# More control, less waste: on-demand content and brand voice learning

Date: 2026-03-13T23:09:06.983Z

When we built Notra, we bet that most teams would rather automate what goes to their audience than tweak it by hand every time. That bet still holds, but we missed something. Some releases don't deserve coverage, some commits aren't worth mentioning and automation doesn't help if the output doesn't sound like you.

This week we shipped three changes to how you generate content. All three let you decide what gets made and how it sounds.

## Generate what you want to ship

For the first two months, Notra worked like this: you set up a trigger, the AI looked at everything since the last run and drafts appeared. That worked for broad coverage and broke down when you wanted to be selective.

We got requests like "I want a changelog for our three customer-facing repos only, without internal tools and only commits from the last five days." Or "I'm writing a blog post about performance improvements and want to pick which PRs matter."

So we rebuilt the generate flow. When you create content on demand, Notra first shows a preview of the commits, PRs and releases it found. You check the ones that belong in your story, turn releases on or off, pick a time window and scope the run to specific repos. You control the scope instead of hoping the AI guesses right.

The preview endpoint paginates your data, because large repos can time out a request. Your filters travel with the generate request, so the AI only sees what you picked.

## Content in your voice

We also kept hearing that drafts read well but didn't sound like the team. People edited our drafts because the tone was off for them.

This week we shipped brand voice learning. Add references from your Twitter account or paste custom text, and Notra's agents study them before generating anything. They match word choice, rhythm, sentence structure and which ideas you put first.

References use Twitter OAuth with encrypted token storage. You connect in one click, import your last 20 tweets and pick the ones that define your voice. You can add custom examples for platforms we don't import from.

You can also scope references by platform. A tweet reference can shape Twitter posts and blog posts, while a blog excerpt only shapes blog posts.

## Instant publish for your schedule

Generated content usually starts as a draft that you review, edit and publish. For teams on daily schedules, that review is one more task.

Triggers now have an "Auto-publish on this trigger" toggle. Turn it on and Notra publishes content as soon as it's generated. On-demand content from your dashboard still starts as a draft, since you probably want to check something you asked for, but scheduled runs can publish directly.

## What comes next

We're working on automatic voice detection, so Notra infers your tone from your references without you tagging them. We're also making filters combine, so you can ask for "commits from the API team in the last week, excluding docs changes" and get exactly that.

Our goal is for you to point Notra at your data once and get drafts that sound like you wrote them.

## Find out what AI agents say about you and change it.

Add a few prompts and run a scan to read the answers. Notra is free to start.

[Start for free](https://app.usenotra.com/signup)

[Book a Call](https://www.usenotra.com/contact?ntr=LUTW2GpWvHqdh4So)