Use cases
What this looks like for an expert brand, an agency, a lead-gen team, a SaaS and the rest.
Guides and references
Everything written about building your own service and product in one, in one index: what the words mean, a recipe for each job, what we have learned running it, and an example you can run for every operation. Start anywhere — none of it assumes you write code.
The LinkedIn API
Connecting an account, what LinkedIn asks for when you do, and the calls that do the work afterwards: what the words mean, a recipe for each job, and what we have learned running it.
- LinkedIn API
- White-label endpoint
- Checkpoint
- In-app validation
- Account id
- API key
- Session cookie (li_at)
- Webhook
- Idempotency
- Pacing
- Proxy and egress
- MCP
- Reply to every new LinkedIn message, automatically
- Turn a list of LinkedIn profile URLs into real data
- Connect your first LinkedIn account
- LinkedIn asked for a code, or a tap in the app
- Your first API call
- Use LinkedIn from Claude or ChatGPT
- A 2xx does not mean it appeared on LinkedIn
- What actually makes LinkedIn ask for extra verification
- Designing around rate limits you cannot see
- Reading webhooks without doing everything twice
- What an AI assistant can and cannot do with a LinkedIn API
Connect your AI
ChatGPT, Claude, Cursor and Grok, each given the whole API as one address: what a connector is, what to answer when it asks how to authenticate, and a recipe per client.
Project rooms and AI agents
Your team, the built-in specialists and any AI agent you bring, in one room: what it is, how to start one, and how to use the same agents outside it.
- Project room
- Built-in AI agent
- Outside AI agent
- AI agent handoff
- Start a project room for your team
- Add your own AI agent to a room
- Bring a third-party AI agent into a project room
- Use AI agents with LinkedIn outside a project room
- Work a LinkedIn inbox with a team of AI agents
- Run project rooms under your own brand
- Let the Google Ads AI agents read your accounts
- Built-in, your own, or outside: three kinds of AI agent in one room
- Keeping AI agents from talking over each other, and over your budget
Messaging channels
WhatsApp, Instagram, Facebook, TikTok, Telegram, LINE, email, SMS, your website chat, X and Discord on one key: what each channel is, how to answer it from code or with AI agents, and what each platform asks of you.
Work trackers and ChatWoot
Asana, ClickUp, HubSpot, Jira and the ChatWoot your team already answers from, on the same key: what one connection reaches, how to keep a room's tasks in step, and how to answer a customer from your own code.
Google Ads
The Google Ads accounts somebody handed you, read and changed through the same key: what the words mean, the calls that report and propose, and why nothing is applied until a person says so.
Campaigns and Top Voices
LinkedIn work that keeps running after you stop calling: posts written and published on a schedule, the people you want to reach engaged with, and the voices your field already reads answered while their posts are still new.
API glossary
The words you meet when you turn the LinkedIn and messaging work your business does by hand into something that runs itself, defined plainly.
Templates
Ready-to-copy recipes that each solve one job end to end, and then keep solving it without you.
- Add your own AI agent to a room
- Answer ChatWoot from your own code
- Answer Discord questions from your own code
- Answer every channel from one inbox
- Answer the top voices in your field
- Bring a third-party AI agent into a project room
- Connect ChatGPT to your LinkedIn
- Connect Claude to your LinkedIn
- Connect Cursor to your LinkedIn
- Connect Gemini to your LinkedIn
Notes
What we have learned turning the LinkedIn and messaging work businesses do by hand into something that runs itself.
- Reading webhooks without doing everything twice
- Built-in, your own, or outside: three kinds of AI agent in one room
- Reading an account is not changing it
- One key for the tools your team already uses
- One key for every channel
- The question everybody gets wrong
- A 2xx does not mean it appeared on LinkedIn
- A score you can argue with
- Designing around rate limits you cannot see
- Keeping AI agents from talking over each other, and over your budget
API examples
Every operation, with a request you can run and the answer you get back.
- Accept a Relation request — Top-Rated Team API
- Add a Chat Participant — Top-Rated Team API
- Add a chat to your website — Top-Rated Team API
- Add a deal to a HubSpot pipeline — Top-Rated Team API
- Add a Message Reaction — Top-Rated Team API
- Add a note to a contact — Top-Rated Team API
- Add a post to a content campaign — Top-Rated Team API
- Add a task to a ClickUp list — Top-Rated Team API
- Add a task to an Asana project — Top-Rated Team API
- Add a webhook — Top-Rated Team API