Build on the fundamental intelligence layer.
Use EquityLayer as the curated research harness behind Claude, Codex, Hermes, or another investment research agent.
View → Change → Act.
Stronger models fit the method to you. We keep the method and the record.
- 01Live
View
One maintained view per theme, built on five judgments: the question, the sources, assumptions and falsifiers, how to read the numbers, what counts as a change.
- 02Live
Change
Official records land. The agent compares them with the saved view and returns what changed and what to check next.
- 03Design
Act
The view as of that moment — version and evidence hash — goes with the user, or an agent they authorize, to their broker. They approve; the broker executes. EquityLayer never executes on its own, never pre-fills side, size, or price, and never holds credentials, positions, or funds.
- 04Planned
Record
What was acted on and what happened, beside the public claim ledger — written before the outcome, never edited after.
Pay per result. Not a subscription, not the methodology.
x402 on Solana devnet, local bridge: 0.05 USDC buys one source-linked Taiwan monthly-revenue refresh, session cap 0.15 USDC, signed result with output hash. No mainnet settlement is live and no settlement asset has been chosen. Revenue is never tied to executed trades.
- 01MCP · live
Load context
The agent starts with a company, theme, or existing investment view instead of a blank prompt.
- 02Routing policy
Route sources
The route tells the agent which evidence it needs. The agent can use approved sources and the financial connectors in its workspace.
- 03Evidence contract
Check change
The agent compares new evidence with the previous view. It keeps facts, inference, missing evidence, and source dates separate.
- 04Local-first
Return state
The agent writes what changed, why it matters, which companies are affected, and what to check next. The Dashboard is the review surface.
Use the tools the agent already has.
EquityLayer does not replace Claude, Codex, Hermes, or their financial connectors. It gives the agent a research starting point, source rules, and an evidence contract.
Research question
↓
EquityLayer context + source rules
↓
Agent selects an approved source or available connector
↓
Evidence check + investment impact + next triggerThe agent keeps control of tool calls. It must ask before it adds a new source or accesses private data.
One update becomes the next starting point.
Each run returns a small research state that the next run can compare. The local Dashboard lets a person inspect that state and its source links.
- Record the company, theme, or investment view.
- Record each source, date, and reporting period.
- Separate facts, inference, and missing evidence.
- Explain what changed and why it matters.
- Name affected companies and counterevidence.
- Save the next trigger for a later check.
ResearchUpdate {
subject, previousView, changes,
facts, inferences, missingEvidence,
affectedCompanies, nextTrigger, sources
}The Dashboard is the review surface.
The reusable layer is the context, routing policy, evidence contract, and research state that move with the agent.