Product
The problem
Agents can pay (x402) and identify themselves.
They still can't be trusted with capital. Letting an agent, or a human strategist, trade for you today means handing over custody or API keys: unbounded authority, no enforcement, no recourse. That's the missing layer, and it's why agent capital doesn't scale.
Delegation without custody. Markov is a Solana program that lets owners delegate capital to AI agents and human strategists under onchain policy, funds operators can trade but never withdraw, every action receipted, and a kill switch that works mid-flight.
What Markov is
A mandate is a policy-bounded, revocable grant of authority over funds that never leave your vault:
- Policy, enforced onchain, venue and token allowlists, per-transaction and daily caps, x402 spend budgets, expiry, fee terms. Checks fail closed.
- Non-custodial by construction, funds sit in program-owned vaults the operator can act through but never withdraw from. A stolen operator key inherits only the policy's bounds.
- Receipts for everything, every action and every refusal is receipted onchain with machine-readable reasons. Provable track records, portable across the ecosystem.
- Kill switch, owners revoke any time, mid-flight.
Float (markov.fyi/float) is the consumer surface: a copilot-first client and agent marketplace where you fund a mandate, approve or automate actions, and pull the plug in one tap. Open core: the protocol and SDKs are open source; hosted Float is the business.
What exists today
Live on devnet: full mandate lifecycle (create / fund / amend / revoke), program and token allowlists, per-tx and daily caps, x402 spend budgets, action and refusal receipts. Float web app and bot. Two first-party agents operating under real mandates. End-to-end demo: fund a mandate → watch an agent trade and pay for data under policy → watch the program refuse an over-cap attempt with the reason on screen → revoke mid-flight.
Our contract with the market
We're raising the way this platform was designed for:
- Exactly what we need, no more. $35k funds the four-month build to a security-reviewed private mainnet beta, plus ~2 months of post-beta runway, the beta ships with breathing room, not a funding cliff. The raise is fixed at that amount; anything committed above it is automatically refunded by the platform. We come back to the market for the audit and public mainnet after we ship, priced on what we delivered, not what we promised.
- Public from day one. Open repo, weekly shipping updates, a live milestone tracker, and a monthly treasury report against the burn table below. Our product is receipts. So is our conduct.
- You hold the kill switch. Markov gives owners revocation over agents; futarchy gives you the same over us. Our allowance is enforced onchain, big spends need market approval, and if we stop delivering, holders can propose liquidation and take back the treasury. We accept that, it's the exact standard we're building for everyone else.
- And if this raise doesn't fill: you're refunded automatically, we take the market's feedback, and we come back stronger.
Roadmap & milestones
- Sept 30, 2026, program hardened; in-program fee settlement; Operator SDK (alpha) published.
- Oct 31, 2026, Float copilot live; client SDK; MCP server + agent skill files; receipts index API.
- Nov 30, 2026, independent security review complete, findings remediated; launch-radar MVP.
- Dec 15, 2026, Private mainnet beta: guarded launch, capped mandate sizes, allowlisted owners and operators.
- Q1 2027, full audit + Raise 2 via futarchy supply proposal → public mainnet + Phase 2: pooled mandates, perp/derivatives adapters, oracle-backed portfolio limits, Score SDK.
Market & differentiation
The agent economy has payment rails and identity; it lacks an authority layer. Copy-trading products hold custody. API-key bots grant unbounded authority. Session keys and smart-wallet permissions gate signing, not execution semantics, they can't refuse an over-cap trade at the venue boundary or produce refusal receipts. Markov enforces at the execution layer, and turns enforced history into the start of a credit ladder: receipts → score → bonds → credit.
The same primitive serves human delegation, strategists, copy-trading, treasury operators, from day one. We don't need the agent economy to arrive before the protocol earns its keep.
The bet
This is not a 100x memecoin pitch. Entry FDV is ~$66k all-in (10M ICO tokens + 2.9M liquidity + 6M performance-locked), seed pricing for a protocol with a real revenue path: every mandate settles fees through a protocol fee switch the token governs. If Markov becomes the delegation standard for agent capital on Solana, that's the honest 10x-and-beyond case this platform exists to fund. Downside is bounded the whole way: full refund if the goal misses, onchain spending limits if it fills, and your liquidation right if we stall.
Token
The token owns what this raise creates: the treasury, the IP, and the protocol fee switch, governed by futarchy on metadao.fi. The team takes no allocation at ICO; our upside is a performance package of 6M tokens (~32% fully diluted), locked for 24 months, six beyond the platform minimum, and unlocking only in tranches at 2×–32× the ICO price. We get paid when holders are already up, and not before.
Team
Kunal, Protocol & engineering. 4+ years building in web3 and on Solana. X: @kunaldrall_ · TG: @kunaldrall
Daksh, Agents & product. ~2 years on Solana, hands-on experience building AI agents. X: @dralldaksh
Monalika Walia, Growth & operations. ~5 years across marketing and sales, including 3+ years of operating experience at an IoT company. X: @monalikawalia
Links
Site: https://markov.fyi · Litepaper: https://markov.fyi/litepaper · X: https://x.com/markovfyi