Desearch AI

What Desearch AI does

Desearch is an open-source, decentralized, and community-driven AI-powered search engine built on Subnet-22 (SN-22) of the Bittensor Network. It provides a live search API for AI agents, offering fresh context from web, social, and crawl sources. Desearch is designed to be censorship-resistant, empowering users with unfiltered, relevant information. It offers various API products including AI Search, X Search, Web Search, Web Crawl, and Forecasting, all with transparent, usage-based pricing. The platform is built for developers and teams shipping agent products, providing tools for API key management, playground testing, usage monitoring, and error tracking.

Products

• AI Search APIProvides AI summaries with source attribution from live web, social, Reddit, arXiv, and open sources, offering multi-source context in agent-ready JSON.
• X Search APIRetrieves X posts, links, conversations, and user signals for monitoring and agent workflows, including real-time posts, conversation context, and advanced filters.
• Web Search APIOffers fast web results with titles, snippets, links, and structured response fields for products, supporting structured SERP data and regional queries.
• Web Crawl APIExtracts clean page text, metadata, and links from URLs for RAG and data pipelines.
• Forecasting APIGenerates probability-backed forecasts for events, risk, markets, and research questions, providing probability outputs and research signals.
• TikTok Search APIAllows retrieval of TikTok posts by ID, profile information, profile posts, comments, hashtag posts, and trending posts.
• Facebook Search APIEnables retrieval of Facebook posts by ID, profile or page information, profile/page posts, comments, and hashtag posts.
• Instagram Search APIFacilitates retrieval of Instagram media by ID, profile information, profile posts, comments, and hashtag media.
• Numinous ForecastingAPI for forecast job creation and result retrieval.

Key capabilities

• Open-source
• Decentralized
• Community-driven
• AI-powered search engine
• Built on Subnet-22 (SN-22) of the Bittensor Network
• Censorship-resistant
• Live search API for production agents
• Fresh context for AI agents (web, social, crawl, cited AI context)
• Concise results with sources
• Clean console for API keys, playground testing, usage, errors, and transparent billing
• Live API playgrounds for agent builders
• Usage-based API pricing
• One credit model across all API products
• Scale-ready with self-serve keys and enterprise credits
• Python SDK
• JavaScript/TypeScript SDK
• Integrations with LangChain, LlamaIndex, CrewAI, OpenAI, Claude, MCP
• Multi-source AI search with links and summaries
• SERP-style web results and crawl output
• Social media search (X/Twitter, TikTok, Facebook, Instagram)
• Forecasting job creation and result retrieval
• Billing metadata in API responses (cost_usd, usage_count, service, currency)

Use cases

• AI answer engines (retrieve current sources, generate grounded summaries)
• Vertical search (focused domain, industry, brand, or topic search)
• Brand and market monitoring (track mentions and trends across web and social sources)
• Research workspaces (collect links, posts, and summaries for analysts or agents)
• Competitive intelligence (monitor product launches, news, discussions, technical research)
• AI agents that need fresh context (before writing, reasoning, or acting)
• Search products (build AI search, web search, social search experiences)
• Market and trend intelligence (monitor discussions, posts, comments, profiles, links)
• Data enrichment (retrieve source URLs, posts, profiles, media, comments, summaries for downstream pipelines)
• AI-driven chatbots (interactive, real-time responses, customer support, virtual assistants)
• Intelligent agent task automation (research agents, sales/market agents, social intelligence agents)
• RAG pipelines (crawl or search fresh sources before embedding or summarizing)
• eBay scraping and price intelligence (monitor marketplace prices, compare listings, track competitors, provide fresh ecommerce context for AI agents)
• Forecasting agents (access up-to-date information, pull evidence from multiple sources, re-query as new data appears)
• Decentralized X marketing (social validation, engagement verification)
• Content intelligence and verification (deliver reliable, real-time data)
• Credibility scoring (real-time social intelligence)

AI approach

Desearch AI provides an open-source, decentralized, and community-driven AI-powered search engine built on Subnet-22 (SN-22) of the Bittensor Network. It offers a live search API for AI agents to retrieve fresh web, social, crawl, and cited AI context. The platform focuses on providing unfiltered, relevant information and includes AI summaries with source attribution. It also supports AI-driven chat and intelligent agent task automation by allowing agents to call Desearch APIs as tools for web search, social search, source discovery, and deeper research. Desearch emphasizes its ability to provide real-time, fresh context for AI agents, contrasting with stale training data.

Tech named: LLM, Bittensor, Subnet-22 (SN-22), TAO, RAG, OpenAI Wrapper, Function Calling with GPT, Function Calling with Claude, LangChain, LlamaIndex, ElizaOs Agents, CrewAI Agents, OpenClaw Agent, Pydantic models, TypeScript

Industries served

• Technology, Information and Internet
• Crypto Analysis
• Reporting
• Research
• Content Marketing
• Creator Economy
• Influencer Marketing
• Ecommerce

What it says sets it apart

• Open-source, decentralized, and community-driven
• Built on Bittensor Network (Subnet-22)
• Censorship-resistant
• Provides fresh, real-time context for AI agents
• Transparent, usage-based pricing without hidden plan gates
• One credit model across all API products
• Strong focus on developer experience with SDKs and integrations for popular AI frameworks
• Visible proof of search quality through benchmarks and evaluation trends
• Direct team access for support and product feedback via Discord
• Supports a wide range of social media data (X, TikTok, Facebook, Instagram) beyond just web search
• Enables self-play forecasting agents by providing real-time data access in sandboxed environments