We know that maintaining a consistent social media presence and activity is important for reaching new clients and staying visible. However, we aren’t marketing specialists, we are problem solvers who enjoy at building solutions to various real-world challenges/problems.
Our slogan is “Web & AI solutions that grow your business” and we do want to apply those same solutions internally. To help us with our own presence on social medias, we built a workflow that lets the AI handle the content creation (mostly advertising rather than announcements or like this blog) while making sure that every post still receives human review before publication.
The Workflow: From Website to Post
This pipeline is created using n8n, a visual workflow automation platform that allows us to connect tools, services, and custom logic into a single automated process (we self-host). Every week, our automated process pulls our website, nordbyte.co, to gather the latest updates, services, and project labs. Then this information gets feds to a LLM (in our case, Claude by Anthropic) to draft a professional LinkedIn post tailored to our brand voice.
Our Control Checkpoint: Slack
While AI is powerful it does have some down sides to it, we are aware of the risks of “hallucinations”, inaccuracies and off topic content can still occur, which is why our workflow includes a human approval.
Before anything goes live on LinkedIn, the system sends a draft of the generated post to our team Slack channel, where we can review it and choose one of two actions: Approve or Decline.
- • If we Decline: The AI receives our feedback and immediately generates a new version.
- • If we Approve: the post is automatically published to LinkedIn and archived in our company wiki for future reference
So, how did it turn out?
You can see the final result in the screenshot below. This is a post that our pipeline pulled together, and it honestly saves us a ton of time. It’s a simple way for us to stay active on LinkedIn without it becoming a chore, all while keeping everything self-hosted so we only pay for the tokens we actually use.
Because the entire workflow is self-hosted, we also maintain full control over the process and only pay for the AI tokens we actually use. The result is a scalable, cost-effective system that supports our “marketing” efforts without becoming another tool that needs constant attention.
Improvements
This workflow is only the beginning. We’re already exploring ways to expand the pipeline and improve both the quality and efficiency of the content it produces:
-
• Cost & Efficiency: We are optimizing our token usage by filtering out “noise” from our website and cleaning data locally before it ever reaches the AI. This ensures we only pay for the most pertinent information.
-
• A “Team” of Agents: Rather than relying on a single AI agent, we’re experimenting with a multi-agent approach. A Critic Agent could identify weaknesses in a draft, a Refiner Agent could improve clarity and engagement, and an Evaluator Agent could verify that the final result meets our quality standards before it reaches our review queue.
Conclusion
This project demonstrates how businesses can move beyond simply using AI tools and begin integrating them into practical workflows that deliver measurable value.
While we’ll never automate away the human judgment that defines Nordbyte, we’re always looking for opportunities to eliminate repetitive tasks and focus our time where it matters most.
Whether it’s cleaning messy data, automating internal processes, or building AI-powered workflows, we collaborate with clients from discovery through delivery to create solutions that improve efficiency and support long-term growth.
-
Thanks for reading!