Early access
AI agents for Instagram
Language models are good at deciding what to look up and terrible at looking it up. Give an agent the Instagram half — profiles, reels, comments, followers, influencer search — and it can research a niche, vet a creator or summarise a comment section on its own.
Build it yourself today
You don't need anything from us to start. Every endpoint on the Instagram Scraper API returns plain JSON, which is exactly what a model's tool-calling interface wants. Register the endpoints you need as tools in the OpenAI or Anthropic SDK, LangChain, LlamaIndex or your own loop, and the model handles the rest.
tools = [{
"name": "instagram_profile_reels",
"description": "Every reel on a public Instagram profile, with view counts.",
"input_schema": {
"type": "object",
"properties": {
"channel_url": {"type": "string", "description": "@handle or profile URL"},
"count": {"type": "integer", "description": "How many reels, up to 1000"},
},
"required": ["channel_url"],
},
}]The guides have working Python for each endpoint — an agent tool is that code with a schema wrapped around it.
What we're building on top
A hosted agent, so you don't have to assemble the loop, the retries and the rate limits yourself. You describe an outcome; it plans the lookups, runs them against the same infrastructure behind the free tools, and hands back a finished artefact — a shortlist, a comment digest, a competitor brief.
- Creator sourcing. "Twenty fitness nano-influencers in Canada, 1k–10k followers, engagement above 3%, with contact emails where public."
- Comment digests. "Read every comment on these six reels and tell me the three complaints that keep coming up."
- Competitor briefs. "Watch these four brands' reels and tell me on Monday what changed and which post broke out."
Request early access
It isn't open to everyone yet — we're onboarding a few teams at a time so the workflows get built around real problems. Tell us what you'd point it at and we'll be in touch when there's a slot.
Request early access
Leave your email and what you'd use it for. No spam, and nothing is charged.
Frequently asked questions
Can ChatGPT read Instagram data?
Not on its own - it has no access to Instagram's data and can't log in. It can, however, call a tool that does. Point an agent at an Instagram scraping API and the model handles the reasoning while the API handles the fetching.
What is an AI agent for Instagram?
A language model plus tools it can call, running on a loop: you describe the outcome ("shortlist ten skincare nano-influencers in Canada and draft outreach"), the model decides which lookups to run, runs them, reads the results and keeps going until it is done.
How do I give an LLM live Instagram data today?
Expose the API endpoints as tools in your agent framework - function calling in the OpenAI SDK, tool use in the Anthropic SDK, or a tool in LangChain or LlamaIndex. Each endpoint returns plain JSON, which is already the shape models handle best. You can build this now without waiting for anything from us.
What stage is the hosted agent at?
Early access, and not generally available. We're taking a small number of teams through it one at a time so the workflows get built around real use cases. Request access below and you'll hear when a slot opens.
What does it cost?
Pricing isn't set yet - that's part of what the early access group is helping shape. The underlying API is priced publicly on RapidAPI today.
In the meantime, the free tools do the individual lookups in your browser, and the API does them in your code.