Templates AI

Disclaimer: I am unable to share direct images or provide further insights, as this work is confidential and must be treated as such.

Project Objective:

As a natural extension of the Toolbox Creation Assistant, Templates AI was designed to make content creation even faster, more consistent, and scalable. By combining AI-generated input with pre-built, reusable page templates, Templates AI helps admins, area managers, and content managers create fully structured, on-brand pages in a fraction of the time. The system builds on everything users do in the Creation Assistant — taking their natural language inputs, auto-generated tags, and images, and intelligently injecting them into flexible, multi-language templates.

Initial Requirements:

The Templates AI project was defined with clear goals:

• Save time for users by merging AI-generated content with ready-made layouts, reducing repetitive page building tasks.

• Ensure templates match each customer’s corporate identity while staying compatible with Toolbox’s modular design system.

• Integrate all supported AI models, translation tools, and open image sources like Unsplash — instead of relying on AI image generation.

• Enable templates to work across multiple languages and deploy seamlessly on all innosabi platforms.

Design Elements:

• Core concept, flow, and visual design for the Templates AI integration — including wireframes, mockups, and early prototypes.

• Modular selection UI to choose templates, with clear preview and customization options.

• Advanced prompt engineering to ensure generated text aligns naturally with each template’s structure and tone.

• Flexible backend logic to map user-generated data to specific content slots in the template (headlines, text blocks, visuals, meta fields).

Process Highlights:

• Researched best practices for reusable template systems and smart content mapping in modern page builders.

• Ran collaborative sessions with product and tech teams to define the underlying data structures and API that enable custom and predefined templates to work seamlessly across different CI variations.

• Built and iterated multiple prototype versions — validating prompt outputs and template population logic through LangSmith workflows.

• Oversaw the end-to-end handover: driving concept and initial prototypes, handing off to a colleague for final design polish, and then re-owning the implementation phase for launch readiness.

Lessons Learned:

Building Templates AI revealed new insights that pushed our platform forward:

• Flexibility is everything: Designing an API and data structure that can host both standard and custom templates — while adapting to different customers’ brand styles — demanded careful planning and future-proofing.

• Human-AI collaboration is practical, not just hype: Combining user prompts, AI output, and predefined design blocks delivered the right balance of automation and creative control.

• Strong prompt engineering is the backbone: Developing prompts that generate content that fits naturally into fixed templates required extensive testing and tuning — but paid off in usability and output quality.

My Role:

I led Templates AI from the initial concept to final implementation. This included shaping the vision, researching user needs, building wireframes and mockups, running early user tests, and developing prompt strategies within LangSmith. After handing over design finalization to a colleague, I returned to oversee implementation, coordinating with developers, AI specialists, and internal teams to ensure a smooth rollout. Templates AI is now a core piece of our Toolbox offering — helping users publish polished, consistent content faster than ever.