Human × AI
Challenge a market claim
Write the claim, evidence and counterevidence. Build a skeptical AI reviewer brief, then revisit the result on the review date.
Start the challengeAI × Web3 research workbench
Bring a question. Leave with a working result.
Bring your records, state the assumptions and keep the evidence. Start with a research challenge or a payment-cost scenario, then share a result someone else can recalculate.
Start here · shareable tasks
Give people a reason to share the next step
These two entry points produce a useful artifact instead of a generic market page. Inputs remain local unless you choose to copy or export them.
Finance scenario
Recalculate stablecoin payment cost
Change amount, fees, FX spread, retries and settlement cost. Share the assumptions so a teammate can test the same case.
Open the finance cardFree beta · no wallet · no token · no automatic posting. Results are assumptions and evidence, not investment advice.
What changed, and what should you check?
Follow AI and compute markets, compare your saved watchlist and keep an evidence note to review later.
Open the research board Read official updates11 tools, all free to try
Choose the task you have today
| Workspace | What you can do | Output |
|---|---|---|
Agent Delivery Lab | Inspect payment challenge shapes and check responses against your declared rules. | Batch and response reports |
Stable Reconcile | Review invoices against your recorded stablecoin transfers. Keep partial payments, wrong assets and pending transactions visible. | Method, examples & report |
Agent Evidence | Review task-specific agent evaluations with sample deduplication, version boundaries and uncertainty intervals. | Method, examples & report |
Route Lab | Compare bounded supplier sequences using quality assumptions, worst-case cost and latency constraints. | Method, examples & report |
Protocol Ledger | Retrieve dated payment, identity and token-interface profiles with official links and explicit interpretation limits. | Method, examples & report |
Data Permit | Check intended AI use against recorded grants, expiry and regions, then model a revenue split without moving funds. | Method, examples & report |
Compute Lens | Bring measured workloads and account for setup, egress, storage and failed outputs before choosing compute. | Method, examples & report |
Incentive Lab | Model capped rewards for accepted work and see whether customer revenue covers operating costs before token support. | Method, examples & report |
Proof Plan | Plan a zkML handoff with measured proving time, verification cost and a pinned model/input/key manifest. | Method, examples & report |
Call Lens | Decode three common token-call shapes offline and compare the recipient, spender and amount with your own expectations. | Method, examples & report |
RWA Notes | Compare tokenized-asset disclosures while keeping units, dates, missing fields and original sources attached. | Method, examples & report |
Evidence comes first
Examples are fictional. Your input drives the calculation. A rule match does not establish a payment, legal right, safe contract or true AI answer.
Built for review
Each tool offers a form, JSON input, a visible report and a downloadable offline engine. No account or model subscription is needed. Optional anonymous feedback is separate from your input.
No matching task. Try a broader term or select all workflows.
A useful place to start
Comparing an AI supplier? Review task evidence, plan a bounded fallback, then check delivery with Delivery Lab. Each worksheet runs independently; you control the handoff.
Share a tool with your team or readers · Tool catalog for agents
Use this workflow in your AI assistant
Connect the MCP server to read sources and run these calculations from a supported client. Remote calls send parameters to the server; the browser worksheet remains local.
Get the MCP connection and citation guideUpdated