To help aspiring entrepreneurs develop these capabilities, the Wadhwani School of Data Science and AI (WSAI) at IIT Madras has announced a one-week, in-person short course titled “Building a Successful AI Startup.”

The programme will be conducted from 10 to 14 August 2026 at IIT Madras. Open to students, researchers, professionals, aspiring entrepreneurs and early-stage startups, the course is designed to help participants transform promising AI concepts into practical and scalable ventures.

Practical Training for Future AI Founders

Many innovators begin with a technically strong solution but struggle to convert it into a sustainable business. A product may perform well in a laboratory or academic setting but still fail to attract customers because it does not solve an urgent market problem or offer a clear commercial benefit.

A major focus of the programme will be the development of a minimum viable product, commonly known as an MVP. An MVP is a basic version of a product that allows founders to test their idea with users, collect feedback and improve the solution before investing heavily in development.

This approach can help founders reduce risk and understand whether their proposed product has genuine market potential.

What Participants Will Learn

The programme will cover essential areas such as customer discovery, market validation, business model development, product-market fit, user acquisition, revenue strategy and startup scaling.

Customer discovery will help participants understand the needs, behaviour and expectations of potential users. Market validation will enable them to assess whether customers are genuinely willing to adopt or pay for their solution.

These stages are particularly important for AI startups, where founders can sometimes focus more on the technology than on the customer problem. A technically advanced AI solution may not become a successful business unless it provides measurable value to its users.

Participants will also study fundraising preparation and term sheet negotiations. These sessions can help aspiring founders understand how investors evaluate startups, how funding agreements are structured and how company ownership may be affected during investment rounds.

The scaling component will examine challenges that growing AI companies may face, including computing costs, data quality, model performance, hiring, customer trust and responsible AI deployment.

An Immersive, Experience-Based Format

Rather than relying only on classroom theory, the one-week programme will combine expert-led sessions, practical exercises, interactive discussions and examples from real AI startup journeys.

This hands-on approach will allow participants to test their business assumptions, refine their ideas and receive constructive feedback. It can be especially useful for students and researchers who possess strong technical knowledge but have limited experience in business strategy, sales, marketing or fundraising.

The in-person format may also encourage collaboration among participants from different professional and academic backgrounds. Engineers, researchers, product managers and entrepreneurs can exchange perspectives and potentially identify future co-founders, advisors or collaborators.

Participants will therefore gain not only structured startup knowledge but also opportunities to build useful connections within the AI entrepreneurship ecosystem.

Expert Guidance from Lakshmi Shankar

The course will be led by Lakshmi Shankar, General Partner at Together Fund and former Vice President of Google Search. He has also held leadership roles at Twitter, IBM and EY and is a Sloan Fellow at the Stanford Graduate School of Business.

His experience across technology, business leadership and venture capital gives participants an opportunity to understand AI entrepreneurship from both an operator’s and an investor’s perspective.

Startup founders must make important decisions about product direction, hiring, fundraising, customer acquisition and growth, often with limited information. Guidance from an experienced technology and investment leader can help participants understand what makes an AI startup investable, how to communicate a strong value proposition and which common mistakes should be avoided during the early stages of venture building.

Participants can also gain insights into how investors analyse market size, founding teams, product differentiation, revenue potential and long-term scalability.

Pitch Day and ₹4 Lakh Prize

One of the programme’s main highlights is Pitch Day, where participants will present their startup ideas before a panel of investors.

The session will allow aspiring founders to test their concepts, practise investor communication and receive feedback from professionals with real funding experience.

A strong startup pitch should clearly explain the problem, proposed solution, target market, competitive advantage, business model and growth potential. The course will help participants bring these elements together into a structured and convincing presentation.

The top-performing team will receive ₹4,00,000 to support the development of its AI venture.

For an early-stage startup, the funding could be used for prototype development, customer testing, cloud infrastructure, market research or initial market entry. Recognition from an IIT Madras programme may also improve the winning team’s credibility while approaching future mentors, business partners and investors.

Who Can Apply?

The IIT Madras AI startup course is open to:

• Final-year BTech students
• MTech students
• PhD scholars
• Faculty members
• Researchers
• Engineers
• Product managers
• Aspiring entrepreneurs
• Early-stage AI startups

The broad eligibility criteria are important because successful startup teams often combine technical, product and business capabilities. A researcher may provide subject expertise, an engineer may build the solution, while a product professional may understand customer requirements and market positioning.

Applications will be evaluated, and selected participants will be informed through email. Since the programme will be conducted in person, applicants should be available at IIT Madras throughout 10–14 August 2026.

Why the Programme Matters

India’s AI startup ecosystem is expanding, but building a sustainable technology company remains challenging. Founders need to move beyond creating impressive technology and focus on customer value, commercial feasibility and responsible growth.

Programmes such as this can help innovators develop a structured approach to entrepreneurship while avoiding common early-stage mistakes. They also demonstrate the growing role of leading academic institutions in connecting technical talent with business mentors and investors.

For students and researchers, the programme offers an opportunity to explore entrepreneurship and determine whether an idea has commercial potential. For early-stage founders, it provides a platform to strengthen their business model, improve their investor pitch and prepare for future fundraising.

Interested candidates can refer to the official course information issued by WSAI, IIT Madras. Programme-related queries may also be sent to events@dsai.iitm.ac.in.

With practical training, expert mentoring, investor feedback and a ₹4 lakh prize, IIT Madras’s “Building a Successful AI Startup” course offers aspiring founders a focused opportunity to move from an AI concept towards a viable and scalable business venture.

Frequently Asked Questions

1. What is the IIT Madras AI startup course?

“Building a Successful AI Startup” is a one-week, in-person short course offered by the Wadhwani School of Data Science and AI at IIT Madras. It will train participants in product development, customer discovery, market validation, fundraising, pitching and startup scaling.

2. When and where will the programme be conducted?

The programme will be held from 10 to 14 August 2026 at the IIT Madras campus. Selected applicants will need to attend the sessions in person.

3. Who is eligible to apply for the course?

Final-year BTech students, MTech students, PhD scholars, researchers, faculty members, engineers, product managers, aspiring entrepreneurs and early-stage AI startups are eligible to apply.

4. What will participants learn during the course?

Participants will learn how to develop an MVP, validate a market opportunity, discover customers, achieve product-market fit, create revenue strategies, attract users, negotiate term sheets, raise funds and scale an AI venture.

5. Is there any funding support for participants?

Yes. Participants will present their startup concepts during Pitch Day, and the top-performing team will receive ₹4,00,000 to support the development of its AI venture.

Researchers at the National Institute of Technology Rourkela have developed an Artificial Intelligence-powered autonomous system that could offer a smarter, more sustainable way to monitor and clean solar power plants.

The patented innovation combines federated learning, artificial intelligence, edge computing, predictive maintenance and autonomous cleaning. It is designed to detect problems in real time, recommend cleaning only when required and reduce dependence on manual inspection.

For a country such as India, where many solar plants operate in dusty, dry or industrial environments, the technology could help improve power generation while reducing water consumption and maintenance costs.

The research team includes:

• Prof. Arun Kumar
• Prof. Bibhudatta Sahoo
• Dr. Lopamudra Hota
• Dr. Biraja Prasad Nayak

The team has secured an Indian patent titled “Federated Learning based Autonomous System and Method for Monitoring and Cleaning Solar Plant.”

The patent represents an important step towards integrating advanced computing technologies with renewable-energy infrastructure. Instead of treating solar panel cleaning as a routine manual activity, the system approaches maintenance as an intelligent, data-driven process.

Why Solar Panel Cleaning Is a Major Challenge

Solar panels need unobstructed exposure to sunlight to generate electricity efficiently. However, panels installed outdoors are constantly exposed to dust, sand, pollution, leaves, bird droppings and changing weather conditions.

In dusty and arid regions, these contaminants can reportedly reduce energy output by as much as 40%. Even smaller efficiency losses can become financially significant when they affect thousands of panels across a large solar farm.

Traditional solar panel cleaning usually follows a fixed schedule. Workers or machines clean panels after a predetermined number of days, regardless of whether every panel actually needs cleaning.

This approach creates several challenges:

• Unnecessary water consumption
• High labour and maintenance costs
• Limited visibility into panel-level performance
• Delayed identification of faults
• Cleaning of panels that may not yet require attention

Water consumption is a particularly important concern. Many large solar plants are located in areas with abundant sunlight but limited water availability. Frequent water-based cleaning can therefore place additional pressure on local resources.

How NIT Rourkela’s Solar Cleaning Technology Works

The NIT Rourkela system uses federated learning to support intelligent monitoring, fault detection and maintenance planning.

Federated learning is a method in which multiple devices or locations can contribute to the training of an AI model without sending all their raw operational data to a central server. Instead, relevant model updates or encrypted information are shared.

In a solar plant, different panels, sensor units or sections of the facility could process information locally. The system could then use this distributed intelligence to identify performance changes, detect possible faults and determine when cleaning is necessary.

Because raw data does not always need to leave the local device, the approach can offer several potential benefits:

• Improved data privacy
• Lower bandwidth requirements
• Better cybersecurity
• Faster local decision-making
• Easier deployment across large solar installations

The use of edge computing is another important part of the innovation. Edge computing allows data to be processed close to where it is generated rather than relying entirely on a distant cloud server.

This can help the system respond more quickly to changes in panel performance or environmental conditions.

Cleaning Panels Only When Necessary

One of the most important features of the patented platform is condition-based cleaning.

Instead of cleaning every panel according to a fixed calendar, the system is designed to evaluate actual operating conditions. It can identify panels or sections that may be affected by dirt, faults or reduced performance and recommend targeted action.

This selective approach could reduce unnecessary cleaning cycles while ensuring that heavily affected panels receive attention at the right time.

The expected benefits include:

• Reduced water consumption
• Lower labour requirements
• Improved solar energy generation
• Faster fault detection
• Lower operating and maintenance expenses
• More efficient use of cleaning equipment

Predictive maintenance may also help plant operators address problems before they lead to major power losses or equipment failure.

A Potentially Cost-Effective Alternative

Many autonomous solar maintenance systems currently available in the market can involve high capital costs. Some systems also provide limited intelligence, focusing mainly on mechanical cleaning rather than advanced monitoring and decision-making.

According to Prof. Bibhudatta Sahoo, the NIT Rourkela technology is expected to provide advanced autonomous capabilities at approximately 10% of the cost of existing systems once it is scaled for field deployment.

This cost advantage could be particularly valuable for small and medium-sized solar operators that may not be able to invest in expensive robotic maintenance systems.

However, the final cost and commercial performance will depend on successful hardware development, field testing, manufacturing and deployment at scale.

Current Technology Readiness Level

The system has been validated through simulation and is currently at Technology Readiness Level 3, commonly known as TRL-3.

At this stage, a technology has demonstrated proof of concept under controlled experimental conditions. It has not yet reached the stage of full-scale commercial deployment.

The next major step for the research team is to develop a physical hardware prototype. This prototype is expected to include Internet of Things sensors and other infrastructure required for real-world monitoring and autonomous operation.

Following prototype development, the system will need to undergo pilot testing at operational solar installations. These trials will help evaluate its reliability, cleaning effectiveness, water savings, cybersecurity and cost advantages under different environmental conditions.

Applications Across the Solar Energy Sector

The technology could potentially be used across several types of solar installations, including:

• Utility-scale solar power plants
• Rooftop photovoltaic systems
• Floating solar farms
• Industrial solar parks
• Smart-city energy infrastructure
• Defence installations
• Remote and off-grid renewable-energy systems

Its distributed and privacy-preserving design could be especially useful for installations where sensitive operational data cannot be freely transferred to external servers.

Remote solar systems could also benefit from autonomous monitoring because frequent manual inspection may be expensive or difficult in isolated locations.

Future Development Plans

The NIT Rourkela research team plans to add more capabilities to the system as it advances.

Possible future features include drone-assisted inspection, multi-agent collaborative cleaning and predictive energy-yield forecasting.

Drones could inspect large solar farms more quickly and identify damaged, dirty or underperforming panels. Multi-agent cleaning would allow several robotic units to coordinate their activities across a solar facility. Predictive forecasting could help operators estimate future electricity generation based on weather, panel condition and historical performance.

Together, these capabilities could create a more comprehensive solar-plant management platform.

Supporting India’s Clean-Energy Goals

The innovation is aligned with India’s broader efforts to expand renewable energy and work towards long-term Net Zero goals.

As solar capacity increases, efficient maintenance will become increasingly important. Poorly maintained panels may generate less power than expected, affecting both financial returns and the reliability of renewable-energy infrastructure.

By combining AI, federated learning, edge computing and autonomous cleaning, the NIT Rourkela system offers a practical example of how digital technologies can address physical infrastructure challenges.

The technology is still at the proof-of-concept stage, and real-world pilot results will ultimately determine its commercial potential. Nevertheless, its focus on targeted cleaning, privacy-preserving intelligence, lower water consumption and affordable deployment makes it a promising development for the solar energy sector.

Frequently Asked Questions

1. What is NIT Rourkela’s AI-powered solar cleaning system?

It is a patented autonomous platform developed to monitor solar plants, detect faults and recommend or perform cleaning based on actual panel conditions. It combines artificial intelligence, federated learning, edge computing and predictive maintenance.

2. How does federated learning help solar plants?

Federated learning allows devices or different sections of a solar plant to contribute to an AI model without transferring all raw operational data to a central server. This can improve privacy, cybersecurity, bandwidth efficiency and scalability.

3. How can the system reduce water consumption?

The system follows a condition-based approach and recommends cleaning only when panels actually require it. This can prevent unnecessary cleaning cycles and reduce the amount of water used for solar panel maintenance.

4. Is the NIT Rourkela system commercially available?

The technology is currently at TRL-3 and has been validated through simulation. A hardware prototype and real-world pilot testing are still required before large-scale commercial deployment.

5. Where can the technology be used?

The system could be used in utility-scale solar farms, rooftop systems, floating solar installations, industrial parks, smart cities, defence facilities and remote off-grid renewable-energy projects.