Areté · AI systems studio · Hyderabad
Custom AI systems that do real work in your business.
Not chatbots. Not demos. Working systems that audit, extract, and
automate — proven on your own data before you commit to anything.
The proof-first offer
I build a working proof-of-concept on your real data before
you commit. If it doesn’t demonstrate value, you owe nothing.
01 What I’m building
Current systems
Revenue-leakage auditor for an 80-store franchise group
Unifies sales and inventory data across the group’s different store software
systems, then hunts the leakage patterns nobody can see from inside one store —
billing manipulation, inventory shrinkage, discount abuse — with rupee figures
attached to every finding.
Pilot in discovery
Case intelligence for a law practice
Case management built around context preservation: scheduling, documentation,
and AI summaries sharp enough to brief a newly assigned associate on any case
in minutes instead of days.
Scoping
Document extraction for messy invoices
Custom extraction pipelines for businesses drowning in inconsistent invoice
formats — built against each client’s real documents, not a generic template.
Discovery conversations
02 Experience
Built and shipped
-
Natural-language → SQL agents
Conversational interfaces over live business databases: ask in plain English,
get governed queries and grounded answers.
-
AI test-automation agents
Agents that drive real software — exploring flows and catching regressions
without hand-written test scripts.
-
An always-on personal agent, in production
A 24/7 assistant on a $6 server: durable git-backed memory, guarded tool
calls, self-healing deploys. Small, boring, and it never loses a fact.
Python · TypeScript · Claude & Gemini APIs · agentic tool use · SQL · Linux + systemd
03 Writing
Ground-level notes
Lessons from building these systems — what worked, what broke,
and what it actually cost. Published as the work happens, not as takes.
Follow along on LinkedIn