SocialRouter
Extracting data from social networks—post likes, comments, followers, profiles—means wrestling with a fragmented market. There are dozens of scraping providers (Lobstr, Evaboot, PhantomBuster, Apify, Bright Data…), each with its own account, its own API, its own authentication, rate limits, and data format.
Our client was facing exactly this challenge: every new provider meant a new integration, and every blocked or discontinued provider forced a rebuild. They needed reliable social data without betting their product on any single source—the exact problem the LLM ecosystem had before OpenRouter: fragmented providers, incompatible APIs, no unified interface, and zero reliability when a source fails.
We built SocialRouter for them—the OpenRouter of social media data: a single API endpoint that routes each request to the best available provider, normalizes the output into one consistent format, and bills pay-as-you-go per request, with no subscription and no lock-in.
The client sends a social URL and what they want—likes, comments, followers, posts—and SocialRouter handles provider selection, failover, normalization, and caching behind the scenes. A LinkedIn like and an Instagram like come back in the exact same shape.
SocialRouter sits as an abstraction layer between the client and the provider ecosystem. Every request flows through a routing engine that checks the cache, filters providers by health and capability, picks the best one, and normalizes whatever comes back into a single schema—so clients never touch a provider-specific quirk again.
On each call, SocialRouter checks the cache for a recent identical request, determines which active providers support the source and data type, ranks them by priority, cost, and latency, calls the winner, normalizes the response, stores it, deducts credits, and returns clean data—falling over to the next provider automatically if one fails.
The platform ships as a full stack: a hosted REST API, a Next.js dashboard for keys and usage, an MCP server so AI agents can extract social data as a native tool, and published SDK, CLI, and MCP packages on npm. Credits, logs, and access are enforced with row-level security in Supabase, and billing runs on Stripe usage-based pricing.
Because routing and caching are built into the core, SocialRouter gives the client reliability no single provider can match while keeping per-request costs low. Cache hits serve repeated requests instantly—the provider is called once, but the client gets the answer every time.
We delivered SocialRouter as an API-first platform so the client's team integrates once and gains access to the whole provider ecosystem—whether they're building lead-gen tools, CRMs, or analytics dashboards, or plugging it into automation platforms like n8n, Make, and Zapier as a single data source.
Crucially, we made it agent-native: the bundled MCP server lets the LLM agents the client builds—on Claude, GPT, and others—extract social data autonomously, finding everyone who liked a post, pulling a profile's recent activity, or enriching a list, all as a first-class tool. No other social data provider ships this out of the box.
For the client, the advantage compounds over time. As the social data provider market keeps fragmenting—more providers, more networks—the unified API we built only becomes more valuable, while the MCP-native design meets the AI agents their users increasingly rely on.
Because routing, caching, and deduplication are built into the core, the platform stays cheap to run as volume grows and turns a thin per-request margin into durable infrastructure economics—reliability and reach no single provider could give the client on their own.
Let's ship infrastructure that scales with your business
Book Your Strategy Call