▤
Fragmented documents
Purchase orders, invoices, delivery challans and inspection records may be reviewed separately.
AI-assisted material inspection
Make every material delivery verifiable.
GridStore AI is a proposed inspection and reconciliation assistant that connects purchase orders, delivery documents and visual evidence into one traceable review workflow.
TISC 2026 concept proposal · Individual innovator · Hyderabad, Telangana
DELIVERY VERIFICATION FLOW● Concept workflow
01
Purchase orderExpected item, code & quantity
↓
02
Delivery evidenceChallan, invoice, labels & photos
↓
03
AI-assisted comparisonOCR extraction + configurable rules
↓
04
Human review & decisionExceptions, notes & audit trail
The opportunity
Incoming-material checks deserve one connected workflow.
The concept focuses on the handoff between supplier delivery, inspection and store acceptance.
⌕
Manual comparisons
Teams may need to compare descriptions, quantities, labels and visible condition across different sources.
◎
Traceability gaps
Evidence, discrepancies, reviewer notes and approvals can be difficult to track consistently.
Proposed solution
An AI-assisted material inspection and reconciliation assistant.
GridStore AI is designed to support inspectors with structured evidence and exception flags while keeping authorised personnel in control.
▧
Capture
Upload or capture purchase orders, invoices, delivery challans, item labels and material photographs.
✳
Extract & verify
Use OCR and configurable rules to compare item identity, quantities, codes and selected specifications.
☷
Review & record
Highlight possible mismatches and create an inspection record with evidence, reviewer notes and outcome.
User workflow
From delivery arrival to a documented decision.
A practical first workflow for a store officer or material inspector.
STEP 01
Select PO
Choose the purchase order or delivery reference.
STEP 02
Capture evidence
Upload documents and photographs of labels, packaging and delivered items.
STEP 03
Extract details
Read item descriptions, codes, quantities and serial numbers using OCR.
STEP 04
Compare
Compare extracted values with PO data and configured criteria.
STEP 05
Resolve exceptions
Let the inspector review flags, add notes and attach supporting evidence.
STEP 06
Record outcome
Generate a report and record authorised acceptance, hold or rejection.
Minimum viable product
Start narrow. Validate with real workflows.
The first milestone is a software-only proof of concept using synthetic or permissioned sample data.
- Upload sample PO and delivery documents
- OCR extraction for item name, code, quantity and serial number
- Configurable comparison rules
- Exception dashboard with evidence references
- Inspection summary with human reviewer notes
PHASE 1 — DEMODemonstrate the end-to-end workflow with sample data.
PHASE 2 — VALIDATIONReview with domain experts and measure extraction and mismatch-detection quality.
PHASE 3 — PILOT READINESSExplore approved ERP / SAP / store-system integration and security requirements.
Responsible by design
Evidence first. Human authority always.
AI outputs are intended to assist review, not replace qualified inspection or statutory acceptance processes.
Human authority
Final acceptance, rejection and safety certification remain with authorised personnel.
Evidence-linked flags
Exceptions should link to the source document, image, extracted field or configured rule.
Data protection
Use approved data with appropriate access controls, retention rules and audit logging.
Computer vision may help identify visible anomalies; it cannot replace laboratory tests, certified electrical testing or applicable technical standards.
KT
Krishnateja Malakajgiri
Founder · Individual Innovator
Hyderabad, Telangana, India
Founder & execution
Domain-led concept, built in stages.
The founder brings 5+ years of Manhattan Active Warehouse Management techno-functional experience, including implementation, configuration, testing and production support.
Current status: idea / concept stage. A working prototype, customer deployment, funding or registered entity is not being claimed.
The next steps are to validate the workflow with relevant stakeholders, build a focused document-to-PO verification demo and define a measurable pilot path.