Video summary

Datathon 2026: Introduction & Problem Statements Explainer Session.

Main summary

Key takeaways

Business

Business-focused summary (Datathon 2026: intro + problem statement explainer)

Event purpose & commercialization intent

  • Datathon 2026 (KSP crime-data use cases) is positioned as more than a “tech-for-tech’s-sake” hackathon.
  • Shortlisted solutions are intended to be taken into production deployment.
  • The stated business outcome is to drastically reduce investigation time across the lifecycle (from FIR lodging through offender jail/closure) using AI + analytics.

Target users

  • Primary end users are police/investigators and senior officers (not citizens).
  • Use cases emphasize operational usability:
    • fast queries
    • multilingual support
    • natural language interfaces
  • Outputs should be decision-ready dashboards for command centers.

Two problem statements (what teams must build)

Problem Statement 1: Intelligent conversational AI for KSP crime database

Core objective

Build an agentic conversational AI (chatbot-style, potentially voice-enabled) that supports multilingual querying (English + Kannada). Investigators should query the crime database using NLP rather than manual SQL.

Key expected capabilities

  • Crime pattern discovery
  • Criminal network analysis
  • Socio-demographic crime insights
  • Criminology profiling
  • Proactive crime prevention and intelligence
  • Contextual, multi-step investigation (clarified in Q&A)
    • Not “simple Q&A”; the agent should help connect an offender to other related cases.

Implementation/data expectations

  • Raw data will not be provided due to sensitivity/privacy.
  • Participants receive:
    • schema/master table structure
    • “header files” to generate synthetic data for prototyping.
  • LLM guidance:
    • LLMs are allowed as long as they run on Zoho Catalyst.
    • Guidance cautions against building/training LLMs from scratch (costly/complex).

Problem Statement 2: State-of-the-art data visualization & AI-driven analytics for crime investigations

Core objective

Transform state crime records-board data (currently in data silos and often handled via manual workflows) into proactive, AI-enabled state-level crime intelligence hubs.

Key expected capabilities

  • Advanced visualizations with drill-down
  • Emerging state analysis
  • Criminology network + link analysis
  • Sociology-focused AI-driven predictive dashboards

Operational emphasis (command/response use)

  • Dashboards should offer real-time/near-real-time situational snapshots to enable:
    • faster response
    • optimal resource allocation
  • Mentions latency/operational constraints considerations.

Tooling/approach guidance

  • They explicitly mention using PowerBI internally/early-stage.
  • Teams are encouraged to go beyond “just BI”—i.e., add advanced algorithms and research-backed statistics.

Frameworks / playbooks / processes mentioned (execution mechanics)

Prototype development & deployment playbook (Catalyst-centric)

  • Solutions must be deployed on Zoho Catalyst (mandatory).
  • If using Catalyst services, teams should prefer them (examples given):
    • serverless functions
    • docker/app services
    • QuickML
  • If not using Catalyst services, open-source tooling is allowed, but deployment must still be on Catalyst.

Submission process (repeatable)

Teams can submit multiple times before deadlines via the participant dashboard:

  • Select one challenge and solve its key aspects (avoid splitting effort across both).
  • Provide:
    • Prototype brief (using a provided submission template)
    • Public GitHub repository
    • Public demo video link
    • Prototype deck upload
    • Deployed solution link (on Catalyst)

Selection/shortlisting principles (quality bar)

Repeated emphasis:

  • Don’t build “quick and dirty.”
  • Evaluation will be by senior officers/domain experts.
  • Solutions should demonstrate:
    • production-grade potential
    • scalability (capacity planning; failure under user scale is a negative)
    • security
    • maintainability for long-term (~a decade) operation

Concrete operational KPIs / measurable success criteria (implied targets)

No numeric KPIs were stated, but “good” was described as:

  • Investigation time reduction (end-to-end speed from FIR to offender jail)
  • Query latency / efficient data pipelines for near-real-time responses
  • For dashboards:
    • a snapshot including trends/accuracy so officers can act quickly
  • Evaluation parameters will be posted (announced they will appear in the resource tab within days once decided).

Key timelines & milestones (dates)

  • June 11: Workshop “Introduction to Catalyst by Zoho” (critical to attend)
  • June 19: Last day to register
  • June 18: Another Catalyst workshop (recommended)
  • June 26: Workshop/prototyping milestone appears in the timeline description
  • July 28: Submissions due / “take the submissions live”
  • After submission: shortlisting → prototype refinement → in-person demo day grand finale
    • (Exact demo day date not provided in the subtitles.)

Data access, privacy, and constraints (major risk mitigations)

  • Raw/real data won’t be shared; participants must work with:
    • anonymized context
    • synthetic data
  • Use of master table structure / schema
  • Dataset will be posted in the Resource tab once ready.
  • Data/test data will not contain real PII; “PII/DPP act wouldn’t arise” per Q&A.

Example scenarios and domains mentioned (business realism)

  • Networked/cyber crimes:
    • cyber/cryptocurrency
    • dark web
    • cross-bank/international money flows
  • Chain-link intelligence:
    • connecting an offender to other crimes/cases to reduce time wasted on manual cross-referencing
  • Command-center style use:
    • officers watching “live” crime dashboards for fast situational awareness

Sources / presenters (as named in the subtitles)

  • Dr. Pranam Mati (Director General of Police, Police Computer Wing, Internal Security Division, Police Computer Wing & Criminal Investigation Department)
  • Shri Pisha Kumar (Deputy Inspector General of Police, State Crime Report Bureau, Bengaluru)
  • Rajiv Dash Sharma (Solution Architect, AI / Cloud Computing / Cyber Security)
  • Arjit (industry practice leader, Analytics & AI; full name not fully captured)
  • Dr. Monty (source/vision/closing; last name not fully captured)
  • Zoho / Catalyst representatives (developer relations / senior developer relations; individual name not captured)
  • Domain experts referenced but not fully named (e.g., Mahesh, Ranganat Sar, plus others)

Original video