The context
Learning and knowledge work includes recurring tasks around source material, drafts, transcription, translation and documentation. These tasks offered a practical setting for exploring where AI assistance might fit.
At Pickado, my technology role connects internal platforms and tools with the needs of learning teams and management. This case focuses on internal-tool work and prototyping, rather than a claim of enterprise-wide AI transformation.
My contribution
I built and explored tools supporting transcription, course creation, translation and knowledge access. My technical foundation includes Python, FastAPI, API integration, LLM applications and RAG prototypes.
The work also sits alongside infrastructure, identity and documentation responsibilities. A tool needs a working environment and a clear place in the team’s process; producing an answer is only one part of that setting.
The approach
The starting point was a specific workflow and its source material. Prototyping made it possible to inspect what a tool produced and consider how it might support a person’s work.
For the next stage of any such project, my preferred approach is to define the intended task, the review process and the information boundary before widening use. That is a working principle, not an assertion that every prototype has passed a formal evaluation or reached production scale.
What the evidence supports
The supported outcome is practical internal tools and prototypes. This public case does not claim measured time savings, a usage total, production RAG accuracy or external customer ROI. Those would require a separate evaluation with an agreed baseline and appropriate data.
This distinction matters because a prototype and an adopted service answer different questions. A prototype can show a possible approach. A service needs evidence about repeat use, ownership, reliability and the experience of the people depending on it.
The lesson I take forward
AI work becomes easier to discuss when the task is concrete. I want to connect technical exploration with discovery, explanation, demonstrations and evaluation — helping people judge whether an idea is useful before calling it a transformation.
No internal configurations, security diagrams, private datasets or application screenshots are included. A shareable demonstration should use a separately prepared, sanitised environment.