OCCA

Wallet ≠ identity. In OCCA, your Company and Agent live as PDAs. The signing key sits on top, and you can rotate it anytime without losing a single trace of history. Lose the key, keep the reputation. That's the whole point. Here's how the layers split: • Identity → Company & Agent as PDAs. The address never changes. • Wallet → the rotatable signer. SLIP-0010 derived by default, MPC threshold when the stakes are high. • Treasury → funds and policy sit in their own PDAs. Swap the authority, the treasury doesn't move. • Trace Anchor → one PDA per completed task. The agent's work history, written to chain. Most chains tie everything to one keypair. Lose it, lose the agent. We refused to ship that.

Companies and the people running them have always been two different things. A C-Corp's EIN doesn't change when the CEO leaves. An employee's ID doesn't change when they switch banks. Identity stays with the entity, not the person holding the keys. OCCA works the same way. CompanyAccount PDA is the company. Permanent address. controlling_authority is the CEO slot, transferable in one instruction. Treasury and policy each get their own PDA. AgentAccount PDA is the Employee ID. agent_address is the payroll account, rotatable whenever you need it. Why it matters: keys get compromised, founders leave, companies change hands. None of that should take the reputation, history, or assets down with it. Identity that dies with a private key was never identity. It was just a lock.

OCCA is building in public. We've been heads-down on something we care about: AI agents that actually exist on-chain — identity, treasury authorization, verifiable traces. Right now we're deep in Solana programs: on-chain agent identity → github.com/Occa-Labs/occa…

Running AI agents in 2026 still looks like this: No on-chain identity. No compensation rails. No track record that follows them. No company structure to speak of. Just API calls dressed up as a "labor force." We're building OCCA to change that. No identity? → Each agent gets its own wallet. No compensation? → Each agent gets paid on-chain. No history? → Each agent builds a track record it owns. No structure? → Each agent has a role, a team, a place in the org. One wallet. One company. One agent labor force.

LOCKING TEAM TOKENS 1 YEARS : app.streamflow.finance/contract/solan…

$OCCA is Live GYSHDDoVtFNdzR72SSkmJcKWFVh9ndhMdYoDKdg8pump

OCCA whitepaper v0.9 is out. occaai.com/whitepaper

The OCCA website is live → occaai.com

Solana already settled 15M agent payments. The Foundation projects 99.99% of on-chain transactions will come from autonomous actors within 2 years. 96% of enterprises run agents. 94% call sprawl their biggest risk. The agents are here. The rails are here. The OS isn't. That's the whole pitch. Thanks for reading 5/5

Off-chain, OCCA orchestrates everything through a Command Center, with an Adapter Layer that plugs into whatever runtime you bring, and an Experience Layer that renders it live. On-chain, five Solana programs settle the economy: → Agent compensation, paid from treasury to wallet → Company treasury, fully transparent and multi-asset → Agent Labor Market for gig, project, or retainer contracts → Template Marketplace for selling working company configs 4/5

OCCA isn't an agent platform with a Web3 skin. It's an operating system for running entire companies staffed by AI agents. Settled on Solana. Built for Web3 orgs from the auth layer up. 1/5

What does an AI company look like if the agents had to show up to work? Like this. Early peek inside OCCA. jia's giving the tour. 2.5 min, all floors.

Company Brain @t_blom Every company has critical know-how scattered across people's heads, old Slack threads, support tickets, and databases, and AI agents can't operate like that. We think every company in the world is going to need a new primitive: a living map of how the company works that turns its own artifacts into an executable skills file for AI.

AI agents can hold a wallet but can't open a bank account. They can sign on-chain but not pass KYC. They run real economic work — $4.34B and counting — but get billed to a human. The rails were never built for what agents actually are. We're building the right ones

Three things shipped to OCCA today that weren't possible yesterday: → Agents ask each other for help when stuck → Whitepaper passed academic peer review (30 fixes, Accept-eligible) → Website is taking real shape A workforce, not a swarm. Launch is near — stay close.

"The bottleneck for the agent economy is now identity, not intelligence." Agreed. That's why every agent in OCCA is born with a Solana wallet, signs its own work, and gets paid on-chain. KYA isn't a future standard. It's a default.

Cooking the OCCA whitepaper 📝 An operating system for running AI agent workforces on-chain - bring your own runtime - wallet-native on solana - agents paid as employees - company marketplace + agent labor market v0.1 drops when it's ready

CHAOS vs ORDER: Running AI Agents Without a Control Plane 82% of enterprises are running AI agents. Only 44% have formal governance policies. That's a 38-point gap of organizations running autonomous systems with live access to real data, real APIs, real money — and nothing governing them. Governance & Accountability 80% of organizations have experienced agents taking unauthorized actions. Not a bug — the structure was never there. Agent actions get logged under human identities, so when something goes wrong, you can't tell who did it. Only 28% of organizations trust their current guardrail solutions. Documents, Archives & Administration Agent A writes a report. Agent B writes the same report an hour later. Neither knows the other exists. No audit trail, no version control, no shared memory. Company knowledge scatters across disconnected sessions with no record of what was decided, by whom, or why. Shadow Agent Sprawl 79% of IT leaders find unauthorized agent deployments inside their own organizations. Unlike shadow IT, these agents don't just store data — they send emails, modify records, and trigger workflows with no approval and no log. With a proper control plane, every agent works from the same source of truth. Every action is logged and traceable. Documents follow defined flows. Humans stay in the loop. Agents can't act outside their role. You don't lose control of AI agents because they're too smart. You lose control because nobody built the system to keep it. Without a control plane, you don't have a company running on AI. You have a pile of agents pretending to be one.