Create Exa Search Feed

User Intent

"How do I search the web with semantic/neural search using Exa? Show me Exa search feed configuration."

Note: Exa API key is not required. Graphlit provides Exa access by default and bills you for usage.

Operation

SDK Method: createFeed() with FeedTypes.Search Auth: None required (Graphlit provides Exa access, bills you for usage)


Code Example (TypeScript)

import { Graphlit } from 'graphlit-client';
import { FeedTypes, SearchServiceTypes } from 'graphlit-client/dist/generated/graphql-types';

const graphlit = new Graphlit();

const feed = await graphlit.createFeed({
  name: 'AI Research Papers',
  type: FeedTypes.Search,
  search: {
    type: SearchServiceTypes.Exa,
    text: 'machine learning transformers research',
    readLimit: 20,
  },
  // Optional: add workflow for content processing
  // workflow: { id: workflow.createWorkflow.id }
});

console.log(`Created Exa search feed: ${feed.createFeed.id}`);

Configuration

text: Natural language search query readLimit: Number of results per execution type: SearchServiceTypes.Exa


What Makes Exa Different

Neural Search: Uses embeddings to find semantically similar content, not just keyword matches

Content Quality: Optimized for high-quality web content (articles, research, documentation)

Recency: Can prioritize recent content


Use Cases

Research Papers:

text: 'transformer architecture improvements 2024'

Technical Articles:

text: 'distributed systems consensus algorithms explained'

Company Research:

text: 'startups using GraphQL in production'

Thought Leadership:

text: 'AI safety alignment research latest developments'

Exa vs Tavily

Use Exa when:

  • You want semantic/neural search

  • Looking for high-quality long-form content

  • Researching specific topics in depth

  • Need technical documentation or papers

Use Tavily when:

  • You want traditional keyword search

  • Need real-time news updates

  • Prefer broader web coverage

  • Want faster results



⚠️ ExaCode Limitation

ExaCode is not available as a feed. ExaCode returns only Markdown content (no URLs to ingest), so it can only be used via the searchWeb() SDK method for direct retrieval, not as a continuous feed.

For code and technical documentation search as a feed, use regular Exa search with code-focused queries.


Last updated

Was this helpful?