LeadFleet Agents
A fleet of always-on agents that watch for buying signals, draft the response, and queue it for your approval.
What it does
Job boards, forums, RFP feeds, freelance platforms and social, polled on your schedule.
Runs against your own model — no prospect data sent to a vendor.
Every outbound message waits for an explicit approval.
One screen for everything sitting in your queue.
Scout, Analyst, Responder, Promoter and Ops, each with its own job.
Systemd units, a FastAPI dashboard, and your own database.
The work of finding people who are actively looking for what you sell is not hard, it is just relentless. Somebody has to read the boards, the forums and the feeds every single day.
LeadFleet runs five agents on your own GPU box. Collectors watch job boards, forums, RFP feeds, freelance platforms and social; an analyst scores what they find against your profile; a responder drafts a reply in your voice; a promoter queues outbound; and an ops agent keeps the whole thing alive. Nothing sends without you saying so.
Specifications
| Runs on | Linux with an NVIDIA GPU |
|---|---|
| Model | Any Ollama or OpenAI-compatible endpoint |
| Dashboard | FastAPI on a port you choose |
| Sources | 10+ collectors included, more are a Python file each |
Questions about this product
Will it send emails on its own?
Only where the platform’s API allows it and you have explicitly turned that on. Everything else waits for approval.
Does my data go anywhere?
No. Scoring and drafting run against your own model on your own hardware.