Your DAO Treasury, Monitored and Protected by AI
Live Demo: https://usewardex.xyz
Demo Video: Watch on Loom
Pitch Video: Watch on Loom
Built by: @jalaal_tweets
Built for the Colosseum Frontier Hackathon 2026 — Powered by Superteam UK
🎓 University submission — Built by an MSc Business Analytics student, Northumbria University. Eligible for the Colosseum University Award.
DAOs collectively manage $24.5 billion in treasury assets — and most of it is structurally at risk.
- 81.67% of DAO treasury assets are held in a single native governance token (Schellinger, Fiedler & Steinmetz, Blockchain Research Lab, 2023)
- Native tokens suffer 70–90% drawdowns in bear markets — the 2022 crash destroyed tens of billions in DAO purchasing power
- Idle stablecoins earn zero yield, leaving millions on the table annually
- The only automated treasury tool for mid-tier DAOs — Parcel — shut down in 2025. Tally shut down March 2026.
- Jito's JIP-24 (passed 2025) routes ~$15M+/yr in Block Engine fees to the DAO treasury. It is accumulating with no deployment strategy.
The structural trap: DAOs want to diversify, but every proposal to sell native tokens signals a vote of no confidence. A governance vote to diversify can suppress the token price before it even passes.
Wardex is an AI agent that reads real on-chain treasury data via Solana SPL Governance (Realms), scores bear market risk, and either generates ready-to-approve governance proposals or — once trust is established — executes pre-approved treasury actions within a non-custodial mandate.
The core insight: DAOs don't need an agent that makes financial decisions for them. They need an agent that executes the financial decisions they've already made but keep failing to act on.
The DAO votes once on a treasury policy. Wardex monitors and executes continuously within that mandate.
RPC Fast → High-performance Solana mainnet RPC for treasury reads
Realms SPL Governance → Real on-chain governance account reads (mainnet)
GoldRush / Covalent → Real SPL token balances with live USD pricing
DefiLlama API → Live TVL and 90-day historical TVL array
DefiLlama Yields API → Live Kamino / Marginfi / Drift APY rates
Jupiter Lend API → Live JupUSD vault supply rates as additional yield venue
TVL Trend Engine → 30d/90d % change + slope from historical TVL data
Risk Analysis Engine → Concentration score, 3-scenario bear market modelling
Claude API (Haiku) → AI-generated Realms governance proposals on demand
SNS Identity → wardex-agent.sol — verifiable onchain agent identity
Agent Execution → Real Solana devnet transaction via Memo program
- On-chain read — reads real governance accounts from Solana mainnet via SPL Governance (Realms) for Marinade, Jito, Jupiter, and Raydium. Powered by RPC Fast for speed;
getProgramAccountsfalls back to a public RPC (not available on RPC Fast Start plan) - TVL data — pulls live total value and full 90-day historical TVL array from DefiLlama
- TVL trend analysis — computes 30d/90d % change and daily slope from the historical array; shown as a directional indicator in the stats grid
- Token balances — fetches real SPL token balances and live USD pricing via GoldRush (Covalent); replaces estimated ratios when coverage is ≥5% of TVL
- Live yield rates — fetches current Kamino USDC/SOL, Marginfi, Drift, and Jupiter Lend APYs; best venue is selected automatically
- Risk scoring — calculates native token concentration, three bear market scenarios (Base −40%, Moderate −60%, Severe −80% based on 2022 historical data), runway estimate, and annual yield opportunity
- Governance proposal generation — Claude Haiku generates a fully-formed, Realms-ready governance proposal from the treasury analysis.
- Agent recommendation — generates a specific, actionable recommendation with the best available yield venue
- Real execution — broadcasts a verifiable Solana devnet transaction via the Memo program with the agent's
wardex-agent.solSNS identity embedded, returning a real Solana Explorer link
| Solution | Chains | Active Management | Min Treasury | Per-Action Vote | AI-Driven |
|---|---|---|---|---|---|
| Karpatkey / kpk | EVM only | Yes (human) | ~$100M+ | No (policy-based) | No |
| Avantgarde / Enzyme | EVM only | Yes (human + vault) | ~$50M+ | No (policy-based) | No |
| Realms (SPL Governance) | Solana | No | None | Yes — every action | No |
| Squads Protocol | Solana | No (manual via SquadsX) | None | Yes — every action | No |
| Castle Finance | Solana | Limited (3% APY vaults)¹ | None | No | No |
| ⬡ Wardex | Solana | Yes (AI agent) | $1M+ | No (policy-based) | Yes |
There is no Karpatkey for Solana. Wardex fills that gap — serving the 50+ Solana DAOs with $1M+ treasuries (representing $15–25B in collectively managed assets, per DefiLlama) that Karpatkey structurally ignores, on the chain where the most active governance communities live.
¹ Castle Finance last verified active May 2026; limited to static vault deposits with no AI-driven rebalancing or governance integration.
| DAO | Realm (Mainnet) | Why Now |
|---|---|---|
| Jito | jjCAwuu... |
JIP-24 routes 100% of Block Engine fees (~$15M+/yr) to DAO treasury. Idle SOL accumulating with no deployment strategy. |
| Marinade | 899YG3y... |
MIP proposals already passed to diversify away from MNDE concentration. mSOL/USDC treasury needs active management. |
| Jupiter | 2Z5BXuR... |
Treasury locked until 2027. New governance model being designed. Wardex is the execution layer for when it reopens. |
| Raydium | 8JZdqeT... |
On-chain governance launched 2025. Treasury management practices still immature. First-mover opportunity. |
All realm pubkeys verified on Solana mainnet, May 2026.
The most common question from DAO treasury managers: "Do you have custody of our funds?"
The answer is no — and the architecture is designed so that trust is earned in stages, never assumed upfront.
Wardex reads the DAO's on-chain treasury, monitors conditions 24/7, and when a policy threshold is crossed, generates a fully-formed Realms governance proposal — risk data, recommended action, yield comparison, rationale. The DAO's treasury committee reviews and approves it. The DAO signs and executes it themselves. Wardex never touches a key.
This is the initial onboarding phase for every new DAO. Zero execution risk. Zero custody. The analytical bottleneck is removed; control stays with the community.
After 3–6 months of clean Phase 1 operation, Wardex can be granted a Squads delegate key — but with hard constraints enforced on-chain:
- Destination whitelist only — Wardex can only call
deposit()on pre-approved protocol addresses (e.g., Kamino Earn vaultxyz...). It cannot transfer funds to arbitrary wallet addresses. Even if Wardex's infrastructure were compromised, an attacker could only deposit the DAO's funds into Kamino — not exfiltrate them. - Daily volume cap — maximum 2% of treasury per 24-hour window, enforced by the Squads program on-chain.
- Timelock — every Wardex-initiated transaction sits pending for 24 hours, visible to all multisig signers. The DAO retains a veto window on every action.
This is the same model Karpatkey uses with Aave and Balancer on EVM — Gnosis Safe role modules with destination whitelists rather than open spending authority. We are building the equivalent primitive on Solana.
After a proven Phase 2 track record, the whitelist expands (Marginfi, Drift, additional venues) and volume caps increase. The DAO still controls the policy parameters. Wardex executes within them.
DAO Governance (Realms)
│
▼
One-time policy vote
┌──────────────────────────────────────────────┐
│ "If native concentration > 70%: │
│ rebalance 20% → stablecoins. │
│ Deploy idle USDC to Kamino Earn. │
│ Monthly report to treasury committee." │
└──────────────────────────────────────────────┘
│
▼
Wardex Agent (monitors 24/7)
│ reads
├─── Realms SPL Governance treasury PDAs (mainnet, via RPC Fast)
├─── GoldRush real SPL token balances + USD pricing
├─── DefiLlama TVL and token breakdowns
└─── Live yield rates (Kamino, Marginfi, Drift, Jupiter Lend)
│ executes within mandate
├─── Phase 1: Realms proposal generation
├─── Phase 2: Squads destination-whitelisted delegate
└─── Real Solana transactions via Memo / SPL programs
│
▼
No vote per action. Policy decision = community.
Execution = agent.
This solves the governance paradox: selling native tokens requires a confidence vote that can suppress price. Wardex separates the strategic decision (one vote, made calmly) from execution (automatic, pre-approved, non-custodial).
- Realms SPL Governance integration — reads real governance accounts and native treasury PDAs from Solana mainnet for all 4 target DAOs
- RPC Fast mainnet reads — all Solana mainnet calls (getMultipleAccountsInfo, getParsedTokenAccountsByOwner, getBalance) routed through RPC Fast for speed and reliability
- GoldRush token balances — real SPL token balances with live USD pricing via Covalent's GoldRush API; badge shows live token count (e.g. "◈ GoldRush — 6 tokens"); replaces estimated ratios when coverage is ≥5% of TVL
- Jupiter Lend yield rates — live JupUSD vault supply rates from Jupiter Lend API alongside Kamino, Marginfi, and Drift; all four venues shown as a pill comparison; best venue selected automatically
- Historical TVL trend analysis — 30d/90d % change and daily slope computed from DefiLlama's historical TVL array; displayed as a directional indicator in the stats grid
- Multi-scenario bear market modelling — three named scenarios (Base −40%, Moderate −60%, Severe −80%) replace the single 70% estimate, shown side-by-side in the risk card
- Claude API governance proposals —
POST /api/agent/proposalcalls Claude Haiku to generate a fully-formed Realms governance proposal from live treasury data; structured template fallback when API key is absent - SNS agent identity —
wardex-agent.solregistered on Solana mainnet via@bonfida/spl-name-service; SOL record verified on-chain to the agent pubkey (47fyhXGzqw2SvATDVLqDNZNNcsMNVnMWCG3B8H4UCitF);/api/agent/identityreturns live registration status (snsRegistered,snsAgentMatch) and embeds the domain in every Memo program transaction - Real devnet execution — agent broadcasts a verifiable Solana devnet transaction via Memo program; returns a real Solana Explorer link
- Risk score 0–100 with native token concentration analysis
- Runway estimation scaled by treasury size and burn rate
- Estimated vs. on-chain data clearly labelled in the UI
- Protocol-specific DAO context (Jito Block Engine fees, Marinade MIP diversification, etc.)
- Security: rate limiting, input sanitisation, CSP headers, HSTS in production, X-Powered-By suppressed
| Layer | Technology |
|---|---|
| Backend | Node.js + Express |
| Governance | Solana SPL Governance (Realms) — mainnet reads |
| Blockchain | @solana/web3.js — mainnet reads + devnet execution |
| RPC | RPC Fast — high-performance Solana mainnet RPC |
| Token Data | GoldRush (Covalent) — real SPL token balances with live USD pricing |
| TVL Data | DefiLlama API — live TVL + 90-day historical array |
| Yield Data | DefiLlama Yields API — live Kamino, Marginfi, Drift APYs |
| Yield Data | Jupiter Lend API — live JupUSD vault supply rates |
| AI | Anthropic Claude API (claude-haiku-4-5) — governance proposal generation |
| Agent Identity | Solana Name Service (SNS) — wardex-agent.sol onchain agent identity |
| Frontend | HTML + CSS + Vanilla JS |
| Deployment | Render |
# Clone the repo
git clone https://github.com/jalaalrd/wardex
cd wardex
# Install dependencies
npm install
# Configure env vars
cat > .env << 'EOF'
# Required for agent devnet execution (optional — auto-generated if missing)
AGENT_KEYPAIR_HEX=<your_hex_encoded_64_byte_secret>
# RPC Fast — high-performance Solana mainnet RPC (rpcfast.com → Mainnet RPC application)
RPCFAST_API_KEY=
# GoldRush / Covalent — real SPL token balances with USD pricing (goldrush.dev)
GOLDRUSH_API_KEY=
# Jupiter Developer Platform — Jupiter Lend yield rates (developers.jup.ag)
JUPITER_API_KEY=
# Anthropic — Claude API for governance proposal generation (console.anthropic.com)
# Optional: falls back to structured template when absent
ANTHROPIC_API_KEY=
EOF
# Start the server
node index.js
# Startup log confirms active integrations:
# [integrations] RPC Fast:✓ | GoldRush:✓ | Jupiter:✓ | SNS: ✓ (wardex-agent.sol) | Claude:✓ (haiku-4-5)
# Open in browser
http://localhost:3000
# Test Realms on-chain integration (reads Jito treasury from mainnet)
http://localhost:3000/api/realms/treasury/jito
# Test live yield rates (includes Jupiter Lend)
http://localhost:3000/api/yield/rates
# Test agent identity (SNS + integration status)
http://localhost:3000/api/agent/identity
# Test agent execution (real devnet TX with wardex-agent.sol identity in memo)
curl -X POST http://localhost:3000/api/agent/execute \
-H 'Content-Type: application/json' \
-d '{"daoName":"Jito DAO","action":"Deploy idle SOL to Jupiter Lend","venue":"Jupiter Lend (JupUSD)"}'
# Generate a governance proposal (Claude Haiku or structured template fallback)
curl -X POST http://localhost:3000/api/agent/proposal \
-H 'Content-Type: application/json' \
-d '{"daoName":"Jito DAO","analysis":{"riskScore":70,"nativeConcentration":63,"stableRatio":18,"totalValue":929000000,"yieldVenue":"Kamino Earn","yieldRate":0.055,"annualYieldOpportunity":9200000,"action":"Deploy idle stablecoins to Kamino Earn."}}'Parcel and Tally both failed because free governance tooling cannot sustain a business. Wardex is designed from day one with a performance-aligned revenue model — benchmarked directly against what the market's leading treasury managers charge.
| Provider | Model | Fee | Notes |
|---|---|---|---|
| Karpatkey (ENS mandate) | AUM + performance | 0.5% AUM/yr + 10% of yield | Industry reference. ENS endowment. Fired by Gnosis 2025 over $700K oracle loss. |
| Steakhouse Financial | Performance only | 10% of yield | Morpho vaults, dYdX SubDAO. Token vesting + forfeiture clause standard. |
| Gauntlet (Aave) | Flat retainer | $2M/yr fixed | Risk management, not yield generation. No performance link — caused governance controversy. |
| Llama | Retainer + advisory | $400K–$800K/yr | Contributor DAO model. Acquired by Uniswap Foundation 2023. |
| Market standard (mid-tier DAOs) | Hybrid | 0.5–1% AUM + 10% performance | Most competitive mandates use a hybrid to align incentives while ensuring baseline viability. |
Karpatkey's 10% performance fee + 0.5% AUM is the industry standard for DAOs above $100M. Wardex targets the segment below — and prices accordingly.
Phase 1 — Intelligence Retainer (zero custody, immediate entry)
$300–$500/month for continuous on-chain monitoring, 3-scenario risk scoring, TVL trend alerts, and monthly treasury health reports. No governance vote required. Serves as the paid entry point before a Phase 2 mandate.
Comparable to Steakhouse's and Block Analitica's advisory retainers for mid-tier protocols — structured as a governance-approved monthly budget, paid in USDC or native token.
Phase 2 — Performance Fee on Yield (execution mandate)
10% of yield generated — aligned with Karpatkey/Steakhouse market standard, with two features competitors don't offer:
- Forfeiture clause: If Wardex underperforms a simple Kamino USDC deposit benchmark in any quarter, the performance fee for that quarter is waived
- On-chain accountability: Every yield deployment is recorded in a Squads transaction, visible to all multisig signers — no opaque oracle or off-chain execution (the exact failure that got Karpatkey fired from Gnosis)
Phase 3 — Governance Proposal Service (add-on)
$1,500–$3,000 per governance proposal drafted and submitted via Realms SPL Governance — risk analysis, yield comparison, rationale, and on-chain submission included. One-time per proposal, no retainer required.
The original range of 10–15% was above the market standard. Charging more than Karpatkey — the $1.8B AUM industry leader — with no track record signals either ignorance of the market or a founder who hasn't spoken to a DAO yet. 10% matches Karpatkey's ENS performance fee exactly, while the forfeiture clause and on-chain transparency give Wardex a structural reason to win on governance proposals against established competitors.
| Metric | Value |
|---|---|
| Target DAO treasury size | $5M average |
| Idle stablecoins (18%) | $900K |
| Live USDC APY (Kamino) | 5.5% |
| Annual yield generated | $49,500 |
| Wardex performance fee (10%) | $4,950/yr |
| Phase 1 retainer ($400/mo) | $4,800/yr |
| Combined revenue per DAO | ~$9,750/yr |
| Break-even DAOs (at $9,750 avg) | ~11 DAOs |
| 50 DAO ARR | ~$487,500 |
| 100 DAO ARR | ~$975,000 |
- Parcel shut down 2025 — direct gap in mid-tier DAO tooling
- Tally shut down March 2026 — "no venture business in governance tooling yet"
- Coinbase Agentic Wallets launched February 2026 — Solana-compatible autonomous execution infrastructure now exists
- MiCA regulation pushing DAOs toward professional treasury management
- Jito's JIP-24 (passed 2025) routes ~$15M+/yr to DAO treasury with no deployment strategy — immediate, named customer pain
Every major DAO treasury mandate in DeFi history — Karpatkey with Gnosis, Llama with Aave, Gauntlet with Compound — started with a governance forum post, not a cold sales call. Wardex follows the same playbook, compressed for Solana's faster governance cycle and our existing live data advantage.
Month 1 — Thought Leadership
Publish treasury intelligence analyses for Jito, Marinade, and Raydium directly on their governance forums — using Wardex's live on-chain data to show, not tell. The Jito Foundation forum (forum.jito.network) is the first target: JIP-24 just redirected 100% of Block Engine fees to the DAO treasury with no deployment playbook. That is a $15M+/yr problem Wardex solves with a working product, today.
Parallel content tracks:
- Post treasury risk analyses on Realms governance hub (1,700+ Solana DAOs)
- Participate in JIP-24 implementation discussions — position as a builder, not a vendor
- Publish "Solana DAO Treasury Health Check" thread on Twitter/X tagging @jito_sol, @MarinadeFinance, @JupiterExchange
Month 2 — Pilot Outreach
Offer one mid-tier Solana DAO a free 30-day Wardex monitoring engagement. Target: Raydium (on-chain governance live 2025, treasury management practices still maturing). Deliver a published case study with metrics: risk score before/after, yield opportunity quantified, governance proposal generated. Share it publicly on the DAO's own forum — free distribution in front of every DAO that watches Raydium.
Month 3 — First Contracts
Convert the pilot to a paid performance-fee mandate. Use the case study to approach Jito and Marinade through their governance forums with a formal proposal — the same governance proposal pathway Karpatkey used to win ENS, Balancer, and Aave (all via public DAO forum proposals, not private BD).
| Channel | Rationale | Priority |
|---|---|---|
| Jito Foundation Forum | JIP-24 = $15M+/yr idle, active governance community | Week 1 |
| Jupiter DAO Forum | Treasury reactivation planned for 2027 — position Wardex as the execution layer now | Week 2 |
| Marinade Realms governance | MIP-11 buyback strategy needs deployment intelligence; Head of Governance publicly accessible | Week 2 |
| Superteam UK Discord | Warm network — same community that sponsored this build; direct line to Kash Dhanda (Jupiter "Cat Herder", formerly Superteam DAO) | Week 1 |
| Realms.today community | 1,700+ Solana DAOs using the platform Wardex reads; treasury tooling posts get attention | Week 3 |
| Twitter/X (build in public) | Governance discussions happen publicly; tagging relevant DAOs builds inbound interest | Ongoing |
Karpatkey took 2 years to formalise its first client (Gnosis, where they were already embedded). Another 12 months to win ENS via competitive RFP. Wardex compresses this because:
- Live data is already running — the first Jito or Marinade demo is not a pitch deck; it's their actual treasury, scored, trended, and yielded in real time
- Mid-tier DAOs move faster — the $1M–$50M segment KPK ignores has shorter governance cycles and lower competition from established players
- Phase 1 is zero-friction — monitoring and proposal generation requires no custody, no multisig changes, and no governance vote. A DAO can start using Wardex before it votes to use Wardex
- Superteam UK network — direct access to builders running governance for several Solana protocols, including the Superteam DAO alumnus now leading Jupiter governance
Research into KPK's 2025 Gnosis firing (88% vote to remove) and Gauntlet's Aave exit reveals a consistent failure pattern in DAO treasury management: high fees + opaque performance + no accountability structure. Wardex's performance fee model (you only pay when yield is generated) and on-chain transparency (every action logged to a governance proposal or Squads transaction) structurally prevent both failure modes.
Colosseum's $250K pre-seed is 18–24 months of focused execution on one goal: get Phase 2 live on mainnet, sign 5 paying DAOs, and hit $30K ARR before a seed round.
| Allocation | Amount | What It Funds |
|---|---|---|
| Engineering | $87,500 (35%) | Founding smart contract engineer hire — Squads Phase 2 destination-whitelist delegate + Kamino Earn CPI integration ($80K/yr prorated). Founder salary: $50K/yr. |
| Security & Infrastructure | $42,500 (17%) | Smart contract audit for Phase 2 Squads integration ($20K — non-negotiable for treasury tooling) + RPC node infrastructure ($12K) + monitoring and alerting ($10.5K) |
| Product Development | $37,500 (15%) | Realms on-chain proposal submission, full SPL token account reads per governance PDA, ARIMA/ETS TVL forecasting, automated monthly treasury reports |
| Legal & Compliance | $25,000 (10%) | LLC formation, SAFEs, terms of service, privacy policy, accounting setup, and insurance — critical for a product handling DAO treasury mandates |
| Go-to-Market | $32,500 (13%) | Founder-led only: 5 DAO pilot incentives ($2K each = $10K), Breakpoint 2026 presence ($8K), governance forum content and community tools ($7K), DAO advisory network ($7.5K) |
| Buffer & Ops | $25,000 (10%) | 3-month contingency + accounting, bookkeeping, miscellaneous operational costs |
Monthly burn: ~$10,400 → 24-month runway on $250K
Milestones this capital achieves — in order:
- Month 2 — Smart contract audit complete; Phase 2 Squads delegate architecture reviewed and cleared
- Month 3 — Phase 2 live on Solana mainnet; first destination-whitelisted execution within a live DAO's pre-approved mandate
- Month 4 — 3 pilot DAOs signed (target: Jito CSD, Raydium + 1 emerging DAO); first real yield deployment event on-chain
- Month 6 — 5 paying DAOs; first performance fee invoices issued; $30K ARR confirmed
- Month 12 — 16 DAOs at $6,188/yr average = break-even; Series A raise begins from a position of revenue
Why this allocation, not 60% on team:
The biggest risk for a DAO treasury product is not distribution — it's trust. A DAO will not hand over a treasury mandate to a product that hasn't been audited. Allocating $20K to a smart contract audit before the first enterprise pitch is not caution; it's the product. Every dollar that goes to security before Month 3 is worth $10 in GTM spend after, because it removes the first and largest objection in every DAO conversation.
GTM is founder-led at this stage. Karpatkey's entire client base from 2022–2024 was won through governance forum proposals — no marketing team, no paid ads, no events budget. Wardex's first 5 DAOs will be acquired the same way: by showing live treasury data for their own protocol and asking if they'd like the agent to monitor it continuously. That costs time, not money.
- Live treasury TVL from DefiLlama
- Realms SPL Governance integration — real mainnet treasury reads for Marinade, Jito, Jupiter, Raydium
- Live yield rates from DefiLlama Yields API (Kamino, Marginfi, Drift)
- Jupiter Lend API — live JupUSD vault supply rates as additional yield venue
- RPC Fast mainnet RPC — high-performance Solana reads, getProgramAccounts via public fallback
- GoldRush (Covalent) — real SPL token balances with live USD pricing
- SNS agent identity — wardex-agent.sol embedded in all agent transactions
- Bear market risk scoring and scenario modelling
- Real agent execution on Solana Devnet — verifiable Memo program transactions
- Competitive gap UI — policy-based onboarding flow
- Non-custodial trust model — phased architecture, destination whitelist design
- Historical TVL trend analysis — 30d/90d % change from DefiLlama historical array
- Multi-scenario bear market modelling — Base (−40%), Moderate (−60%), Severe (−80%)
- Claude API governance proposal generation — Realms-ready proposals via claude-haiku-4-5
- Realms governance proposal generation — submit proposals directly on-chain via SPL Governance
- Squads destination-whitelist delegate integration — Phase 2 execution model
- Full SPL token account reads per governance PDA — GoldRush currently reads realm pubkey; extend to all derived treasury PDAs
- Time-series forecasting (ARIMA/ETS) on historical TVL for runway prediction
- Kamino Earn Vault integration — direct idle stablecoin deployment via CPI
- Automated monthly treasury health reports
Jalaaldeen Akinola — MSc Business Analytics, Newcastle Business School, Northumbria University
Twitter/X: @jalaal_tweets · GitHub: jalaalrd
I'm building Wardex because I spent months researching on-chain governance and DAO financial behaviour for my MSc dissertation — and what I found was embarrassing: protocols generating tens of millions in protocol fees are letting that capital sit idle in a wallet, because the governance process required to deploy it is so friction-heavy that it never happens. Jito's JIP-24 is the clearest example: $15M+/yr in Block Engine fees now routing to the DAO treasury with no deployment strategy at all.
I'm the right person to build this because I sit at the intersection of two worlds most builders occupy separately: quantitative financial analysis (my MSc programme covers forecasting, risk modelling, and portfolio theory) and Solana on-chain development (I've shipped the Realms integration, GoldRush token reads, SNS identity, and Claude-powered proposal generation in this build). Wardex is not a general "AI for DAOs" product. It is a specific answer to a specific failure mode I observed in the data.
Previously founded Zakatchain — a blockchain-based charitable giving platform that raised funds for 300+ families in Nigeria. That experience taught me how to build trust with communities who have never used a financial tool like this before — the same trust problem Wardex solves for DAO treasury committees.
Wardex integrates four sponsor stacks as core infrastructure, each solving a real gap in the product:
| Track | Sponsor | Integration | Endpoint |
|---|---|---|---|
| RPC Fast | RPC Fast | High-performance Solana mainnet RPC — all treasury reads route through solana-rpc.rpcfast.com when RPCFAST_API_KEY is set |
All /api/treasury/* reads |
| Not Your Regular Bounty | Jupiter | Jupiter Lend API (api.jup.ag/lend/v1/earn/tokens) — live JupUSD vault APY as an additional yield deployment venue. See JUPITER_DX_REPORT.md for full integration notes. |
/api/yield/rates → jupiter_lend_usdc |
| Build with GoldRush | Covalent | GoldRush Balances API (api.covalenthq.com/v1/solana-mainnet/...) — real on-chain SPL token balances with live USD pricing for treasury PDAs |
/api/treasury/:protocol → goldRushConnected |
| SNS Identity | Bonfida | wardex-agent.sol registered on Solana mainnet; SOL record verified on-chain to agent pubkey via getDomainKeySync + NameRegistryState.retrieve + resolve(); /api/agent/identity returns live snsRegistered and snsAgentMatch fields |
/api/agent/identity |
| Claude API | Anthropic | claude-haiku-4-5 generates Realms-ready governance proposals from live treasury analysis data; structured template fallback when key absent |
POST /api/agent/proposal |
Treasury risk data: Schellinger, Fiedler & Steinmetz (2023), Blockchain Research Lab. TVL data: DefiLlama. Yield data: DefiLlama Yields API. Built for the Colosseum Frontier Hackathon 2026, powered by Superteam UK.