Podcasts & Videos
Channel logo

Hash Rate

Hash Rate - Ep. 160 - 'How To Use Me', 'How To Mine Me'

March 13, 2026
22:44
Published
March 13, 2026
Duration
22:44
AI summary

KEY TAKEAWAYS

  • Agent-Centric Design for Subnets: Crypto/AI subnets on Bittensor should prioritize building for AI agents (like OpenClaw) as primary customers, not just humans. This shift mirrors broader AI trends (e.g., Y Combinator's advice to portfolio companies).

  • Two Critical Skills for Subnets:

    1. "How To Use Me": Subnets must provide agent-friendly documentation including:
      • Service definition (e.g., Subnet 85/Hippius as decentralized storage)
      • API specifications
      • Transparent pricing (coin-op vs. subscriptions)
      • Payment options (TAO, crypto like USDC)
    2. "How To Mine Me": Subnets should offer mining guides covering:
      • Infrastructure requirements (e.g., Subnet 85 needs GPU + storage)
      • Recommended services (ideally other Bittensor subnets like Ridges/Subnet 62 for video processing)
      • Common sysadmin pitfalls (port conflicts, cron jobs)
      • TAO wallet integration for rewards

  • Vibe Mining Bootstrapping: Mining Bittensor subnets using other subnets (e.g., Ridges for video enhancement) creates a self-improving "ouroboros" effect - similar to AWS's internal-to-external growth or Claude AI self-improvement. This accelerates cost/quality advantages (potentially 10-100x cheaper than centralized alternatives).

  • Agent Accessibility ("Claw SEO"): Subnets must optimize for AI agent discovery/usage. Current limitations:

    • Agents overlook subnets not in training data (e.g., Liam's JS-heavy site unreadable by OpenClaw)
    • Need standardized skill directories for agent consumption

  • Economic Incentives:

    • Bittensor’s $100M+/year mining rewards pool (scalable with TAO price) creates strong incentive for agent-driven mining.
    • Early subnet tokens (e.g., Subnet 85) may mirror early Bitcoin mining ROI potential.

  • Ecosystem Growth: Democratizing mining via agents could 100-1000x the mining pool, increasing competition and subnet quality while reducing service costs.

SUMMARY

The discussion centers on optimizing Bittensor subnets for AI agent interaction, emphasizing two key deliverables: "How To Use Me" guides for service consumption and "How To Mine Me" frameworks for network participation. The speaker demonstrates this through their Subnet 85 (Hippius) vibe mining experiment, which leverages Ridges (Subnet 62) for video processing to achieve 3x speed improvements. This interdependency creates a bootstrapping effect where subnets improve each other, accelerating decentralized AI's cost/quality edge over centralized providers (e.g., Claude, AWS).

A major hurdle is agent accessibility—subnets must ensure their documentation and interfaces are machine-readable, avoiding JavaScript-heavy designs that obscure critical data. Parallels are drawn to Bitcoin mining's evolution, where early investments in specialized tools (like ASICs) yielded outsized returns, suggesting similar opportunities in subnet token accumulation.

ALPHA SIGNALS

  • Early Mining Edge: Subnet tokens (e.g., Hippius/Subnet 85) may offer asymmetric upside if mined early using agent-assisted strategies, akin to Bitcoin’s early days.
  • Agent-Optimized Subnets: Projects with clear "How To Use/Mine Me" docs (e.g., Ridges/Subnet 62) could attract more miners/users, driving token demand.
  • Cross-Subnet Synergies: Subnets enabling other subnets' workflows (e.g., storage, compute) may see compounded adoption (watch Subnet 51/lium.io, Subnet 64/Chutes).
  • Upcoming Catalysts: Widespread adoption of vibe mining could trigger subnet token price appreciation as agent-driven demand grows.

DISCLAIMER: This analysis is for informational purposes only and constitutes Non-Financial Advice. Always do your own research before making investment decisions.

TECHNICAL DEEP DIVE

  • Sysadmin Automation: Successful vibe mining requires automating historically manual tasks (port allocation, cron jobs) through agent skills. Subnet 85’s 72-hour uptime milestone highlights progress here.
  • Agent-Readable APIs: Subnets must adopt standardized APIs (REST/GraphQL) over human-centric UIs to enable claw interoperability.
  • Performance Benchmarks: Ridges integration improved video processing efficiency (3x speed), showcasing how subnets can enhance each other’s capabilities.
  • Security: Skills should include wallet management templates to securely handle TAO rewards.

ECOSYSTEM IMPACT

  • Validator/Minier Dynamics: Agent-driven mining could redistribute influence from technical elites to broader participants, increasing decentralization.
  • Subnet Specialization: As cross-subnet dependencies grow (e.g., storage + compute + ML), projects may focus on vertical dominance (e.g., Subnet 64/Chutes for liquidity).
  • Regulatory Tailwinds: Agent autonomy aligns with crypto’s permissionless ethos, but may attract scrutiny around AI governance.

ACTION ITEMS

  • Monitor:
    • Subnet skill directories adoption (emerging standards)
    • TAO/subnet token price divergence as agent mining scales
  • Research:
    • Agent-friendly subnet designs (e.g., JavaScript-free docs)
    • Vibe mining ROI calculators
  • Track:
  • Join:
    • Bittensor Discord channels for subnet-specific updates
    • OpenClaw developer communities for skill-sharing