> ## Documentation Index
> Fetch the complete documentation index at: https://aisa.one/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Build a Market Intelligence Agent with AIsa

> Combine financial data, filings, news, prediction markets, social signals, and model synthesis in a verifiable market-research workflow.

A market intelligence agent should separate **market data**, **source evidence**, and **analysis**. Prices and metrics describe what happened; filings and news provide primary context; social or prediction-market signals may help explain expectations but should not be treated as verified outcomes.

## Choose the capability

| Need                                                          | Start with                                                     |
| ------------------------------------------------------------- | -------------------------------------------------------------- |
| Equity fundamentals, filings, analyst data, and macro context | [MarketPulse](/docs/agent-skills/marketpulse)                       |
| Broad market and company data                                 | [AIsa Market](/docs/agent-skills/market)                            |
| Cross-source trend forecast                                   | [Trend Forecast](/docs/agent-skills/trend-forecast)                 |
| Prediction-market data                                        | [Prediction Market Data](/docs/agent-skills/prediction-market-data) |
| Application-controlled financial calls                        | [Financial API Reference](/docs/api-reference/financial/get_prices) |
| Final comparison or report                                    | [Model inference](/docs/guides/models)                              |

## Recommended workflow

### 1. Define scope

Specify the asset, company, market, event, and time horizon. Distinguish between:

* Historical analysis.
* Current-state monitoring.
* Scenario analysis.
* Forecasting.

A forecast should never be presented as a verified fact.

### 2. Collect structured data

Retrieve only the data required for the question, such as:

* Prices and volume.
* Financial metrics or statements.
* Filings and filing sections.
* Earnings or analyst information.
* Relevant macro indicators.

Record the timestamp and units for every time-sensitive value.

### 3. Add primary-source context

Use filings, company releases, and official data when available. Use news to discover developments, then verify material claims against the most direct source possible.

### 4. Add optional market signals

Prediction markets and social discussion can indicate expectations, attention, or disagreement. Label them as signals rather than ground truth.

Do not combine values from different sources without checking their definitions, currency, frequency, and observation time.

### 5. Synthesize with a model

Ask the model to separate:

* Observed data.
* Source-backed events.
* Interpretation.
* Scenarios and assumptions.
* Risks and missing evidence.

### 6. Verify before delivery

* Recheck prices and time-sensitive metrics.
* Link filings or primary evidence.
* State the cutoff time.
* Mark forecasts and opinions clearly.
* Avoid presenting the output as personalized financial advice.

## Read versus write behavior

Market research is normally read-only. Do not place trades, alter portfolios, publish content, or trigger alerts unless the user explicitly requests that separate action and the relevant authorization path exists.

## Minimal documentation set

* [Capabilities by Goal](/docs/by-goal)
* One selected market Skill or API family
* [Authentication](/docs/guides/authentication)
* [Models](/docs/guides/models) for synthesis
* [Security](/docs/guides/security)

## When AIsa may not be necessary

A dedicated market-data provider may be simpler when the application only needs one dataset and already owns the analysis, model, and billing layers.