. All rights reserved. Mailbox 3306, 112 73 Stockholm, Sweden. All donations are handled by WRLD Foundation Sweden (registered with org ID No "802526 - 9328") and WRLD Foundation US (registered 501(c)(3) charity).
© Milkywire AB, 2026. All rights reserved. Mailbox 3306, 112 73 Stockholm, Sweden. All donations are handled by WRLD Foundation Sweden (registered with org ID No "802526 - 9328") and WRLD Foundation US (registered 501(c)(3) charity).