Tendium

Background

In 2019, before AI tools like ChatGPT and Grok became widely available, I was invited to support Tendium AB, an AI-focused startup, for a couple of months during a formative stage of its technical development. The company’s growing team included part-time students from KTH and master students, many of whom were transitioning from academic research to commercial product work for the first time.

My role was to act as a technical advisor and mentor. Together with the team we establish structured, sustainable development processes while evaluating suitable tech stack and workflow methodology for the company.

(Note: This collaboration was personal and not conducted under any formal agreement with my current company, TenSrc AB, which wasn’t founded yet.)

The Challenge

The newly formed data science team brought strong analytical and academic capabilities but limited experience in commercial software production - including areas such as workflow management, testing, and delivery cycles.

The goal was to introduce clear processes and tools, build shared terminology, and create a foundation for efficient collaboration, scalability, and knowledge transfer.

‍

Our Approach

We began by understanding each team member’s background, preferred working style, and strengths. From there, we co-created processes that laid the foundation for communication, documentation, and delivery efficiency.

Key steps included:

• Communication: We categorized communication types and assigned the right tools. Temporary updates and casual chats went to Slack, while long-term or sensitive information was documented in Confluence for easy retrieval.
• Terminology & Naming Conventions: We emphasized consistent naming across the code base, UI, and database, reducing confusion and improving on-boarding and handover processes.
• Project Naming: To make projects memorable and organized, the team chose astronomy-inspired names.
• Handover Process: Teams documented their projects in Confluence and tracked progresses in Jira, then exchanged documentation to test clarity. Exercises produced comprehensive, reusable documentation and strengthened internal communication.
• GUI Framework: When a new developer with Vue.js expertise joined, we pivoted to that framework and used peer training to spread the knowledge.
• Agile & Meetings: We introduced Scrum gradually, rotating the Scrum Master role every few months to develop leadership skills across the team.
• Database & API Structure: We implemented clear documentation standards for data fields, formats, and naming conventions, minimizing redundancy and technical debt.
• Automation & Testing: Build automation routines and test data sets were established, covering GUI stress tests, edge cases, and special characters.
• Internationalization: We separated human-readable text from source code to prepare for future localization.

The Outcome

Through this mentorship and collaboration, the team evolved into a well-structured, confident, and highly functional group capable of managing their own workflows and processes.

The improved documentation and knowledge-sharing practices reduced dependency on individuals and strengthened long-term scalability.

The company gained a clear, maintainable technical foundation and continues to grow on that base.

Why It Worked

The success came from a collaborative, educational approach—not imposed processes. By combining technical guidance with practical mentoring, the team gained confidence, organization, and sustainable development habits that carried forward into future projects.

Summary

Context: Early-stage AI startup with a mixed team of students and developers transitioning to commercial product work.

Contribution: Personal mentorship and technical advisory—introducing agile methods, communication structure, and documentation standards.

Outcome: Independent, efficient team with scalable workflows and minimized technical debt.

Collaboration Type: Personal collaboration (not under TenSrc AB).

‍