Subnet 71: Leadpoet
Leadpoet is Bittensor Subnet 71. The Leadpoet README source describes it as a decentralized AI sales-agent subnet focused on lead generation and intent-driven sales workflows.
What Leadpoet Provides
The Leadpoet README describes the subnet as a decentralized AI sales-agent network focused on the top of the sales funnel, starting with high-quality lead generation and aiming toward a more automated sales engine over time. The subnet’s task is to find companies that match a buyer’s ideal customer profile (ICP) and show genuine buying-intent signals, with verifiable evidence for each.
According to the README, Leadpoet runs two complementary miner tracks on one subnet. In the Model Competition track, miners submit models that surface in-market companies from the open web for a given ICP; a model that scores above a daily open-source reference baseline by a set margin becomes the “champion” and earns rewards until a higher-scoring model takes its place. In the Fulfillment track, miners compete on real, paid client requests for fully enriched leads — company, contact, and intent evidence — and the top-scoring leads for each request earn rewards over a fixed runway.
Miner and Validator Roles
The README describes miners as participating in either or both tracks: building a qualification model that returns the best-matching companies for an ICP, or fulfilling live client lead requests end-to-end. Person-level contact details are out of scope for the Model Competition track and are layered on by Fulfillment miners.
Independent validators evaluate both tracks with the same scoring pipeline, judging results on ICP fit, data accuracy, and the quality of intent evidence, and set weights on-chain from those scores. The README notes that fabricated or mismatched evidence is penalized, so that scoring rewards genuine, verifiable signals.
Evidence Quality Context
Leadpoet’s central evaluation problem is not only whether a miner names a plausible company. The README frames the subnet around matching a buyer’s ideal customer profile and providing intent evidence that validators can score for fit, accuracy, and quality.
The model evaluation rules make that boundary more explicit. They describe validator checks for fabricated intent signals, wrong-company URL matches, generic marketing copy presented as intent, and other patterns that make a result unsuitable as a genuine client lead.
This gives Leadpoet a different scoring context from ordinary search or contact scraping. A miner’s submission needs to connect a real company to a buyer-specific intent signal, and the supporting evidence has to remain grounded in the source material rather than merely sounding commercially relevant.
Data Boundary Context
The Leadpoet privacy policy describes the subnet as handling B2B lead data rather than personal consumer data. It also separates participant identity from lead data: miner and validator participation is associated with wallet addresses, while submitted leads carry business contact information, source metadata, validation outcomes, and audit trails.
That distinction matters for understanding Leadpoet. Miner work is about surfacing and supporting business leads, while validator work is about checking provenance, accuracy, and usefulness. The public policy documents frame those checks as part of network operation, attestation, auditability, fraud prevention, and data-subject request handling.
The result is a subnet where reward weight depends on both commercial relevance and verifiability. Lead evidence that cannot be traced to a real source, does not match the requested intent, or presents generic copy as buying intent is treated differently from a well-supported lead that fits the buyer’s profile.
On-Chain Identity
Live SN71 subnet data is available on TaoStats. The source-backed lead-generation, evidence-quality, and data-boundary details in this article come from public Leadpoet materials rather than from live identity fields.
Relationship to Yuma Consensus
Subnet 71 uses Yuma Consensus to convert the lead-quality weight vectors that validators submit into the emission shares distributed to miners and validators within the subnet each tempo. The Yuma Consensus documentation describes how validator weight submissions are aggregated into consensus weights for each miner registered on the subnet.
In Leadpoet’s context, independent validators evaluate miner submissions across the Model Competition and Fulfillment tracks, scoring results on ICP fit, data accuracy, and the quality of intent evidence. Fabricated or mismatched evidence is penalized, so weight vectors reflect genuine, verifiable lead quality rather than volume alone. The Emission documentation describes how those consensus weights determine each participant’s share of the subnet’s accumulated emission each tempo.
Development Stage Context
The Introduction to Bittensor describes subnet development as moving from localnet to testnet and then mainnet. For Subnet 71, that sequence applies to the standard Bittensor lifecycle: localnet for isolated development, testnet for shared non-production testing, and mainnet for live operation with real emissions.
On mainnet, Subnet 71 is registered as the live production subnet at netuid 71. The Bittensor Networks reference separates mainnet, testnet, and localnet. Participation examples or emission outcomes from one environment should not be read as representing production subnet performance in another environment.
Reader Boundary
Subnet 71 Leadpoet should not be read as generic Bittensor subnet documentation, a CRM or outbound-email service, or a guarantee that any lead will convert. It names one subnet’s decentralized AI sales-agent market — miners surface companies and enriched leads matching a buyer profile and validators score them on fit, accuracy, and intent evidence — on netuid 71 (Leadpoet repository, Understanding Subnets, Glossary: Netuid).
A miner’s reward reflects validator-scored fit, accuracy, and intent evidence of the surfaced leads, not sales outcomes from contacting them (Leadpoet site — leadpoet.com).