Hello AI Enthusiast,
What if posting on LinkedIn felt as easy as leaving a voicemail for a colleague?
Ben Libberton, a scientist and biotech communicator, joined our AI Agent Bootcamp to build exactly that. The result is a Telegram-based assistant that lets salespeople voice their ideas, stores them automatically, and generates polished LinkedIn posts on demand — no writing required.
The Problem
Ben's target user was "Eddie" — a Sales Manager at a biotech company with a strong network and plenty of ideas, but no time to write. Eddie gets told to post more, spends half a day agonizing over it, and eventually publishes something underwhelming. Rinse and repeat.
The thing is, Eddie has ideas all the time. He just has no easy way to turn them into content.
How Ben Built It
Ben worked with our AI coaches to nail down the simplest flow that would actually solve the problem: one pipeline to capture ideas by voice, another to generate posts on demand.
Here's the full automation flow, built using Make, Telegram, OpenAI, and Google Sheets:
User sends a voice note → triggers automation (TELEGRAM)
AI transcribes audio and extracts post topic ideas (OPENAI WHISPER)
Topics are stored for later use (GOOGLE SHEETS)
User requests a post → AI retrieves a topic and writes it in the user's tone (OPENAI)
Finished post is delivered back to the user (TELEGRAM)
The full automation scenario in Make
Step 1: The Voice Brain Dump
Eddie opens Telegram and sends a voice note — 30 seconds of rambling about a client call, a product update, whatever's on his mind. The automation transcribes it with OpenAI's Whisper API, then pulls out specific, post-worthy topics with a short synopsis for each.
The Telegram bot receiving a voice note
Step 2: Topic Storage
The extracted ideas get logged into a Google Sheet automatically — a content library that grows every time Eddie does a brain dump. With ideas stored, he never starts from a blank page again.
Step 3: Post Generation on Demand
When Eddie wants a post, he asks for one in Telegram. The system grabs a topic and runs it through a writing prompt that knows his tone, his industry, and what makes a LinkedIn post perform. The opening lines are built to stop the scroll; the body delivers the substance.
Step 4: Review and Post
Eddie reads the draft, tweaks anything he wants, pastes it into LinkedIn, and posts. Minutes instead of half a day.
What Ben Learned Along the Way
Building this project taught Ben lessons that go beyond the technical side:
Scoping is everything. Learning to identify the core flow that delivers value — and build that first — is a lesson every bootcamp participant walks away with.
Model choice matters. Ben tested multiple AI models and found real differences in output quality. Picking the right one is as important as writing a good prompt.
Prompt engineering takes iteration. The LinkedIn post prompt went through several rounds to get tone, structure, and formatting right.
Cost is a real variable. Ben calculated the cost per completion at around 15 cents — worth knowing before you scale anything.
Ben's LinkedIn writing assistant is the kind of practical project you'll build in our AI Agent Bootcamp. Next cohort starts March 9, with 1:1 coaching to help you build whatever AI prototype solves your actual problem.
Want to get even more practical? Explore hands-on AI learning with AI Academy:
AI Academy Membership: Get 12 months of access to all our cohort-based programs and on-demand courses.
AI Agent Bootcamp: Accelerate processes and solve business problems by building AI Agents, without coding.
Corporate Training: Equip your team with the skills they need to unlock the potential of AI in your business.
Practical Introduction to ChatGPT: A free course on using ChatGPT confidently, understanding its workings, and exploring its potential.
We'll be back with more AI tips soon!
