> ## Documentation Index
> Fetch the complete documentation index at: https://developers.t2000.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Research with citations

> A question in, a sourced answer out. Exa finds the primary sources, Perplexity grounds the answer in live web citations — paid as you go, no scraping setup.

Ask a real question and get back an answer you can trust, with sources. Exa surfaces the most relevant primary pages by meaning, Perplexity returns a web-grounded answer with live citations, and Claude (optionally) fuses them into a tight, cited brief. Self-contained — just a question. \~\$0.06.

***

## The prompt

```
Use t2 services. What are the leading approaches to stablecoin yield on Sui right now, and how
do they differ? Use Exa to find primary sources and Perplexity for a grounded
answer — cite everything.
```

***

## What runs

1. `POST /exa/v1/search` — semantic search for the best primary sources (\~\$0.02)
2. `POST /perplexity/v1/chat/completions` — web-grounded answer with citations (\~\$0.02)
3. `POST /anthropic/v1/messages` — Claude fuses into a cited brief (optional) (\~\$0.02)

***

## Run it

### Claude Desktop (MCP)

```bash theme={null}
npm install -g @t2000/cli && t2 init && t2 fund && t2 mcp install
```

Paste any question — the agent decides whether one source pass is enough or whether to fuse Exa + Perplexity for breadth.

### SDK

```typescript theme={null}
import { T2000 } from '@t2000/sdk';

const agent = await T2000.create();
const question = 'Leading approaches to stablecoin yield on Sui, and how they differ';

const [sources, grounded] = await Promise.all([
  agent.pay({
    url: 'https://mpp.t2000.ai/exa/v1/search',
    method: 'POST',
    body: JSON.stringify({ query: question, numResults: 5, contents: { text: true } }),
  }),
  agent.pay({
    url: 'https://mpp.t2000.ai/perplexity/v1/chat/completions',
    method: 'POST',
    body: JSON.stringify({
      model: 'sonar',
      messages: [{ role: 'user', content: `${question}. Answer with inline citations.` }],
    }),
  }),
]);

const brief = await agent.pay({
  url: 'https://mpp.t2000.ai/anthropic/v1/messages',
  method: 'POST',
  headers: { 'anthropic-version': '2023-06-01' },
  body: JSON.stringify({
    model: 'claude-sonnet-4-5',
    max_tokens: 700,
    messages: [{
      role: 'user',
      content:
        `Write a 250-word answer to: "${question}". Keep every claim tied to a source URL.\n\n` +
        `SEMANTIC SOURCES: ${JSON.stringify(sources.body)}\n\nGROUNDED ANSWER: ${JSON.stringify(grounded.body)}`,
    }],
  }),
});

console.log((brief.body as { content: { text: string }[] }).content[0].text);
```

***

## Expected output

```
2–3 calls · ~$0.06 · ~7s · 0 taps
Cited brief + source URLs
```

***

## Extend it

* Swap **Exa** for **Firecrawl** (`/firecrawl/v1/scrape`) to read a specific known source verbatim
* Add **Firecrawl map** (`/firecrawl/v1/map`) to discover every page on a site before extracting
* Have a second model (**Gemini** `/gemini/v1beta/models/gemini-2.5-pro`) review the brief for unsupported claims
* Pipe the cited brief into the **Write code, then run it** recipe to turn data points into a verified chart or table