Why we built summaries and action items into Scribr

Three months after launch, a therapist emailed us. She'd been using Scribr to record client sessions, and our on-device transcription was solving her privacy problem. But then she wrote something that stuck with me: 'I spend 45 minutes after each session writing my notes. What if I didn't have to?'

The problem we kept hearing

Before we shipped summaries and action item extraction, I'd get into calls with early users and hear the same thing over and over. A sales director would say, 'I record every call, but then I have to sit down and manually extract what actually matters.' A legal researcher mentioned she was transcribing interviews but spending twice as long reading through them as she did conducting them. A consultant told me his team would have transcripts but no way to track who was meant to do what.

These weren't edge cases. They were the core of why people record in the first place. They need a record of what was said, sure. But they also need to know: what's the summary I can send to someone who wasn't there, and what do I actually need to do on Monday morning?

We'd launched Scribr with on-device transcription, which solved the privacy piece. But transcription alone is just a transcript. It's passive. It sits there until you need it, which often means it doesn't get used at all.

Why extraction, not just transcription

The technical side was straightforward enough. We already had the transcript. Building summaries and action item extraction on top of that transcript was the logical next step. But the real question was why it mattered as a product decision, not just a feature.

Here's what changed for us: we realised that transcription is a solved problem. Deepgram does it well. Apple's Speech framework does it well for on-device work. The differentiation isn't 'we can turn audio into text.' The differentiation is 'we can turn your meetings into actionable insight without you having to read through 40 minutes of rambling to find it.'

So we built extraction into Pro. When you upgrade from free (which keeps everything on your device and gives you transcription), you unlock cloud transcription for longer audio and, more importantly, AI summaries and action item extraction. We set it at 500 calls a month because that's roughly what a knowledge worker doing a lot of meetings would burn through without hitting walls.

The Action Items Widget matters here too. It's the piece that brings extraction into your workflow. Once we extract what you need to do, you see it immediately. You don't have to open the app, find the note, re-read it. It's there, waiting.

A decision point on privacy and trust

There was a choice we had to make early on. We could extract everything by default. Give every user access to summaries and action items, paid or free. But we didn't, and I think that was right.

Scribr's free tier is fully private. Your audio never leaves your phone. That's the whole point of free. The moment you ask us to extract summaries and action items, we're running that against cloud infrastructure. That's where we need cloud transcription (via Deepgram) and where we need explicit consent. You're opting into something different.

This is boring to talk about, but it matters. A therapist recording a sensitive session needs to know what data leaves their device. A lawyer recording a deposition needs to understand the chain of custody. So we made it a tier boundary, not a surprise feature.

On Team, we add Contact Intelligence and Note Sharing, which means you're moving into a collaborative space where your notes are being accessed across a team. That gets GDPR Compliance Modes with an audit log. Not because lawyers told us we had to, but because if you're a team using Scribr, you need to know who accessed what and when.

The moment it clicked

Launch week of summaries and action items, a freelance consultant used Scribr to record a client call. She had six meetings that week. Each one generated a summary she could send to the client within minutes, and each one flagged the action items she'd committed to. She sent us a message saying she'd saved roughly eight hours that week just by not having to manually write up her notes.

That's when I understood what we'd built. It wasn't a feature. It was time. Every person whose work lives in conversations was getting a chunk of their life back.

That's also why we built the Action Items Widget and the Siri Get Action Items shortcut. We wanted extraction to live in your phone's ecosystem, not just inside the Scribr app. You're working through your week; you need to see what you've committed to without leaving your workflow.

Why this mattered more than we expected

Looking back, I underestimated how much of knowledge work is repetitive transcription and summary-writing. Researchers manually listing interview themes. Sales teams copying call outcomes into a CRM. Legal professionals writing case notes. Teachers noting down student progress from recorded sessions.

We built summaries and action item extraction because users asked for it. But we built it as a paid feature because it's a meaningful shift in what Scribr does. Free is privacy first; paid is productivity first. Both matter. Both have their place.

The constraint of a monthly extraction allowance (500 on Pro, 1,500 on Team, unlimited on Enterprise) exists for the same reason our upload limits exist on free. We want you to use Scribr for what matters, not as a dumping ground for every conversation ever recorded. Quality over quantity.

The real question isn't whether summaries and action items are nice to have. It's whether you're willing to record your work if nobody's reading the transcript. Once extraction is there, suddenly recording makes sense again.

Want to try Scribr?

Visit Scribr →