What works in AI: The leading companies turning AI into real-world impact

The World Economic Forum’s MINDS (Meaningful, Intelligent, Novel, Deployable Solutions) programme recognizes companies that use artificial intelligence (AI) to deliver real impact across sectors.

The first cohort – comprising 18 AI industry convergers from 23 countries – was unveiled at the Annual Meeting of New Champions 2025 in Hangzhou, China. The second cohort now brings together 15 additional global industry cases.

The programme’s strength lies in its organizations, which together provide a rich repository of case studies and impact metrics that enable broader learning and replication. These insights are captured in the report, Proof Over Promise: Insights on Real-World Adoption from 2025 MINDS Companies, which serves as a practical framework for successfully scaling AI applications.

Here are selected members of the second cohort.

Several organizations are pioneering the use of AI to deliver practical, climate solutions spanning batteries, power grids, energy markets, buildings and clean generation.

CATL is reshaping electric vehicle battery design with an AI-powered platform that combines physics-based electrochemical modelling and machine learning to automate and accelerate development.

Processing more than 50 million data records, the platform integrates structured and unstructured data to deliver optimal designs in minutes rather than weeks, cutting data operations by 99% and shortening prototype cycles by nearly 50%.

State Grid Corporation of China is using AI to manage Shanghai’s power grid more efficiently.

The platform brings together forecasting, trading, regulation and settlement, allowing distributed energy resources to be coordinated in real time. With sub-second response times, it supports over 15,000 users and demonstrates how megacities can build more resilient, sustainable energy systems.

NICE’s AI energy-market foresight system connects live regulatory updates with market data to improve accuracy and reduce risk.

By combining an AI model trained on energy policy with market forecasting, it continuously monitors how policy changes affect markets, cutting energy usage by up to 95%.

Schneider Electric is improving building comfort and efficiency with on-device AI that learns, adapts and delivers sustainable energy savings in real time.

Its room controller predicts thermal behaviour, adjusts settings to maintain comfort, and minimizes energy use—achieving 5–15% savings in just two weeks while reducing emissions.

China Huaneng Clean Energy Research Institute, together with its partners Huaneng Jilin Power Generation and Co., and China Huaneng Group Jiangsu Branch, is making offshore energy more efficient through a real-time, AI- and advanced analytics-powered platform.

It predicts equipment needs, optimizes operations and boosts clean energy output – enough to power over 10,000 homes annually.

As renewable energy infrastructure becomes more distributed and complex, maintaining reliability requires closer integration between data, diagnostics and control.

Organizations are also embedding AI across healthcare systems, expanding access, improving outcomes and enabling more resilient, human-centred models of care.

Landing Med is improving women’s health in China through AI-driven cytology, bringing cancer screening to communities worldwide. Its platform automates cell analysis and connects clinics to remote pathologists.

The system processes volumes of medical imaging data and uses cloud and edge computing for real-time diagnosis, enabling over 12 million screenings across 91% of China’s remote telepathology network.

The Ministry of Health of Saudi Arabia and Amplifai Health are piloting their Thermal Foot Scan, using computer vision and large language models to interpret thermal patterns and generate clinical risk scores in under a minute.

By enabling nurses to conduct AI-assisted screenings, the programme bridges the podiatry workforce gap, allowing specialists to focus on high-risk cases while increasing screening capacity by up to 12× without expanding specialist headcount.

Early results show nearly 80% lower treatment costs, demonstrating a shift from reactive care to scalable, preventive models that support Saudi Arabia’s vision for equitable, technology-enabled healthcare, including deployment in remote and resource-constrained settings.

Phagos offers an AI-driven alternative to antibiotics, protecting animal health and food systems. Its platform predicts phage-bacteria interactions from genomic sequences and designs adaptive multi-strain phage cocktails.

It has achieved 95% accuracy and is accelerating discovery by up to 10 times compared to manual methods.

OAO and Sanofi are building an AI-first enterprise where employees contribute to discovery and innovation.

Their multi-agent system captures insights across all levels, generating over 1,300 AI use cases, faster model development, and measurable commercial uplift.

Fujitsu and Social Medical Corporation Genshukai are partnering to apply AI to streamline hospital management, tackling challenges like medical record standards and bed allocation optimization.

The new operating model has already seen over 400 hours saved across hospital management and a revenue uplift of $1.4 million of 10%.

AI can modernize critical industrial and infrastructure systems, for example, by making cities safer or chips more efficient, embedding intelligence directly into the foundations of any economy.

Cambridge Industries empowers cities to keep roads safer and sites compliant through locally governed AI, making complex engineering knowledge accessible to everyone.

Their flagship applications, Orbit and Ndege, use private client-driven large language models to analyse road conditions and monitor construction-site safety via smartphones and drones, achieving nearly 50% reduction in emergency road-repair costs and construction incidents.

Dmall and Wumart are streamlining retail operations end-to-end, improving pricing, reducing shrink and cutting energy use across their store network.

Their unified retail operating platform connects point-of-sale data, inventory, cameras and Internet of Things (IoT) sensors, enabling dynamic, event-driven workflows. The system automates pricing, monitors operations and manages energy use in real time.

Deep Principle uses AI to bring chemistry, computation and data into a single, continuous discovery process. Its system rapidly predicts how chemical reactions will behave, automates more than half of materials simulations and reduces experimental costs.

By speeding up candidate identification and supporting real-time, transparent decisions, the platform enables faster, more collaborative materials research.

Synopsys and AMD are integrating agentic technology into electronic design automation (EDA) tools, improving the efficiency of semiconductor design. These systems support human designers by understanding design goals, running tools, and evaluating results, helping teams work more productively.

The approach has doubled productivity while cutting design costs and approval times, accelerating innovation in chip design.

Tech Mahindra is building multilingual AI models that support Hindi, Bahasa and regional languages often overlooked by global systems.

Trained on large-scale local data, the models now power 3.8 million monthly queries across citizen services, banking and healthcare for more than 10,000 active users.

With 92% conversational accuracy, they outperform generic global models and could deliver up to $2 billion in productivity uplift. The initiative aligns with OECD AI Principles and the Hiroshima AI Process, supporting inclusive AI adoption across the Global South.

Across the two cohorts, common attributes emerge. Each organization demonstrates a commitment to social and environmental progress: empowering the workforce, expanding access to education and upskilling, and leveraging AI to reduce energy consumption.

Collectively, MINDS companies report double-digit gains in productivity and revenue, driven by improved operational efficiency. This success reflects their ability to scale responsibly, moving from pilots to enterprise-wide adoption by embedding AI into core processes and modernizing technology infrastructures.

It is the interplay of strategy, workforce, data, technology and governance that enables amplified, measurable impact.

As the first two cohorts shape the global conversation on responsible AI, preparations for the third cohort are underway, with stakeholders across industries – from agrifood to cyber defence – invited to apply and join the growing community.

The second MINDS cohort makes clear that impactful AI is being embedded in real systems and delivering measurable results. The opportunity now is to replicate what works so that responsible AI becomes the norm rather than the exception.

Thank you to Federico Capaccio, Fatima Gonzales Novo, Eric Enselme, Na Na and Chiharu Nakayama for their collaboration and leadership on this project.