Build Fast with AI: AI Build Challenge 2026

Overview Looking for an opportunity to turn your AI skills into something real? Build Fast with AI: AI Build Challenge 2026 gives you the chance to build a working AI system, compete for ₹18,000 in prizes, showcase your project, and get considered for up to 10 internship opportunities through the post-hackathon selection process. Prize selection and internship selection are separate decisions. For Further Details and Updates, join the WhatsApp group: https://chat.whatsapp.com/LMrDqE2tKTy3PRAyHIdPLH Eligibility Open to 3rd-year students, final-year students, and freshers from an IT background. Individual and Team participation is allowed. Inter-college teams are allowed. Students from different specializations and colleges can form a team. Each team can select and build one track only. Participants must meet the eligibility criteria individually. Participants from different engineering specializations are allowed to collaborate within the same team. Format Mode: Online Participation: Individual or Team Build Duration: 24 September 2026, 10:00 AM IST to 26 September 2026, 10:00 PM IST Total Build Time: 60 hours There are 6 problem tracks, and each participant must choose one track. Tracks Track 01: Live Web Agent Build an AI agent that can complete a real multi-step task on live websites. The agent should be able to: Compare information across websites Fill forms Gather information Complete multi-step workflows Recover from errors during execution Provide a verifiable step-by-step trace Track 02: Meeting-to-Living-Memory Build an AI system that ingests transcripts across multiple sessions and answers questions that require understanding what changed or contradicted between meetings. The system should be able to: Process transcripts from multiple meetings Maintain information across sessions Identify what changed over time Detect contradictions Resolve conflicting information Answer questions using temporal context Track 03: Contract/Policy Redliner Build an AI agent that reviews a document against a supplied rulebook or playbook. The agent should: Identify risky clauses Flag issues based on the supplied rules Cite the exact rule supporting each finding Suggest appropriate redlines Ground recommendations in the supplied rules and document Track 04: Code Archaeologist Build an AI system that explores a messy and undocumented code repository and helps developers understand the codebase. The system should: Explore the repository Generate an architecture diagram Create an onboarding guide Identify dead code Flag risky code patterns Help developers understand unfamiliar parts of the codebase Track 05: The Un-Boring Textbook Build a system that turns a dry PDF chapter into a short narrated explainer video. The system should: Understand the source PDF Generate an engaging explanation Create a short narrated explainer video Ground every factual claim in the source Cite every claim to the relevant source page Track 06: Inbox-to-Action Butler Build an AI agent that takes a messy inbox or support-ticket export and turns it into actionable work. The system should: Triage incoming messages or tickets Prioritize them Identify the required action Draft ready-to-send replies Keep a human in the loop for approval before sending Submission Requirements Each participant must submit: A runnable project with a deployed link GitHub repository with setup instructions A 3-minute demo video A PPT presentation explaining the project (maximum upto 10 slides) Disclosure of AI tools used The PPT should clearly explain the problem, approach, solution, key features, technology stack, architecture or workflow, results, and demo. The project should be submitted before the final submission deadline. Evaluation Criteria Submissions will be evaluated based on: 30% Working system 25% Output quality and tests 20% Reliability 15% Code quality and reproducibility 10% Usability Rules Individual and team participation are allowed. Inter-college and inter-specialization teams are allowed. Each participant or team can participate in one track only. All changes and commits must be made within the hackathon timeline. Any changes made before or after the hackathon timeline may result in automatic disqualification. The project must be submitted before the final submission deadline. The submitted project must be runnable, include a deployment link, and provide clear setup instructions. Participants or teams must submit a 3-minute demo video. Participants or teams must submit a PPT presentation with a maximum of 10 slides. Participants or teams must disclose the AI tools used in building their solution. Participants or teams must provide the required output for their selected track. Shortlisted participants or team members may be required to explain their implementation. Participants are responsible for ensuring that their submissions are original and that external resources, datasets, models, APIs, and AI tools used are appropriately disclosed. Prize selection and internship selection are separate decisions. Up to 10 internship selections will be made following the evaluation and interview process.