AI Agents That Disagree
So You Don't Have To

An AI-powered token security platform built on Solana that deploys specialized AI agents to evaluate any token's risk. The agents independently analyze on-chain data, market signals, and social sentiment — then engage in adversarial debate until reaching consensus. Think of it as a courtroom trial for every token, with AI prosecutors, defenders, and a jury.

Most scanners use a single model or static rules. VerdictSwarm runs a Proprietary Adversarial Consensus Engine that coordinates specialized AI agents powered by Gemini, Grok, Kimi, Claude, and ChatGPT (Codex). Agents independently analyze, cross-check, and enter adversarial debate when they disagree — then converge using our proprietary scoring methodology. This multi-model approach consistently outperforms single-model analysis, and every verdict is stored immutably on Solana.

Yes. No wallet, no signup, no catch. Paste any Solana SPL token address or EVM contract address and get a real AI-powered analysis with 2 agents. Free forever — we want you to see the quality before upgrading.

Connect your Solana wallet to access Pro tier with 6 agents + adversarial debate. Higher tiers with up to 20+ agents and frontier AI models are coming soon. Free tier requires no wallet — just paste an address and scan.

Five top-tier AI providers: Google Gemini (security analysis), xAI Grok (social signals & sentiment), Moonshot Kimi (research & deep analysis), Anthropic Claude (reasoning & architecture), and OpenAI Codex (code analysis). We continuously upgrade to the latest models — our architecture is model-agnostic, so when a better model drops, we integrate it immediately.

That's the plan. We're building a B2A (Bot-to-Agent) API so any AI agent or trading bot can get swarm intelligence before every trade. Drop a SKILL.md into your OpenClaw agent, or hit the SSE API directly. One call, instant verdict. API access is coming soon — follow @VswarmAi for launch updates.

Solana (native, first-class support), Ethereum, Base, Arbitrum, Polygon, BSC, Optimism, and Avalanche. Solana token scanning uses Helius RPC for deep on-chain analysis.

Free (2 agents): ~70%. Tier 1 Investigator (6 agents + Devil's Advocate + debate): ~85%. Tier 2 Prosecutor (adversarial multi-pass): ~90%. Tier 3 Grand Jury (full swarm): ~92%. Tier 4 Consensus (4 frontier AIs debate to consensus): ~95%. Accuracy improves continuously as we integrate newer, smarter AI models and expand on-chain data sources.

Yes — every scan verdict is stored as an immutable PDA (Program Derived Address) on Solana via our Anchor program. This creates a verifiable, permanent record of AI security analysis that anyone can audit on Solana Explorer.

Yes — this is core to our value. AI evolves weekly. When Google releases a new Gemini, when xAI upgrades Grok, when Anthropic ships a new Claude — we integrate it. Our swarm architecture is model-agnostic, so upgrading is seamless. We also continuously optimize our agent prompts and analysis strategies with each new model release — extracting maximum intelligence from every provider. Your scans automatically get smarter over time.

Yes. VerdictSwarm's adversarial consensus engine, prompt architecture, and scoring methodology are proprietary technology. Think of it like PageRank for Google — continuously tuned, continuously upgraded, and central to why our verdict quality compounds over time.