Investigate. Verify. Assess.
Smart India Hackathon (SIH) — Problem Statement #188
TEHQIQAAT is an agentic AI-powered document screening platform designed to assist authorized personnel in analyzing passports, visas, identity documents, permits, and other travel documents.
Instead of following a fixed verification pipeline, TEHQIQAAT uses an AI agent that can decide what to check next, adapt to intermediate results, handle failures, and generate an explainable risk assessment.
High-volume border checkpoints process thousands of identity and travel documents every day. Manual verification can be slow and may miss sophisticated document tampering, identity impersonation, altered photographs, inconsistent information, and expired or invalid documents.
Existing systems often rely heavily on manual inspection and basic database lookups.
TEHQIQAAT turns document verification into an adaptive investigation workflow.
The agent can coordinate multiple tools such as:
- OCR & information extraction
- Document validation
- Tampering detection
- Face verification
- Cross-document consistency checks
- Image enhancement
- Risk assessment
The agent dynamically selects actions based on what it discovers during the screening process.
For example, low OCR confidence can trigger image enhancement and re-extraction, while suspected photo manipulation can trigger additional forensic analysis and face verification.
TEHQIQAAT demonstrates:
- Goal-Driven Execution — works toward completing a defined screening objective.
- Dynamic Action Selection — chooses appropriate verification tools based on evidence.
- Multi-Step Execution — coordinates multiple verification operations.
- Adaptation — changes its workflow when new or unreliable results appear.
- Robustness — retries failed operations or escalates unresolved cases for human review.
- AI-assisted identity and travel document screening
- OCR-based document information extraction
- Document validation and consistency checking
- AI-assisted tampering and forgery analysis
- Face verification
- Cross-document analysis
- Explainable risk scoring
- Adaptive investigation workflow
- Human-in-the-loop review
Detailed documentation covering the system architecture, agent workflow, modules, decision logic, AI models, risk assessment, and implementation details is available in the docs/ directory.