📦 GitHub: paoloronco/makecom-openai-gcp-instagramreel
📣 Publishing on ReadyTensor: Automated Instagram Reels with AI

Introduction: More than a social project

The “AI Instagram Reels” project was born with the aim of automating the creation of content for social media, but in reality it turned out to be much more: an advanced exercise in orchestration between generative AI, cloud computing, and no-code automation.
During the development, I acquired transversal skills in:

• Complex automations with Make.com
• Using OpenAI APIs (ChatGPT, TTS, DALL·E)
• Managing files and buckets on Google Cloud Platform
• Container deployment on Google Cloud Run
• Integrate multiple tools into a single, seamless workflow

More than a project “for Instagram,” it was a training ground for building a scalable and intelligent infrastructure capable of automatically generating and publishing multimedia content, starting from a text prompt.

---

Technical infrastructure: AI, Cloud, automation and file management

Generative AI: Original content at the heart of the project

The creative heart of the project is OpenAI's artificial intelligence. Using GPT-4o, I set a specific prompt to generate short but engaging scripts, optimized for narration in a Reel. The text is then transformed into speech using the TTS-1 model, resulting in an audio file. .mp3 Clear and natural. To complete the visual aspect, DALL·E 3 generates a vertical, flat-style image, designed as a visual cover for the content.

Google Cloud Run: A Tailor-Made Video Microservice

One of the most interesting and advanced aspects of this project was undoubtedly the development of the backend on Google Cloud Platform, in particular the use of Cloud Run combined with Docker.
I made a custom microservice, containerized via Docker, capable of dynamically merging the file .png (cover image) with the file .mp3 (voice-over) and return a video .mp4To do this, I have:

• wrote a Python script (main.py) for merging files
• configured a Docker environment to run
• created a Docker repository on Artifact Registry
• deployed the service on Cloud Run, making it publicly accessible via HTTP POST

The infrastructure is highly scalable and serverless: Cloud Run runs the service only when called, reducing costs and ensuring consistent performance.

To connect Make.com to this endpoint I also created a Service Account with the appropriate permissions, configured OAuth and defined the necessary policies to allow the secure communication between the two environments.

File recovery and management: everything goes through the cloud

Once the video is generated .mp4, is saved on Google Cloud Storage, which serves as a central repository for all temporary and final files. To get the public link to the most recent file, I integrated an API call directly from the Make.com pipeline, which queries the bucket and returns the files sorted by creation date. This way, I can unambiguously locate the correct output.
This feature is essential to keep automation running smoothly, especially when the system runs multiple times a day.

Google Drive: Backup and traceability archive

Google Drive is not used as an active work environment, but as backup space, useful for always having a copy of the generated content available. In the Make.com pipeline, I've inserted two specific steps that upload the raw image and audio to a shared Drive folder.
To achieve this integration, a simple connection was not enough: I had to go through the Google Cloud Console, create a project, enable the necessary APIs, configure an OAuth authentication system and create a Dedicated Service Account to allow Make.com to access Google Drive in a secure and controlled way.
This phase was very formative, because it allowed me to understand how the complete credential cycle between third-party environments and Google Workspace works.

---

Conclusion

“AI Instagram Reels” was not just a creative project, but a true distributed architecture, built through the deep integration of advanced tools such as OpenAI, Make.com and Google Cloud Platform.

Far from being a no-code project, this work required real technical skillsFrom writing Python code for video generation, to managing Docker containers, to configuring service accounts, OAuth permissions, and deploying to Cloud Run.

The result is a fully automated, yet highly controllable, scalable and customizable, capable of producing video content from scratch, and doing so with precision, speed and consistency.

More than a flow to create Reels, it is a intelligent multimedia production machine, ready to be reused, extended, or adapted to other contexts — educational, branding, social media automation, or content marketing.

This project strengthened my ability to manage complex cloud environments, connect external APIs, design robust pipelines, and write code that communicates with distributed infrastructures. This experience is something I'll bring with me to every new project.