Changelog

Record anywhere — even offline. That's the headline of this release, alongside big upgrades to shared pages and embeds, and AI summaries showing up everywhere. Here's what's new.

Record offline, upload later

The mobile app now works fully offline. Record in a basement, on a flight, or at a client site with no signal — everything is saved safely on your device and uploads automatically the moment you're back online. Uploads also got much more resilient along the way, so flaky connections and network drops no longer get between you and your recordings.

You can now download your files straight from the mobile app, too.

Bulk url imports just got easier

You can now paste urls directly in our web app without having to create a CSV file and import multiple urls quickly using the same settings

Embeds you can make your own

Embedding transcripts on your own site got a serious upgrade. You can now customise the look to match your brand and set the exact height that fits your page. Want more than a transcript? Show Insights and the AI chatbot right inside the embed. And sticky tabs keep transcript, insights and chat one click away while scrolling.

Shared pages that do more

People you share a recording with can now copy the transcript, download it, and search inside it with the new Find button — no account needed. And opening the share dialog now generates your link instantly, one less click to get it out the door.

AI summaries, everywhere you look

Every new recording now gets an AI summary automatically. You'll see it at a glance in your recordings list, and the "transcript ready" email includes it — so you know what's inside before you even open it. The Summaries tab is now called Insights, to better reflect everything it can do.

Auto-rename recordings

Tired of "New Recording (34)"? Turn on auto-generated titles and recordings name themselves from what was actually said. It's opt-in — per workspace or per upload — and you can add custom naming instructions in workspace settings, like always prefixing the client's name.

Smaller improvements

• Export any transcript as Markdown — ready for Notion, Obsidian, or your docs

• Each workspace now remembers your last selected transcription language

• Workspace search now uses a unified date-range picker

• Fixed dark mode colours in embeds, the transcript menu hiding on scroll in shared pages, folder picker scrolling with many folders, and a scroll-position reset in the mobile app

That's the lot. Tell me what you want us to improve next.
Praveen

Full list of changes

Features

• Add "Create your own transcript" button in shared links

• Add clipboard-write permission to iframe in share button component

• Add customizable embed height option in share button

• Add Find button to transcript menu in shared transcripts

• Add founder-interview email sequence for ICP validation outreach , closes #1

• Auto-generate recording titles from transcripts (opt-in)

• Custom instructions for auto renaming in workspace settings

• Customisable look for embedded pages

• Enhance clipboard functionality with error handling and internationalization support

• Implement clipboard copy functionality for transcript text with fallback support

• Improve reliability of uploads in mobile app and recovery for network issues

• Include AI summary in the transcript-ready email

• Mobile app works offline for recording and uploads when online again

• Option to show insights and chatbot in embedded pages

• Remember last selected language for each workspace

• Show copy and download buttons in the shared transcripts

• Show summary of a recording in the list

• Simple one url per line input for bulk importer

• Sticky tabs in embeds to switch to other pages easily

• Support for exporting transcripts in markdown format

• Support for file downloads from the mobile app

• Workspace preference and upload option for auto-generated titles

Bug Fixes

• Bug fix for folder picker scrolling with many folders

• Bug fix for transcript menu hidden on scroll in shared pages

• Bug fix with theming in dark mode embeds

• Fix emails using non production links

• Fix for mobile app list scroll position reset

Your recordings just became a knowledge base your AI assistant can actually use. Connect Transcript LOL to Claude or ChatGPT, and your agent can search your meetings, pull transcripts, read insights, organise recordings, and more — all through natural conversation.

Ask questions, get answers from your meetings

Once connected, your agent has access to your entire workspace. It can search across every recording, fetch full transcripts and translations, read AI-generated insights, and discover your folders and speakers.

Some things you can ask:

• "What did Sarah commit to in yesterday's standup?"

• "Find every mention of the Q3 budget across my recordings this month."

• "Translate the transcript of my interview with the Madrid team and pull out the action items."

Your agent can tidy up, too

The agent doesn't just read — it can organise and act on your workspace. Try:

• "Read the transcript of this morning's interview, figure out who's speaking, and label the speakers."

• "Set up a folder for each client and file my recordings from this quarter into them."

• "Create a new insight prompt that extracts action items with owners and deadlines, and save it to my workspace."

The speaker one deserves a special mention: agents can read the conversation, infer who's who from context — introductions, names mentioned, who's asking versus answering — and assign the labels for you. What used to be a manual cleanup chore is now a one-line request.

A smarter search to power it all

Agentic access is only as good as the search behind it — so we rebuilt it.

Search now supports semantic mode: instead of matching exact words, Smart mode understands what you mean. Searching "budget concerns" will surface the moment someone said "I'm worried we're overspending." Prefer literal matching? Toggle to Exact mode.

The search dialog got a full redesign too — filter by speaker and date, see match scores on results, and jump straight to the exact moment in a recording with timestamped seek links.

Full REST API for developers

Everything the agents can do, you can build with. This release adds REST endpoints for:

• Search — query your recordings programmatically, with the same semantic search

• Prompts — manage your workspace's insight prompts

• Speakers — create and manage speakers

• Organisation — move recordings between folders

Plus transcript and translation fetching now supports time-range offsets, so you can pull just the segment you need.

Smaller improvements

• Speaker names now appear in transcripts and translations fetched by agents — no more "Speaker 1" when you've already labeled everyone

• Timestamps in the transcript view are now seekable, with fixed auto-scroll and deep links

• Polished search result cards

Getting started

Connecting takes about a minute — Transcript LOL runs as an MCP server, so it plugs into Claude, ChatGPT, or any MCP-compatible client. Add it as a connector, sign in, and start asking. Follow the setup guide here.

Your meetings have always held the answers. Now you can just ask for them.
Praveen

We've overhauled how transcription works under the hood — and the result is a meaningful jump in accuracy for everyone, with no change to how you work.

Tested against the competition

We didn't just test on clean studio audio. We benchmarked our new defaults against leading models and competing services across 16 test suites — from clean baselines (LibriSpeech) to the hard, messy stuff that reflects real recordings: distant-mic meetings with overlapping speakers (AMI), real earnings calls with heavy jargon (Earnings-22), podcasts (Rev16), and diverse real-world media (Kincaid46, People's Speech). We also tested across four languages beyond English.

The takeaway: on par or better than every model and competitor we tested — and the gap is widest exactly where it's hardest, on noisy, real-world audio. Same speed you expect, noticeably better output. 

Choose the balance that fits the job

You can now pick a transcription mode for the trade-off you care about — from Standard (our most accurate, and the new recommended default) to Balanced and Fast when you need speed.

There's a Verbatim style that keeps filler words, plus experimental modes for those who want to test the latest.

What's next

This work isn't just about today's numbers — it lays the groundwork for domain-specific models. Specialised fields like medicine come with terminology that trips up general-purpose models, and we're now positioned to ship models tuned for exactly that. More on this soon.

As always, nothing to configure.

Just start transcribing,
Praveen