ZobiqueContext
Vastav Intellect works in a specialised space where understanding a prospect matters as much as finding one. Two employees researched every incoming lead by hand, about ₹40,000 a month in staffing, and the work grew with every new lead.
Problem
The founder wanted to hand over a spreadsheet of leads and get back a short brief on each prospect and what to offer them. A general-purpose AI tool could do it for a few leads, but at hundreds or thousands it became too expensive.
What we built
We costed the options first, then split lead research into three stages, each handled by the cheapest tool that does the job well, run by two workers on a Django and MongoDB backend.
Discover: Serper
Web search as its own stage, not bundled into an all-in-one AI research call.
Extract: Jina AI
Turns the pages found into usable material for analysis.
Interpret: Gemini
A lighter model classifies findings and spots opportunities without paying for premium reasoning on every step.
Two-worker execution
Research and processing run without blocking each other.
- Upload a lead list
- Search the web
- Extract the content
- Classify and brief the sales team
Result
Estimated research cost fell to about ₹0.80 per lead, from about ₹2–₹3 with a web-search agent (roughly 60–73% less) and ₹5–₹8 with general-purpose AI research (up to about 90% less).
Our own estimates for this workflow, not independently audited. Actual cost depends on research depth, retries and usage.
See it
