Bold projects and initiatives we get to support.

The challenge

A shared AI assistant was needed for two different consulting scenarios, in-person and online. The core requirements first had to be worked out with the team and translated into a shared software solution.

How we solved it

In a workshop, we worked with the team to define the core requirements for both usage scenarios using the Design-to-Code framework. In an initial prototyping phase, we tested the software directly in practice and incorporated the results iteratively.

The result

A production AI assistant that covers both consulting formats. We also built a scraping tool for automatically updating the AI model's database using the Design-to-Code framework. We provided the design system via Figma tokens. Everything was built with a tool chain of Claude, Git, and a locally hosted server setup.

The challenge

The marketing department needed to be able to build and maintain the new website and all further venture sub-pages independently going forward, instead of depending on external development for every change.

How we solved it

Across several modules, we worked with the project and marketing team to define the requirements for the new website and build its content structure. Building on the existing CMS, the marketing team independently implemented and coded all further venture pages using the Design-to-Code framework. This was based on a design system hosted as HTML tokens, extended and maintained for each new venture page.

The result

A web platform continuously developed and maintained via Claude, built from Wix Studio modules and ChatGPT as the LLM, used to communicate current news and updates on venture initiatives and holdings. The marketing team has maintained and extended the pages independently ever since.

The challenge

Agenda wanted to integrate DesignOps and the user-centered approaches of the Design-to-Code framework into their existing, automated development process.

How we solved it

In a workshop format, the strategic gaps were discussed from a developer perspective together with DtC experts. Building on that, the existing process was combined with the Design-to-Code framework, and the key tasks for adapting the workflow were defined.

The result

An update to the existing, automated software process aimed at introducing Human Stage Gates as well as accessibility and usability requirements — for a better, more user-centered output. Dev and PO were also trained remotely on UX requirements, so that as AI Product Engineers they could better answer the questions AI raises during automated coding.

The challenge

For a digital rehab program, a solution was needed that digitally accompanies and supports patients between in-person appointments at the practice.

How we solved it

We trained the responsible product team step by step while simultaneously coding the mobile application with AI support. Regular user testing accompanied the process, with the results continuously fed back into refining the application.

The result

A personalizable mobile application for patients with customizable training sessions. We also built a headless CMS for creating training plans and managing patients. Steering and further development happen via the Design-to-Code software, supported by an internally trained AI Product Engineer.