Back to products

Live product case study

My Personal AI Job Agent

A personal AI agent that reads 8 job platforms every morning, filters against my real career context, and drops 50 ranked roles into my inbox so I spend time applying, not searching.

Live in production AI agent INR 0/month
Read the PRD View architecture

The one-liner

Built for one real user, with real constraints.

Generic job tools optimize for volume. This agent optimizes for signal: seniority, AI/product fit, geography, freshness, duplicate removal, and whether a role is actually worth Harshitha's morning attention.

8 Job platforms scanned daily
50 Ranked roles in the digest
70% Apply rate on surfaced roles
~5 Minutes to triage each morning
7 Days to build v1
INR 0 Current monthly running cost

Problem

Generic AI job autopilots do not know me.

Most job products are built to serve everyone, so they hedge. Out of 20 recommended jobs, only 5 to 7 felt genuinely relevant. The rest were wrong seniority, wrong domain, or wrong stage of career.

User

One person. By design.

The primary user is a senior specialist mid-search: sharp about what she wants next, short on time, uninterested in auto-apply, and looking for a shortlist that respects attention.

Experience

A daily email is the interface.

At 08:15 IST, one digest arrives with a clean spreadsheet attached. The top roles are already ranked, nothing repeats from yesterday, and every row has a direct apply link.

Why it is different

Personalization, not aggregation.

Generic AI autopilots

  • Built for every user on the planet.
  • Optimize for volume of matches.
  • Ranking is broad and shallow.
  • Auto-apply flows inflate activity metrics.
  • Noise rate: about 65% to 75% of results.

This agent

  • Built for exactly one real person.
  • Optimizes for signal over volume.
  • Ranking uses a real resume and preferences.
  • Human applies, agent curates.
  • Noise rate: about 30% and dropping.

Scope

What is in, and what is deliberately out.

The product is strongest because it is opinionated. It helps with daily discovery and ranking, but it does not replace judgment or pretend to be the person applying.

In scope

  • Reading across multiple job platforms every day.
  • Filtering stale, irrelevant, or out-of-context roles.
  • Scoring against a real resume and preferences.
  • Remembering seen roles so nothing repeats.
  • Delivering a clean, sortable shortlist by email.

Out of scope

  • Auto-applying on the user's behalf.
  • Writing cover letters or editing resumes.
  • Impersonating the user in recruiter conversations.
  • Replacing the judgment of the person searching.
  • Scaling to many users before the one-user loop is strong.

Production state

What is already live.

Live

Daily shortlist

Eight platforms are read every morning and the strongest roles are ranked into one email.

Live

Seen-role memory

The agent remembers what it has already shown, so repeated roles do not keep wasting morning attention.

Live

Fit-based ranking

Roles are scored against the real resume, seniority, AI/product relevance, location, freshness, and apply-readiness.

Live

Zero-cost operations

The system runs on free-tier services with GitHub Actions, SQLite-backed state, and email delivery.

Product proof

See the product thinking behind the build.

The PRD captures the product choices. The architecture page shows how the agent runs every day.

Read the PRD View architecture