Case Studies

Built for ourselves first.

Dictio was born inside Zehntech, and our own teams were its first users. These are the two problems it solved for us before it ever had a website.

EngineeringAI coding workflows

Prompting at the speed of thought

The problem

Our engineers spend a large part of their day directing AI coding tools like Claude Code and Cursor. A good prompt is long: the architecture, the constraint, the edge case, all spelled out. Typing 200 to 500 words at around 40 words per minute, several times an hour, meant our fastest thinkers were working at the speed of their fingers. On long days, wrists felt it too.

What we did

We put push-to-talk dictation on every engineer's machine: hold the hotkey, describe what you want as if you were explaining it to a colleague, release into the tool's text field. The cleanup pass strips the filler and fixes the punctuation, so a spoken ramble arrives as a readable brief.

What changed

Prompts now happen at speaking speed. A 300-word prompt takes about two minutes out loud instead of ten at the keyboard. The surprising part: prompts got better, not just faster. When describing context costs nothing, people include more of it, and the AI's output improves with it. Dictation is now the default way our engineers talk to their tools, and the reason Dictio exists as a product.

150 vs 40words per minute, speaking vs typing
~2 minfor a 300-word prompt that took ~10 typed
Dailydefault input method across the engineering team
Delivery teamsDocumentation

From call recordings to same-day notes

The problem

Client calls, sprint reviews, and internal demos were being recorded and then quietly piling up. Notes got written from memory hours later, details drifted, and decisions were paraphrased instead of quoted. Subtitling product demo videos meant sending files out and waiting.

What we did

Recordings now get dropped straight into Dictio. Long calls come back as transcripts; the ones nobody will replay come back as summaries. Demo and tutorial videos get SRT or VTT subtitle files from the same tool. Everything lands in a searchable local history, so "what exactly did we agree in that call" is a search, not an archaeology project.

What changed

Call notes go out the same day, with decisions quoted from the transcript instead of reconstructed from memory. Subtitles happen in-house, the day the video is cut. And because transcription runs on our own key and nothing passes through extra servers, recordings of client conversations stay exactly where they should.

Same daycall notes and meeting summaries
5 in, 3 outMP3, M4A, WAV, MP4, MOV in; TXT, SRT, VTT out
0extra servers your client calls pass through
Why it matters: recordings of client conversations are sensitive. Dictio's bring-your-own-key design keeps them between you and OpenAI, with nothing stored on our side.

Your team has the same two problems.

Start free and see what changes in a week.

Download Dictio