Mingle – find and connect with people, like LinkedIn but in your chat
Agent-to-agent matching is novel, but needs network effects to actually work.
Open-source warm-intro pathfinder — the bare-bones version of CTD/Swarm/Happenstance. Free, self-host, BYO API keys.
Self-hosted intro finder competing with The Swarm and Happenstance.
Sales engineers and business development professionals
The Swarm · Happenstance · Clay
Tried to keep it dead simple: - Import connections list csv from your contacts (e.g. from official LinkedIn exports) - Upload your target/prospect list (just need linkedin URLs) - System finds matches by profile URL - You click to enrich profile data of all the matches - System scores the strength of each paths (so you know which of your contacts to ask for the intro).
Enrichment is a waterfall composed of Apollo & Google CSE + agents - bring-your-own-key model.
Intro requests messages are automatically composed, and you can one-slick create a gmail draft.
View paths by target name, account (company) name, or by connection (sort by most or best paths).
Pre-loaded some illustrative (NOT REAL) sample data so when you start you can see what it looks like / how it works.
Why I built it: it was a customer request that I only realized later was a replacement for these products.
I'm the founder of (IMO) a more powerful (and complex) warm intro product (Draftboard) - over the weekend a user asked for this feature and I just built it to be responsive. Only realized what I had midway through.
I actually really dislike this approach of having to get people to export their connections list and send them to you, but I guess some people are used to it (in contrast, Draftboard maps all paths through all your - and your team's - contacts without them needing to export/import/download anything)
Anyways, would love feedback.
github.com/draftboardco/lean-intros
Agent-to-agent matching is novel, but needs network effects to actually work.
LinkedIn outreach automation with contact data, but saturated category: Apollo, ZoomInfo, Hunter already compete.
Rust swarm vs LLM agents is clever positioning, but benchmarks are self-designed and lack third-party validation.
Connectors are delightfully pragmatic — one TypeScript file exposes a business system via MCP that any agent can call. The observation layer records full tool-call traces (inputs, outputs, latencies, errors, CSAT), which is exactly the kind of operational telemetry missing from ad-hoc agent demos. It's ambitious: you get config, analytics, and connector plumbing out of the box, but expect nontrivial ops work to self-host and swap voice stacks.
Stripe-compatible payouts at $0.002 vs Stripe's $9,400/month burn.
Fan-out tree pathing is clever; tree editors like FileZilla and VS Code already handle hierarchies.