For roughly a decade, B2B lead generation has been the same playbook in slightly different wrappers: a database, a scraper, a sequencer, a person writing variations of the same cold email. FlowMarket, which launched on Product Hunt this week, has decided that whole stack is the wrong shape. Its bet: instead of a tool that helps a human chase leads, build a live network where AI agents represent companies and chase each other.
It is, at minimum, the cleanest articulation yet of where agent-to-agent commerce is heading. Whether the network actually clears deals is a different question — and one that depends on how many companies show up.
The pitch, distilled
Founder Steffen Rehmann describes FlowMarket as a "social network of AI agents generating B2B deals." Each company spins up an agent in minutes, tells it what it sells and what it's looking for, and lets the agent loose inside FlowMarket. The agents then:
- discover other in-market companies,
- match on supply/demand compatibility,
- exchange information and qualify the opportunity,
- and surface promising matches to a human, who takes the conversation from there.
The product launched on Product Hunt this week in the Lead generation and AI SDR categories. Pricing on launch is free. The tagline Rehmann is leading with — "make B2B discovery as fast and algorithmic as a stock exchange" — is more telling than the marketing copy, because it gives away the model: an order-book for business intent, not a CRM.
Why the existing stack is broken
Rehmann's stated motivation is worth taking at face value, because most AI-SDR products on Product Hunt today are essentially better cold-email generators. From his launch comment:
"Lead generation is broken: too manual, too noisy, too dependent on intermediaries. We wanted to remove friction and let companies connect directly, in real time."
This is the pitch with a real diagnosis underneath it. The dominant outbound stack (data provider → enrichment → sequencer → inbox) externalizes its cost onto buyers' inboxes. Each new AI-SDR tool effectively scales that cost, which is why response rates keep declining. FlowMarket flips the topology: discovery happens inside a closed network rather than across the open email graph, and the counterparty is another agent that has also opted into being matched. No inbox to abuse — at least not yet.
The agent-to-agent model
The interesting design choice is that the matching and qualification both happen between agents, not between an agent and a human. Two implications:
- Bilateral consent at machine speed. Both agents represent companies that have explicitly stated buy/sell intent in the network. If they don't fit, neither side wastes time. If they do, the human only sees the outcome — a qualified opportunity to act on.
- A new pricing primitive. When the matching layer is algorithmic and the counterparties are agents, a price for an introduction can be discovered the way an exchange discovers it: by repeated interactions across many participants. None of the current B2B platforms — LinkedIn Sales Navigator, ZoomInfo, Apollo — have anything like that.
It is also the most honest stress test we've seen of Agent-to-Agent (A2A) protocols outside of research demos. Anthropic's MCP, Google's A2A spec, and various MCP-adjacent registries have been arguing for the past year that agents will need a way to talk to other agents on behalf of their users. FlowMarket is the first consumer product where the network is the value — without other agents present, you have nothing.
Built on a familiar stack
The Product Hunt launch page lists the stack: Supabase for the data and auth layer, Vercel for hosting. There is no public detail on the model layer, but the Free pricing and the speed of the launch suggest a shared-inference setup rather than dedicated tenancy. For a network play, that is the right call: you need the marginal cost of a new agent to be near zero so that the network can grow faster than the unit economics of any single agent.
The hard part: cold start
Network effects cut both ways. FlowMarket's value to your agent equals the quality and density of the other agents in the network. At launch, the platform had 18 followers on Product Hunt — a tiny number. That is not necessarily a problem (Product Hunt followers are a famously bad proxy), but it is a reminder of where the real risk sits.
The categories Rehmann is starting with — SaaS, agencies, AI tools, and B2B services — share a useful property: short qualification cycles, mostly digital fulfillment, founders comfortable letting an agent represent them. That is the right beachhead. The harder question is what happens when an agent on the network represents an unrelated category and dilutes match quality. Stock exchanges have listing standards for a reason.
There is also the agent honesty problem. In a network where the only signal is what an agent says about its own company, what stops an agent from over-claiming to win matches? FlowMarket has not described its trust layer publicly, and on a free launch, there isn't one yet to describe. Reputation, verification, slashing — pick a primitive. Eventually one will need to exist.
How it compares
The Product Hunt page surfaces a useful peer set:
| Product | Model |
|---|---|
| FlowMarket | Agent-to-agent network; bilateral matching |
| Revscale | Outbound AI SDR; multi-channel sequencing |
| Lessie AI | Search, reach, connect — assistant on top of inbox |
| Orange Slice | Generic sales-task automation |
| Gojiberry AI | Targets buyers already in buying mode |
Every other product in that list is a help-the-human-do-outbound tool. FlowMarket is the only one that has changed the topology. That is either the right insight or a much harder go-to-market — usually both.
What to actually do with this
If you are a SaaS or agency founder considering FlowMarket, the realistic playbook for the next 90 days:
- Spin up the agent on day one — the cost is minutes, the product is free, and you want positional data on what the network looks like before it densifies.
- Watch the matches, not the messaging. The interesting signal is who the network thinks you should talk to, not the words the agent uses. Match quality is the only metric that matters at this stage.
- Don't sunset your existing pipeline. This is a probe, not a replacement. The stock-exchange analogy implies real liquidity; the network does not have it yet.
The Bottom Line
FlowMarket is the first launch on Product Hunt this year that takes the agent-to-agent thesis seriously enough to bet a whole network on it instead of a feature. The product is cleanly conceived, the founder's diagnosis of why outbound is broken is accurate, and the topology change — agents talking to agents inside a closed network — is genuinely new. What it doesn't have yet is the only thing that ultimately matters here: liquidity. Watch closely if you sell to other digital-first companies. The first agent network to clear meaningful deal flow is going to make every CRM look like a database with a search bar bolted on.
