c0mpiled_
← back to articles
hack·march 31, 2026·15 min read

c0mpiled-9: Bangalore x Magicball x Razorpay

by fiyin olayisade577 reads

Overview

There's a peculiar irony in India, a country that built entire industries around constraints, becoming one of the most interesting places to watch people eliminate it.

150 builders. Razorpay HQ. Four hours. Over 500 applied, 150 showed up. If you've ever tried to get people to give up a Saturday in Bangalore, where traffic alone is reason to stay home, you'll appreciate that number.

The theme was agentic AI. Not the kind that summarises your emails marginally faster. Systems that actually do something: receive an input, make a decision, use a tool, close the loop.

Four hours is a revealing constraint. No roadmaps. No "here's what it will eventually do." It either works or it doesn't.

India produces exceptional engineers. What was interesting was watching that instinct adapt to a format that rewards speed over sophistication. Some found it uncomfortable. The best teams found it clarifying.

We partnered with Magicball, Razorpay, and Emergent AI to make it happen.

Reflections

  • Good: Magicball, Razorpay and Emergent AI were fully present throughout the day — not just as logos on a banner but as active participants in the room. That kind of partner engagement makes a real difference and we are grateful for it.
  • Good: Most teams arrived pre-formed and ready to go, which set a great tone from the start and pushed the quality of what teams were able to produce. We hope to see this become the norm across all c0mpiled editions.
  • Good: The technical depth in the room was exceptional. Bangalore's builder community is the real deal and it showed across the submissions.
  • Learning: Venue access created some early friction. Going forward we will work more closely with venues ahead of time so builders can hit the ground running from the moment they walk in.
  • Learning: Team registration will be moving to a group model. Having individuals register separately meant teams were sometimes split across acceptance decisions. From the next edition, registering as a team means the whole group moves through together.
  • Learning: We will be dedicating time at the start of each event to walk participants through the submission process. A number of strong projects were disqualified due to avoidable errors and we want to make sure every team's work gets a fair hearing.
  • Learning: Judging needs more structure. A clear timeline upfront, explicit guidance on pacing, and an agreed process for handling ties will make the final stages of the event run better for everyone involved.
  • Observation: We will be sharing curated examples of strong hackathon submissions & pitch videos with participants ahead of future events. Seeing what great looks like sets a clearer bar than describing it ever could.
  • Observation: As building shifts from raw coding to prompting, orchestrating and tooling, we are thinking about how our judging framework can better reflect those skills. The definition of technical excellence is changing and our criteria should keep pace.
  • Observation: Taking c0mpiled international has surfaced real differences in how builder cultures approach the format. Bangalore's technical depth was immediately apparent, but a four hour window is a different discipline from the 24 or 48 hour formats many were used to — one that rewards prioritisation over completeness. Designing for that mental shift is something we are actively working on.
  • Observation: Some participants struggled to build directly against YC's Requests for Startups, which raises an interesting question for future Bangalore editions. The greatest YC companies from India, Razorpay included, started by solving local problems before scaling globally. There is no shortage of high friction problems in India worth building on, and a locally anchored theme may unlock more natural and compelling ideas from the room.

Judges & Partners

Prizes & Recognition

1st Place ($3000): Vikings

2nd Place ($1500): Ring Zero

3rd Place ($500): Argus

All Submissions

2-Techies (AI-Native Agencies (Spring 2026))
Description: RefundShield AI is an autonomous system that investigates e-commerce refund claims for fraud. It analyzes submitted evidence using AI vision, checks for reused or AI-generated images, verifies merchants, and generates a fraud risk score to approve or reject claims automatically.
Team: Sujay D, ShrihariKulkarni
Partners Used: Crustdata, Razorpay
Video: https://drive.google.com/drive/folders/1bROOao0TUyE2X0phuhxaOtGtYwY3g-8I

Anshika Vijay (Cursor for Product Managers (Spring 2026))
Description: AI that helps product teams iterate on UX faster
Team: Anshika Vijay
Video: https://www.loom.com/share/21acfd393da342709dfc66e718020c59

Argus (AI-Native Agencies (Spring 2026))
Description: ARGUS is an AI-powered security auditing agent that autonomously discovers vulnerabilities in AWS cloud infrastructure. Unlike traditional scanners, it forms security hypotheses, investigates them through a reasoning loop, and correlates findings across IAM, network, compute, and data layers to surface composite attack chains.
Team: Prakhar Khatri, Omm Animesh Mishra
Partners Used: Gearsec (Browserwire), Safedep
Video: https://www.loom.com/share/c9f7d58248d74a53bcabb50f0bd996d1

Automatic Form Filler for Govt (Arun + Laxmi) (AI for Government (Spring 2026))
Description: Agentic AI for automatic form fills with multi lingual support
Team: Arun, Laxmi
Video: https://drive.google.com/drive/folders/1a4JQMZfl6yxgZPcBitujNnO4ezvUobCP?usp=drive_link

Barcelona (AI-Native Agencies (Spring 2026))
Description: Agentic AI on Mobile Device
Team: Ujwal KV, Sahana BS, Ayush, Naresh
Partners Used: Razorpay
Video: https://drive.google.com/drive/folders/15Q7LjJW0NAZmkPfTMfdyOV6UPC2UXwNd?usp=sharing

ChaosProbeX (AI-Native Enterprise Software (YC Fall 2025))
Description: ChaosProbeX is an API resilience and observability platform that bridges traditional testing and chaos engineering. It subjects services to real-world fault scenarios to identify breaking points, then uses AI-driven root cause analysis to help teams harden infrastructure before production.
Team: Chetan
Partners Used: Safedep, Concierge
Video: https://www.loom.com/share/cfc03ddc3bdc4357a8435d80d717c0b4

Claimly (AI-Native Agencies (Spring 2026))
Description: Claimly puts an AI agent on a phone number. You call it — or it calls you — and it already knows your policy, your documents, your claim history. It can answer questions, file claims, schedule adjusters, update records, and hand you to a human the moment it can't handle something. All over a regular phone call. No app, no portal, no hold music.
Team: Kshitij, Chirag
Partners Used: Emergent, Razorpay
Video: https://www.loom.com/share/8a4d1128071e4399b76fbfa3511f4e38

Code Alpha (AI Guidance for Physical Work (Spring 2026))
Description: MediGuide AI is a healthcare operations platform that helps pharmacists, nurses, and clinic staff check medicine availability, view stock across sources, find substitutes, and handle urgent requests — reducing delays and improving workflow accuracy in real time.
Team: Eshwar B N
Partners Used: Emergent
Video: https://drive.google.com/file/d/1QvmQGXF33Hh7CXy_IaVvvhihnq9tOMuW/view?usp=drive_link

Hack_4 (AI for Government (Spring 2026))
Description: This an AI agent used to reduce the ambiguity to check the right eligibility for the people to their exact eligible scheme
Team: Shivayogi, Kishan, Sougandh, Shreyas
Partners Used: Emergent
Video: https://drive.google.com/file/d/1veGfXzfiJmpU5BTTXnoGWXokiulGA5Lt/view?usp=sharing

HedgeHogs (AI-Native Hedge Funds (Spring 2026))
Description: TryHedge is a quant hedge fund for carbon credits. Its AI engine scores every credit like a credit rating using satellite imagery, wildfire risk, registry data, and audit trails, then executes trades autonomously — going long on underpriced, high-quality credits and shorting overpriced ones.
Team: Surya Tiwari, Anish
Partners Used: Emergent
Video: https://drive.google.com/drive/folders/14XTegOspL2OCdeBc8RVHkS2QPE5xeJXR?usp=sharing

Hierarchy (Cursor for Product Managers (Spring 2026))
Description: https://docs.google.com/document/d/1LrnqqM7WzszPmZsvFlkMKD_h5oL4BI0OKv1IZ-eOvSY/edit?usp=sharing
Team: K Hemant Kumar, Mayank Saini, Subal Bain, Smit Patel
Partners Used: Emergent
Video: https://drive.google.com/drive/folders/1hKU6O24L1w6U7LnapryBl52MqGWs5ALy?usp=drive_link

High Agency Capital (AI-Native Hedge Funds (Spring 2026))
Description: AnalystOS compresses hours of analyst work — reading concalls, extracting guidance, building assumptions, and comparing opportunities — into one source-backed sector analysis workflow.
Team: Harsh Chandak
Partners Used: Emergent
Video: https://www.loom.com/share/c9f39f9f818340e4835a1c92a068b5da

Innovative Engineer (Cursor for Product Managers (Spring 2026))
Description: PMBrain AI is a Cursor-like system for Product Managers that continuously analyzes customer feedback, product usage, and codebases to discover unmet needs. It produces implementation-ready specs, timelines, and execution plans — turning product strategy into an autonomous AI-driven loop.
Team: Williyam M
Partners Used: Emergent
Video: https://drive.google.com/file/d/1Ew1lHdV8rtjNYthp-JIh78jsvHx2oLO5/view?usp=sharing

KhataPe (AI for Government (Spring 2026))
Description: KhataPe is a WhatsApp CFO agent for Indian SMBs. When a customer pays via Razorpay, it automatically calculates CGST and SGST, logs the transaction, and sends a GST breakdown to the merchant on WhatsApp. No human involvement, no manual entry — targeting 1.4 crore GST-registered businesses that still reconcile payments manually.
Team: Amogh Sunil, Akilesh Gowda
Partners Used: Crustdata, Emergent, Razorpay, Unsiloed, s2.dev
Video: https://www.loom.com/share/cd55bf0f12c74c988445aa99579f6826

KubixAI (AI-Native Enterprise Software (YC Fall 2025))
Description: Kubix.ai is an AI-powered GPU kernel optimization platform that profiles, diagnoses, patches, and verifies GPU workloads making every team's models run 2–10x faster without needing $2M/year CUDA engineers.
Team: Mihir, Naman Shankar Srivastava
Partners Used: Emergent, Razorpay, Unsiloed
Video: https://www.loom.com/share/f4e7724ced14485eb6ffa63459e451fc

LeanPay AI (Infra for Government Fraud Hunters (Spring 2026))
Description: CleanPayout sits between every invoice and every payment. Before any money moves, it automatically checks if the vendor is real, if their GST registration is legitimate, and if the price is fair. If something is wrong, it blocks the payment and tells you exactly why in plain English.
Team: Laxman Srivastava, Manit Bohra, Gagan
Partners Used: Crustdata, Razorpay, Unsiloed
Video: https://drive.google.com/file/d/1gJ5S7EihYUGZGrNz92-mgZthbo3rVwhG/view?usp=sharing

Lynx (Infrastructure for Multi-Agent Systems (YC Fall 2025))
Description: Infrastructure for Multi-Agent Systems
Team: Aditya Rana
Partners Used: Emergent
Video: https://www.loom.com/share/155717f478e747119da6f3764d0c6c43

Memorystack (The First 10-person, $100B Company (YC Fall 2025))
Description: Nikolas is an autonomous AI scientist agent that discovers novel drug molecules and catalysts without human chemists in the loop. Powered by PRISM, a 5-layer meta-cognitive architecture, it generates candidate molecules, validates drug-likeness, and simulates molecular docking — getting measurably smarter with each experiment. Results span CO₂ catalysis, cancer, aging, and Alzheimer's, with all discovered molecules verified novel against PubChem's 130M+ compounds.
Team: Pandurang Mopgar
Video: https://www.loom.com/share/c9bb621788d7490e80b54151feee4824

MSJ News | Tokenised.News (AI for Government (Spring 2026))
Description: We are building a social news platform the that helps fight disinformation and isinformation by fact/Opinion checking news content by journalists to begin with, but eventually more too. We also plan to pay both publishers as well as users
Team: Tanmaya Sharma
Partners Used: Emergent
Video: https://drive.google.com/file/d/1V6QUCqkz7YT2Eym1x3iHY_coOVdjGM8b/view?usp=sharing

NeuroNova (AI-Native Enterprise Software (YC Fall 2025))
Description: Diligence OS is an AI-native platform that fully automates venture capital due diligence. When a startup submits their data room, a multi-agent system extracts claims from pitch decks and documents, then verifies them against live firmographic databases, public records, and codebase security scans. The result is a deterministic "Trust Score" that highlights discrepancies between what a founder claims and what the data proves.
Team: Kunal Jaiswal, Atharva Agrawal, Ashdeep Singh, Nilpat
Partners Used: Emergent, Safedep
Video: https://drive.google.com/drive/folders/1b4nWESHKHQZULk0-H73EgWS7YlSRYBhO?usp=sharing

OpenTender (AI for Government (Spring 2026))
Description: OpenTender is an AI-powered interface for government procurement. It aggregates public tenders, structures the data, and lets vendors search using plain language. The system matches vendors with relevant tenders, explains eligibility requirements, and guides them through the application process.
Team: Aravind, Vikas C
Video: https://www.loom.com/share/fa3237eab8ac42479253de1a00063279

PiedPiper (Infrastructure for Multi-Agent Systems (YC Fall 2025))
Description: InBots is the communication layer for AI agents. It enables agents to discover and communicate with each other through MCP (Model Context Protocol), while giving humans full real-time oversight via a live dashboard. Enterprises create isolated workspaces where agents register, exchange messages, and trigger webhook notifications — with direct and group messaging, OTP-based auth, and subscription billing built in.
Team: Rahil Mavani
Partners Used: Crustdata, Emergent, Razorpay, Safedep
Video: no video

Plenvo (Cursor for Product Managers (Spring 2026))
Description: figureout.ai is Cursor for Product Managers — an AI-powered tool that turns raw startup ideas into validated, scoped MVPs. It guides users through a structured 7-phase process from idea articulation to competitive research, TAM sizing, and MVP scoping, outputting a downloadable PRD with validated market data and a realistic build plan.
Team: Yatharth Lakhera
Partners Used: Emergent
Video: https://www.loom.com/share/6aa743f1dd8a410d9c1e294b03339439

potato rangers (AI for Government (Spring 2026))
Description: AgentPay automates cross-border B2B payments. Drop a PDF invoice into a folder and the agent extracts vendor details, runs OFAC sanctions screening and compliance checks, submits a USDC transfer on-chain, and generates a tamper-evident audit trail — removing the human bottleneck from every payment cycle.
Team: Pruthvi, P S Lochan, Sai sugeet, Tanushree G, Dwiraj S
Partners Used: Crustdata, Emergent
Video: https://drive.google.com/drive/folders/1iStzLt7rXY6gW3vV01XrENlGci9lElKh?usp=drive_link

RAIL (AI-Native Agencies (Spring 2026))
Description: A real-time proxy that intercepts every AI agent tool call, evaluates it against 8 responsible AI dimensions and regulatory compliance frameworks (India DPDPA, EU AI Act), and blocks unsafe or biased actions before they execute.
Team: Rohit Panwar, Sumit verma
Partners Used: s2.dev, Gearsec (Browserwire), Safedep
Video: https://www.loom.com/share/eea15a45c105455594b479dc7508d0a1

RecoverAI (AI-Native Agencies (Spring 2026))
Description: RecoverAI automatically recovers failed payments in India. When a payment fails on Razorpay, our AI analyzes the failure reason, sends a personalized WhatsApp message in Hindi, generates a fresh payment link, and logs everything — all in under 8 seconds. No human required.
Team: Asad khan
Partners Used: Razorpay
Video: https://www.loom.com/share/f37a8ecf6ab249e19e670f17e220c6c4

Ring Zero (AI-Native Agencies (Spring 2026))
Description: FlowPay is an AI‑native payments ops agent for small businesses in India. You type what you want done in plain English, FlowPay handles the Razorpay links, Google Sheet updates, emails, and follow‑ups.
Team: Mohit Paddhariya
Partners Used: Razorpay
Video: https://www.loom.com/share/5f4b4baed68541828917bbcac7c05fa1

Rockcoderz (AI-Native Enterprise Software (YC Fall 2025))
Description: AutoOps automates IT operations using AI agents. It autonomously resolves tickets in seconds, integrates with tools like Slack, GitHub, Jira, and PagerDuty, and automatically creates searchable knowledge base articles — zero human intervention for 80–90% of common IT issues.
Team: Shashank T S
Partners Used: Emergent
Video: https://www.loom.com/share/8ed5a88496374683905c5854449ebe15

Santhosh Aditya (AI-Native Agencies (Spring 2026))
Description: End to end sales pipeline that starts with prospect discovery and ends with high quality leads to your sales people
Team: Santhosh Aditya
Partners Used: Crustdata, Emergent
Video: https://www.loom.com/share/9dfdb471f37043c986fe98d3852463f5

Shanks (Stablecoin Financial Services (Spring 2026))
Description: Shanks Market Intelligence is a financial analytics platform that converts complex crypto and US stock market data into simple, visual insights. It features a real-time dashboard, a proprietary Shanks Score (0–10) for investment strength, Smart DCA signals for optimal accumulation zones, and an AI assistant that explains market signals in plain language.
Team: Sumeet Joshi
Partners Used: Crustdata, Emergent, Razorpay
Video: https://drive.google.com/file/d/1sjuPRfOI_vZO6vnxQFHA_fmLKz7F46Qi/view?usp=sharing

shiplog (AI-Native Agencies (Spring 2026))
Description: ShipLog automatically generates and sends stakeholder updates when a developer pushes code to GitHub. Every push triggers AI that reads commits, understands business impact, and writes three tailored updates simultaneously — investor emails, user changelogs, and team standups — sent via Gmail and Telegram with zero human involvement.
Team: Aanchal Pranjal Rathi
Partners Used: Crustdata, Razorpay, s2.dev, Safedep
Video: https://drive.google.com/drive/folders/1Cer2xWh-bXBY5qrS_7AbhoCeIjIulWZu?usp=drive_link

Team hawkins (AI Guidance for Physical Work (Spring 2026))
Description: RealityShield Autopilot is an autonomous AI system that reviews pull requests, detects security and reliability risks in dependencies, generates and applies remediation strategies, and returns a merge-ready PR with machine-verifiable proof.
Team: Mohan Sri Vamsi, Leemoji Reddy, Niteesh Sai Guddenti
Video: https://drive.google.com/file/d/1-DFf9IT-XRTbEhBfV5VddVcur4_kHluj/view?usp=sharing

Team Titans (AI-Native Agencies (Spring 2026))
Description: 3d property intelligence and viewer
Team: Amey, Naman, Hemapriya, Prerna
Video: https://www.loom.com/share/b423b9d4fb064c8b8784f166044cf021

TeamYM (AI-Native Enterprise Software (YC Fall 2025))
Description: An automated CI/CD healer that scans open source GitHub repos for bugs ranging from low to critical severity, generates fixes, and lets you create a pull request directly from the tool.
Team: Yash, Madhvan
Partners Used: Emergent
Video: https://drive.google.com/file/d/1730-6m4JKXrCTYHNsnm_LipJEI0kXW5T/view?usp=sharing

Technutpam (AI-Native Enterprise Software (YC Fall 2025))
Description: ComplianceShield is a compliance layer that sits inside AI coding tools like Cursor and Claude Code. Before code is written, it injects the relevant legal rules (GDPR, HIPAA, India DPDP, SOC2). After code is written, it scans for violations and offers one-click fixes. It also generates audit-ready compliance reports, scans dependencies for supply chain risks, and stays current with regulatory changes in real time.
Team: Mohamed Ali.A, Saad H Ilkal, Nagarjun Yogesh
Partners Used: Crustdata, Emergent
Video: https://drive.google.com/file/d/1ozCoYeGz4bKx8-YtgzUNQGr8fakOfLeC/view?usp=sharing

Terminator (AI-Native Hedge Funds (Spring 2026))
Description: AlphaAgent | AI-Powered Investment Signal Engine | Democratizing quantitative hedge fund-level analysis for individual investors and small fund managers. AlphaAgent scans 78K+ companies, scores growth signals, generates AI investment theses, and delivers real-time alerts — all in one clean interface.
Team: Bhagya Rana
Partners Used: Crustdata, Emergent, Razorpay, Unsiloed, s2.dev
Video: https://www.loom.com/share/ac3272d253f145caa4a7b2ec8a7477ca

The Node Runners (AI-Native Agencies (Spring 2026))
Description: A personalized job search dashboard powered by AI agents. It finds relevant job listings, scrapes application details from company websites, extracts resume data for proper formatting, and helps users improve their responses to HR questions — automating the end-to-end job application process.
Team: Yash Daga
Partners Used: Crustdata, Emergent, Gearsec (Browserwire)
Video: https://drive.google.com/file/d/1miPCJzF5GS-MCMdaYTDbtVGvlg3lQj_v/view?usp=sharing

Vikings (Make LLMs Easy to Train (Spring 2026))
Description: Tune8n brings n8n-style visual workflow automation to LLM and vision-language model fine-tuning. Design, connect, train — no code required.
Team: Akash Jha, Anshuman Singh
Partners Used: Unsiloed, s2.dev
Video: https://drive.google.com/file/d/19DXqANz5voCOD_5MOdAGfOSdyNCRZc9A/view?usp=sharing

XeroD (AI-Native Agencies (Spring 2026))
Description: NexusAI is an AI agent that replaces the entire sales outreach process. Type one company name and take one photo of your product — the agent researches the company, finds the right decision maker, writes a personalized email based on real signals like recent posts and hiring activity, converts your product photo into an interactive 3D model, and sends it all in 12 seconds.
Team: Ashish Singh, Shivam Raj, Saatwik
Partners Used: Crustdata, Emergent
Video: https://drive.google.com/drive/folders/1Oawh1_MvMkE3en0KgmLTHoaRSUW8ipmm?usp=sharing

Zapi (AI-Native Enterprise Software (YC Fall 2025))
Description: Zapi is a tool for teams to talk and vibe to there salesforce data with natural language and get back the data in forms of tables,records,charts,dashboards, so you can also create accounts,leads,in your salesforce account with natural language
Team: Alankrit Khatri
Video: https://loom.com/share/8151c6be0d10472aaccb74d0ff7ec56c

Zion (AI Guidance for Physical Work (Spring 2026))
Description: BoloBill is voice-first billing for Bharat. A shopkeeper speaks in Hindi or Kannada and in 4 seconds a professional bill is created, a Razorpay UPI payment link is generated, and the customer can pay by scanning a QR code. No typing, no forms — targeting 60 million small shops in India that still run on pen, paper, and memory.
Team: K N Nidhi
Partners Used: Emergent, Razorpay
Video: https://www.loom.com/share/f246b8527a9846349d87728fd6340392

All Photos









![(https://imagedelivery.net/rScUDlx4qgA-3tUZCbl67w/445418ea-ea0a-404a-5123-65347e718900/public)





* * *