Launching imast.ai: an AI HR agent for sourcing and outreach

I’ve been quietly building imast.ai for the last several months and it’s now live. imast is an AI HR agent for sourcing, screening, and engaging talent — one chat box that searches 800M+ profiles across LinkedIn, GitHub, Crunchbase, and 30+ other sources, ranks them against your criteria, and runs personalized outreach on autopilot. 400+ recruiting teams are already using it.

Modern sourcing is split across half a dozen tools. You write boolean strings in LinkedIn Recruiter, dump candidates into a spreadsheet, hand-write outreach in a separate sequencer, and chase replies in your inbox. The pipeline works, but most of the recruiter’s day is glue work between tools.

The shape of an agent is the right fit for this problem. Search, ranking, and outreach all need the same context about a role and a candidate. Once the agent owns that context, the recruiter’s job collapses from operate ten tools to steer one.

Three steps from open req to first reply:

• Describe the role in plain English. imast parses a sentence or a JD into a structured search across 800M+ profiles, unifying every candidate into a single record across sources.
• Review the AI-ranked shortlist. Define your criteria once. The agent assesses every candidate against your bar and returns the best fits with reasoning per skill — no boolean gymnastics, no keyword roulette.
• Engage on autopilot. Multi-step email sequences are written candidate by candidate. imast tracks replies, books intros, and learns what converts in your industry.

Every chat saves into a candidate table you can sort, enrich with custom columns, and continue without losing context. Push to Greenhouse via the Harvest API today; Ashby and Lever native push are on the 2026 roadmap. CSV export works with every ATS.

Solo recruiters running a few reqs, in-house teams running many, and agencies that need to move fast. Three plans:

• imast Start — $50/mo, 25 searches.
• imast Pro — $200/mo, 100 searches.
• imast Enterprise — custom, for in-house and agency teams.

A search returns up to 100 ranked candidates. Refining the same search — adding filters, narrowing geography, or asking the agent for “more like candidate #3” — does not consume an extra credit. Monthly billing, cancel any time.

The interesting research questions around imast all live at the intersection of agents and trust: how do you give a recruiting team enough control to feel comfortable letting an agent send messages on their behalf, while still keeping it autonomous enough to be useful? That’s where most of the upcoming work is — tighter feedback loops on ranking, more transparent reasoning per candidate, and safer outreach defaults.