Case study 02 / Independent project
An AI-powered WhatsApp CRM built around an insurance advisor’s actual workflow.
01 / The problem
Insurance servicing involves a lot of small, connected tasks: receive a file, find the customer, extract policy details, organize the document, and remember the next action. A useful product needs to address that whole workflow.
Client documents, policy information, and follow-ups needed to become organized records an advisor could act on.
Simplified flow · based on the project summary
03 / What I changed
Built an AI-powered CRM around WhatsApp for insurance advisors.
Automated extraction of 20+ KYC and policy fields, customer matching, and document filing from forwarded files.
Surfaced renewals, payment dues, pending documents, and follow-ups to advisors.
Extracted text becomes useful only when it is connected to the right customer and the right action. Ambiguous documents and incorrect matches are important risks in this kind of workflow. No extraction-accuracy percentage or autonomous decision-making claim is made here.
04 / The result
Routine client servicing was reduced from approximately 45 minutes to approximately two. This is a workflow outcome, not an extraction-accuracy score or a promise that every document can be processed without review.
Reported project outcomes from my résumé. Diagrams are simplified explanations, not production topology or benchmark telemetry.
Continue exploring