Looking for Make specialist — Claude API + Notion + Google Drive workflow
We’re a small content strategy and ghostwriting firm building our first automated workflow using Make. Looking for someone to help us design and build it, with the possibility of ongoing work as we add more workflows over time.
The first build
A transcript file lands in a designated Google Drive folder. Make detects it, extracts the text, sends it to the Claude API with a structured prompt, receives a JSON response containing four structured outputs (action items, content ideas, client commitments, decisions made), creates a new Notion page, populates it with the formatted outputs, and sends a Telegram notification to the team via a bot we will set up on our side before the build begins.
We also want basic conditional logic — if the transcript is under a word count threshold, the workflow flags it as incomplete rather than processing it.
What we’re looking for
Someone who has built Make workflows that call external APIs — ideally the Claude or OpenAI API specifically, since the response structure matters. Notion and Google Drive experience is essential. We’re not a technical team on our side, so we need someone who can map the logic clearly before building and explain decisions without jargon.
Scope and timeline
We’d like the first workflow built and tested within a week of starting. We have two additional workflows we’d like to build after this one, and we’d prefer to work with the same person if the first engagement goes well — but the immediate brief is the workflow described above. We’re happy to pay for a small paid test task before committing to the full build if that’s how you prefer to work.
To respond
If you’ve built something similar — a Make workflow calling an external API and writing output to Notion — we’d love to hear about it in a few sentences. What did it do and what was the interesting part?
1 Like
Hello @gcain2086 , welcome to make.com community, I have worked and have experience with Make.com and l will love to collaborate with you on this you can schedule a call Here and you can checkout my upwork profile Here, for my pastworks and certifications
Hi gcain2086,
This is exactly the kind of structured automation I build.
I’ve designed multiple Make workflows that call external APIs (OpenAI and Claude-style structured responses), parse JSON outputs, and write formatted results into Notion databases. The most similar build was a content automation system where:
• A file (or form input) triggered the scenario
• Text was sent to an AI API with a strict JSON schema
• The response was validated and parsed using Make’s JSON tools
• Structured outputs were mapped into Notion properties and formatted page sections
• Conditional logic handled incomplete inputs and fallback flows
• Notifications were sent via Telegram/Slack
The interesting part wasn’t the API call — it was enforcing response structure reliability. I implemented schema-controlled prompting + JSON parsing + error handling to prevent malformed outputs from breaking the workflow. That’s usually where these builds succeed or fail.
For your workflow, I would structure it as:
- Google Drive “Watch Files” trigger
- Extract transcript text
- Word-count check (router: process vs flag incomplete)
- Claude API call with strict JSON schema
- Parse + validate structured response
- Create & format Notion page (sectioned blocks for each output type)
- Telegram bot notification
- Error handling route
Since you’re not technical, I’ll map the logic clearly before building and explain each decision in simple terms. I also document workflows so your team understands how they work.
I can complete and test this within a week, and I’m open to a small paid test task first.
You can reach me at: fopefoluwaakinola@gmail.com
Happy to discuss next steps.
Hi gcain,
This is exactly the kind of workflow I work on regularly.
I’ve built production-ready Make automations that trigger from Google Drive, send structured prompts to Claude or OpenAI, handle JSON responses properly, and create clean, formatted pages in Notion — all with solid error handling and conditional logic in place.
For example, I recently built a workflow where meeting transcripts were uploaded to Drive → Make extracted the text → sent it to Claude → parsed a structured JSON response → created categorized sections in Notion (tasks, insights, risks, next steps) → and notified the team automatically.
The key part was making sure the JSON structure was reliable and adding fallback handling so the workflow wouldn’t break if the AI response slightly changed.
For your build, I would:
Clearly map the logic before development
Add the word-count threshold check
Structure the Claude prompt for consistent JSON output
Format and populate the Notion page cleanly
Set up the Telegram notification
Include proper testing and error handling
I can start right away and deliver within your one-week timeline. I’m also happy to begin with a small paid test task if that works for you.
If this sounds aligned, I’d love to review your Notion setup and transcript format so we can plan it properly and get started.
You can message me here also
+2349046842148
With experience integrating APIs like Claude and OpenAI, and working with Notion and Google Drive, I’m confident I can create an efficient solution for your content strategy needs.
I specialize in building clear, scalable workflows and explaining technical details in an easy-to-understand way for non-technical teams.
Check my profile: https://www.upwork.com/freelancers/farhana401
Book a quick call:Calendly - Automaxion
Hi @gcain2086,
Welcome to the community.
Your workflow is clear and very achievable in Make. I’ve built similar automations where Make watches Google Drive, sends structured prompts to Claude or OpenAI, enforces clean JSON responses, then maps the outputs properly into Notion databases with conditional logic and notifications included. The key part is validating the JSON and adding fallback handling so the scenario doesn’t break if the API response shifts slightly.
A few quick questions:
• Will transcripts always be plain text files?
• Do you already have a defined JSON schema for Claude’s response?
• Is the Notion page going into an existing database with predefined properties?
One week is realistic for building and testing this properly.
Website Portfolio: Check out my portfolio for related automation setups here.
Book a call: Schedule Jobscope walkthrough session here (P.S: Adjust to your timezone when booking)
Fiverr Workspace: http://www.fiverr.com/s/o8rbqXN
Email: folafoluwaolaneye@gmail.com
Best regards,
Folafoluwa Stephen
Flowedge Digitals
Hi @gcain2086,
Your brief is clear and well-structured — the fact that you’ve mapped out the exact flow (Drive trigger → Claude API → JSON parse → Notion + Telegram) tells me you’ve thought carefully about what you need.
**What makes me a strong fit for this:**
I work with the Claude API daily — not just calling it, but specifically designing structured JSON prompts with schema enforcement. The distinction matters because Claude’s response structure behaves differently from OpenAI’s, particularly around nested JSON and array handling. I’ve built production systems that route between Claude Haiku (for speed) and Sonnet (for complex analysis) based on input complexity — something that could optimize your costs as volume grows.
**A recent similar build:**
I built an AI-powered business intelligence system for an Amazon seller — it ingests data files, sends them to Claude with structured prompts requiring specific JSON schemas, parses the structured outputs into categorized sections (performance metrics, optimization recommendations, action items), writes everything into a database, and notifies the team via Slack. The interesting part was designing the prompt to return reliable, parseable JSON every time — I use a combination of explicit schema instructions in the system prompt, response validation, and a retry mechanism for malformed outputs.
**For your workflow specifically, one thing I’d flag:**
The word-count threshold check should happen *before* the Claude API call, not after — saves you API costs on incomplete transcripts. I’d also suggest adding a hash check on the transcript filename to prevent duplicate processing if the same file gets re-uploaded.
I’m comfortable delivering the first workflow within a week and happy to start with a paid test task. I’m based in India (UTC+5:30), IIT incubated, and have been building custom AI automation systems for 1.5+ years.
Happy to discuss specifics — you can reach me at priyanshukumarmaurya2224@gmail.com