Hi there,

A few months ago you downloaded the open-source version of Doza Assist. I want to tell you what's happened since, because you're part of the reason it happened.

Fifty-plus people paid for the full version after that first release. Every one of them got a line to me. They sent bug reports, screen recordings, half-finished story briefs, and "why doesn't it do this" emails at 2am. Almost everything in 1.1 traces back to one of those messages. This isn't a release I made for them. It's one they made with me.

Here's what's in it.

Claude can sit in on your project. This is the one.
Doza Assist now ships a Model Context Protocol connector for Claude Desktop. Turn it on for a project, open Claude, and ask it anything about the interview. "Where does she talk about her father?" "Pull me every moment that would work as a cold open." "Which of my selects contradict each other?" It reads the transcript, searches it, looks at your selects, and adds labeled clips straight into your project. And it's almost instant. You ask, and the clips are just there. The first time I watched it drop six labeled selects into a project I hadn't opened in a week, it felt like magic. It still does.

It's off by default and you enable it per project. Claude can't delete anything, can't see projects you haven't turned on, and your media never leaves your machine. It reads the words. You keep the footage.

Fix the transcript without re-transcribing.
Double-click any word and correct it. Insert the word the model missed. Split a paragraph where the speaker changes. Reassign who said what. Timecodes hold across every clip and export.

Collections.
Group multiple interviews and work on them as one. Transcript, Chat, Story Brief, Story Builder, Clips, and Export all run across the whole set. Ask a question and chat pulls from every interview at once. Multi-subject docs finally have a home.

Chat got a real upgrade.
It reads the whole interview now, answers in plain language, and comes back in seconds on interviews that used to time out. Ask for "no clips" when you just want prose.

The smaller things that add up.
Speakers get their real names pulled from introductions instead of "Speaker 2." Clip trims are word-by-word with an audition preview. Exports name themselves sensibly and Story Builder sequences go straight out. Quote sheets match press formats. Final Cut clips carry the full transcript in the Notes field. And 36-hour recordings transcribe without falling over. Dark mode is the default now, which about half of you asked for and the other half will forgive me for.

Full notes are here: https://doza.ai/changelog/1.1/

If you've been on the fence.
Code LAUNCH50 takes the full version from $199 to $149. If you want to feel the MCP connector before you buy, it works in the trial on the sample transcripts. Open Claude, ask it one question, and you'll get it.

I'd rather you upgrade because 1.1 solved a problem you actually have, not because of a discount. So if you open it and something still isn't right, reply to this email. That's exactly how everything above got built.

Thanks for downloading it in the first place. It mattered.

Chris
Doza Visuals