AI Agents Go Autonomous: Who’s building the Agent economy in crypto?

The Spaces explored what AI agents mean in crypto, where value will accrue, what agents can already do, and what must exist for broader trust and adoption. Tim hosted a panel with Mike (Unity Labs), Joey (Fogo/Flogo), Ora (Kiora), Slava (PureFi Protocol), Kalish (WebProof Wizard/Cardell), and Peter (Perceptron). Speakers agreed agents are autonomous systems that perceive, decide, and act toward goals, well beyond chatbots. Enterprise adoption is early, constrained by compliance, security, and payments; blockchain can provide verifiable execution and value transfer. Value is likely to concentrate in infrastructure (networking, operating systems, execution rails), payments, and high‑quality data—now a “liquidity layer.” Practical wins today include monitoring, alerting, market/data scraping, DeFi co‑pilots, and airdrop automation; execution autonomy remains fragile. Trust requires identity, permissions, reputation, and better data quality and provenance; guardrails and scoped wallet access are essential. Tokens must link to real usage and revenue; recent AI token launches underwhelm, and sound tokenomics and market discipline matter. Looking 12 months out, panelists want to see agent‑to‑agent payments at volume, enterprise attribution to agent workflows, micro‑payments for content, stronger UX via LLMs, and tangible network KPIs.

Peanut Market Talks: AI Agents, Autonomy, and Crypto Value

Participants and Roles

• Tim (Host, Peanut Market Talks)
• Mike (CEO, Unity Labs) — rebuilding crypto primitives; launching Astro OS for agents; focus on verifiable execution, networking, and payments for agents
• Ora (Kiora) — building user-augmenting AI agent experiences integrated with existing wallets; developing Empire wallet extension
• Joey (Fogo) — growth/community; Fogo is a high-performance SVM L1 (Firedancer iteration) targeting ~40ms block times; building a native perp DEX; agent-friendly infra
• Kalish (“WebProof Wizard”, Web3 identity protocol; host referenced “Cardell”) — identity and trust layer bridging Web2/Web3 with ZK and MPC; building “QCat” AI-powered prediction market agent
• Slava (PureFi Protocol) — on-chain KYC since 2020; using agents/LLMs to replace manual workflows in compliance
• Peter (Perceptron Network) — decentralized data vantage network with ~700k nodes for distributed intelligence and lower-cost, higher-quality data access
• HyperGPT (connection issues, no remarks captured)

Working Definitions: What Is an AI Agent?

• Mike (Unity Labs): Enterprise lens. Autonomy means removing the human-in-the-loop while meeting compliance, policy, governance, and security constraints. Key blockers: compliance, security (prompt injection, DLP), and payments. Blockchains enable verifiable execution and value transfer but are only part of the stack.
• Joey (Fogo): More than a chatbot. A task-focused autonomous system that can execute specific, complex tasks on your behalf.
• Ora (Kiora): Simple test—if it completes a task on behalf of the user (even just retrieving answers from docs), it qualifies as an agent because it saves user effort.
• Slava (PureFi): Broadly, anything that can perform tasks autonomously with minimal initial human input. Shared the compliance challenge from practical experience.
• Kalish (WebProof Wizard): Agents perceive, decide, and act towards goals with minimal human intervention; they don’t just inform—they participate in economic processes. Most current “agents” are advanced automation; the trajectory is towards genuine economic actors.
• Tim (Host): A “real” agent couples autonomy, context, and action—decision-making and system interaction beyond Q&A.

Automation vs. Autonomy

• Kalish: The line is crossed when systems evaluate information, adapt to change, and independently carry out tasks end-to-end. Many marketed products are still automation; true autonomy reduces ongoing human decision inputs.

Where Will Value Accrue in the Agent Economy?

• Mike: Still too early to tell. Enterprises mostly operate agents in “single-player” mode (data analysis, support), not agent-to-agent ecosystems. Until inter-agent collaboration at scale exists, value allocation is unclear. Compliance is today’s primary blocker.
• Ora: Data and payments are valuable layers, but as agent capabilities advance, agents may increasingly build or replicate needed components in-house, reducing reliance on third-party agents.
• Tim: Data is becoming a new “liquidity layer.” Value likely distributes across trust/identity layers and payment rails rather than residing solely inside agents.
• Joey: High-quality data access and “bulletproof” execution infrastructure are critical. Failure modes in payments or state transitions are catastrophic; robust rails will capture significant value. We’re very early—akin to “Bitcoin at $10.”
• Slava: Macro caution. Large-scale tech layoffs suggest AI will compress costs for previously high-cost roles (e.g., developers). The macroeconomic transition—what displaced workers do next—will be a major societal impact of agents.

What Agents Can Usefully Do Today (Beyond Demos)

• Peter: Monitoring and alerting where trust and speed matter—e.g., funding rates across venues, whale wallet movements, mempool activity—surfacing actionable signals faster than humans. Execution is improving but still fragile; agents can route swaps, manage bridging/slippage, and orchestrate multi-step DeFi positions. Today they are excellent co-pilots for data-heavy workflows.
• Joey: Persistent data scraping (e.g., X/Twitter) for daily digests, risk flags, opportunity scans, and portfolio tracking. Effective for market research (filtering noise to find high-signal commentary), especially when “following clusters” of credible sources.
• Slava: Massive acceleration in blockchain data analysis (e.g., labeling, clustering, bad-actor detection). Agents/LLMs enable overnight reruns and rapid error correction vs. legacy months-long workflows. Given crypto’s sensitivity to infra failures (RPCs, gateways, DBs), agent-driven data reliability is a vital backbone.

Project Roles and Near-Term Roadmaps

• Kiora (Ora): Augmenting user workflows within existing wallets—no separate custody/delegation required. Building the Empire wallet extension with vertically integrated AI features. 2024 focus: ship something users love; iterate on user feedback; prioritize user-facing agent utility over agent-to-agent networks.
• Unity Labs (Mike): Launching Astro OS—an open-source agent “kernel”/OS that unifies memory, tools, and execution across frameworks (e.g., Hermes, Cloud/Claude, etc.). Aims to standardize secure, composable agent capabilities and power networking, payments, and verifiable execution layers.
• WebProof Wizard (Kalish): Trust and identity layer for agents: verifiable identities, permissions, accountability, and reputation with ZK and MPC. Building “QCat” AI prediction market agent. Belief: trust—not intelligence—will be the defining bottleneck.
• Perceptron (Peter): Distributed data vantage network (~700k nodes) leveraging user bandwidth/IP vantage points to deliver diverse, locale-accurate data at lower cost than centralized providers. Targeting business-intelligence products and cost-quality improvements for AI workloads.
• Fogo (Joey): SVM L1 with Firedancer iteration achieving ~40ms block times—suitable for high-throughput agent activity (trading, micro-payments). Building a native perp DEX; encouraging agent use cases on Fogo.
• PureFi Protocol (Slava): On-chain KYC/AML since 2020. Using agents to replace manual compliance workflows and to expand on-chain data labeling, clustering, and monitoring.

Early Revenue Touchpoints and UX Implications

• Ora: Clear near-term fit in trading automation, DeFi automation, and payments; uncertain on gaming.
• Mike: Crypto UX is a “nightmare.” Conversational/agentified UX for DeFi is low-hanging fruit (e.g., airdrop-hunting agents; “agent banks”; DeFi protocols rewritten for agent-to-agent interactions). Expect emergent behaviors once large user communities’ agents can network and transact.
• Tim: Regulatory/compliance acceptance will be pivotal for institutional trust in DeFi-agent implementations.

Trust, Security, and Data Preconditions

• WebProof Wizard: Users need a trust layer mediating agents and wallets—transparent rationale, permissions, auditability, and enforceable boundaries.
• Peter: Data quality and multi-vantage access are core bottlenecks. Centralized data licensing is prohibitively expensive (e.g., OpenAI’s large payments to X/Reddit). Decentralization can cut costs and broaden access. Strong guardrails are needed; he would not currently grant agents wallet access or broad data access on personal machines.
• Slava: Each additional layer expands the attack surface. Scoped permissions are key (e.g., allow gas top-ups/internal transfers but disallow withdrawals). Security is a double-edged sword: defenders and attackers both benefit from better tooling. Expect supply-chain risks (agents auto-fetching vulnerable libraries) to rise.
• Tim: Trust is layered and earned. Limited, structured autonomy is prudent before granting broader control.

Token Value: Where Should It Come From?

• Ora: Tokens should accrue value from real product usage and revenue. Classic crypto value accrual (e.g., buybacks) can complement, but utility must be genuine. Recent AI token launches mostly underwhelmed—likely market conditions.
• WebProof Wizard: Value should flow from actual activity—infra usage, execution fees, data services. Tokens should support the ecosystem, not substitute for a business model.
• Mike: Example of a strong infra token launch (Kite) for agent-to-agent commerce. Envisions new token use cases around verifiable execution (blockchain as a proof system), networking, and payments. As agent use cases evolve, novel token utilities may emerge.
• Joey: Skeptical unless the token has a clear, compelling purpose (e.g., network fees, feature access, revenue sharing tied to agent usage). Tokenomics scrutiny is much higher this cycle; justification is mandatory.
• Peter (added context in closing): Founders should deeply understand tokenomics, MM, and exchange dynamics; poor structuring can tank charts regardless of product quality.

12-Month “Real Market” Milestones the Panel Wants to See

• Peter:
  • Meaningful volumes of recurring agent-to-agent transactions beyond hackathon demos
  • Enterprises (funds, trading desks, insurers) publicly attributing decisions to agents running on decentralized infra
  • Perceptron KPIs: tens of thousands of daily paid data tasks; node yields; active “data questing” app participation
• Joey:
  • Tangible life improvements beyond crypto (e.g., health), and within crypto (risk management to avoid liquidations)
  • Builders shipping agent use cases on Fogo’s high-throughput L1
• Mike:
  • Transaction growth as the core metric of the agent economy
  • Early real economy: micro-payments for content/media/weather as agents replace ad-based models (agents have no “eyeballs”)
• Slava:
  • Drastic UX simplification via agents/LLMs for mainstream onboarding (explaining chains, tokens, fees, and bridging). Improved UX could meaningfully expand crypto’s user base.

Key Takeaways

• Definitions converged on autonomy, context, and action: Agents do work, adapt, and transact—not just chat.
• We are very early. Enterprises mostly use single-player agents; compliance is the main blocker to autonomy in production.
• Data is the new liquidity: quality, diversity of vantage, and cost-efficient access are fundamental; decentralization can help.
• Infra and trust layers matter: robust execution rails, identity, permissions, and verifiable execution will capture value.
• Practical utility today clusters around monitoring/alerting, research, and co-piloting complex DeFi flows; execution is improving.
• Security must be scoped and explainable: permissioning, audits, and guardrails before granting wallet control; anticipate supply-chain risks.
• Token value should tie to real usage and revenue; tokenomics maturity and market mechanics are decisive for sustainability.
• Near-term proof points: meaningful agent-to-agent commerce, enterprise adoption signals, micro-payments for content, and UX breakthroughs that make crypto approachable.