MiniMerkle: from brief to campaign draft

At Merkle Norge, I helped develop MiniMerkle, an AI prototype for turning campaign briefs into draft content. My work focused on the OpenAI integration and the Next.js interface.

November 17, 2024

MiniMerkle logo in dark blue and red on a white background.

MiniMerkle started with a practical problem: turning campaign briefs into draft content for different channels. At Merkle Norge, we built a prototype that used AI (artificial intelligence) to help with that first pass.

A brief could become a campaign outline with platform guidance, placeholder images and suggested content. The idea was to give the team something concrete to work from.

Merkle's 2025 write-up explains how GPT-4o analysed the brief, a prompt library adjusted the brand's tone, and generated search terms found images in a DAM (digital asset management) system.

My contribution

I joined the project later in development. I helped connect its core functionality to the OpenAI API (application programming interface) and worked on the Next.js interface. This was a team project at Merkle Norge and dentsu Norge.

Adapting it for different brands

The app used Tailwind CSS for styling. JSON (JavaScript Object Notation) configuration files held client logos, design settings and system prompts, so the same tool could be adapted for different brands.

MiniMerkle in use

The demo shows brand switching, suggested campaigns with images, and an editor with options for email, Instagram and banners.

MiniMerkle demo from 2024. The interface shows a brand menu, proposed campaigns with images, and editable content with email, Instagram and banner options.

The result

MiniMerkle was still a prototype when Merkle reported its results in 2025. The article describes a campaign being reviewed and ready to publish in under an hour, compared with a usual 7 to 20 hours. Those are Merkle's reported figures for the project.

MiniMerkle won silver in the AI-generated content category at the ANFO MarTech Awards 2024. ANFO credits Merkle Norge and dentsu Norge for the project.

For a new campaign, a tool like this could give the creative team a first draft to discuss, edit and develop further.

Keep reading

GenAI / MCP / Productivity

How I connected my assistant to ComfyUI

I built Comfy2MCP so my coding assistant can run approved ComfyUI workflows, change configured prompts, and save the results locally.

Terminal demo showing local CrewAI agents requesting an Entur journey from Grorud to Røverkollen.
GenAI / MCP / Productivity

How I connected local AI agents to Entur

I built an MCP prototype using Entur's open APIs, CrewAI, and Llama 3.1 8B through Ollama to plan Norwegian public transport journeys.