v1.1
Audio uploads, an extension, and the new pricing ladder
May 17, 2026Two weeks after launch, v1.1 lands the changes most-requested by early customers: a refreshed pricing ladder anchored by a $1 trial, audio file uploads, browser-tab capture via a Chrome and Edge extension, a real To-Do dashboard, two new languages, a Cornell-style Study Note template, pause and resume on every device, and a meaningfully faster mobile experience.
Added
- Public pricing reset to the 60-minute ladder, Trial 7-Day for $1 (30 minutes), Basic 60 at $2.99/mo, and Pro 60 at $4.49/mo. Annual versions of Basic and Pro save four months.
- Top-up packs with bonus minutes, 60 minutes for $1.99, 360 minutes (300 + 60 free) for $9.99, and 780 minutes (600 + 180 free) for $19.99.
- Pro 60 carries a 1-month rollover bucket, unused minutes roll into the next period and the remaining balance plus expiry date are shown in the dashboard, billing settings, sidebar, and mobile home.
- Cancellation grace period, keep using your account for three months on Basic 60 and six months on Pro 60 after cancelling.
- Audio file upload, drop in MP3, WAV, FLAC, M4A, OGG, or WEBM (up to 200 MB) and Loka transcribes it the same way as a live recording.
- Loka Capture browser extension for Chrome and Edge, one-click tab-audio capture for Google Meet, Zoom Web, and Microsoft Teams.
- Pause and Resume on the meeting recorder, on desktop and mobile, the produced audio stays as one continuous file and the paused interval is excluded from your usage.
- Crash-safe recording, chunks are kept locally in your browser, and an interrupted session auto-resumes after a refresh, tab close, or network drop.
- To-Do dashboard with List and Kanban views, action items pulled from your meetings become a real, customizable workspace with drag-and-drop columns.
- In-browser notification sounds when transcription or analysis finishes (and a distinct alert on processing failures), with an on/off toggle in Notifications settings.
- Study Note template, a Cornell-style analysis option (keywords, test questions, structured notes, summary, key takeaway) tuned for lectures and study material.
- Vietnamese and Chinese (Simplified Mandarin) added to the transcription, analysis, and default-language pickers.
- Premium redesign of the summary PDF export.
- Audio-reactive recording overlay on mobile, a level-reactive central mic replaces the old static bar.
- Copy-transcript action on every finished meeting.
- Network-quality probe on the recording setup screen, multi-sample median plus jitter, so you know your connection will hold before you press record.
- Password reset flow, explicit Terms and Data Processing Agreement acceptance for new sign-ups, and an optional marketing opt-in.
- Short-recording warning, confirmation before stopping a very brief recording, plus an in-summary banner if it slips through.
Improved
- Transcription quality on noisy recordings, every audio source now passes through a server-side cleanup pass (high-pass, mains de-hum, denoise, gentle compand, loudness normalization) before reaching speech-to-text.
- Mobile performance, cold-visit LCP dropped by roughly 300–800 ms across dashboard and meeting routes thanks to a smaller logo asset, deferred fallback fonts, and lazier analytics scripts.
- Mobile bottom navigation now surfaces To Do and Upload Audio for plans that include them, Trial and Pro users can finally reach both from their phone.
- Out-of-minutes upgrade dialog distinguishes “you have zero minutes left” from “this file is longer than what's left this period”, so the suggested fix matches the actual problem.
- Recording Setup gives an honest first-tap message (“Allow microphone access to record on this device”) and warns Messenger and Instagram in-app browser users about microphone restrictions with a path to open Loka in a real browser.
- Re-analyzing a meeting now uses a single quiet cap across every plan, no more visible counter or upgrade nag during normal use.
- Upload Audio source-file cap raised from 100 MB to 200 MB.
Fixed
- Stop-recording could fail with a 500 on accounts routed through audio cleanup, the post-recording pipeline now completes without error.
- PDF download hang on long meetings.
- Mobile settings redirect bug.
- Recording Setup panel scroll on short viewports.
- Echo and phrase-loop transcripts on unstable connections.
Changed
- MP4 removed from the Upload Audio accepted-file list, pick M4A for audio-only files (the OS file picker was letting through video MP4s that the server then rejected).
- Paid Trial is now one-time-only per account.
v1.0
Loka, live
Released May 4, 2026The official launch. After four weeks of building, shipping, and refining, Loka stepped out of private preview and opened to every team that wants their meetings to do more than fill a calendar. v1.0 was the product we'd been telling ourselves about since the first commit, fast, mobile-ready, and honest about what AI can and cannot do for a conversation.
At launch
- Self-serve signup on lokanote.com with secure integrated pricing, a $1 paid trial, plus Basic and Pro plans available monthly or annual (annual saves four months).
- Browser-native meeting capture, no bot, no install, no calendar plugin, no one sitting in the room who doesn't belong there.
- Real-time transcription in eleven languages, powered by AI speech-to-text with an automatic safety net.
- AI-generated summaries, action items, and key decisions grounded in the transcript, never a model guessing at what you meant.
- Conversational AI chat that treats every finished meeting as a queryable document.
- Drag-and-drop meeting folders with realtime sync across tabs and devices.
- Email notifications for every processing milestone, with per-category preferences and one-click unsubscribe.
- Optional email digests for product updates and changelog notes, opt in from notification settings, unsubscribe any time.
- Full mobile parity, record, review, chat, and organize from any phone without a second thought.
v0.9.5
Pre-launch performance & pricing
May 2 – 4, 2026The final days before launch were spent rewriting the pricing catalog to support annual plans, enforcing a mandatory payment gate, and significantly upgrading landing page performance and SEO.
Added
- Mandatory payment gate before dashboard access, ensuring all users have an active plan.
- Annual plans added to the pricing catalog, offering four months free on Basic and Pro tiers.
- Comprehensive SEO enhancements including metadata, structured data, canonical tags, and a sitemap.
- Pyidaungsu font support for Burmese characters across the app and PDF exports.
- Social links added to email templates and new Security and Data Processing Agreement (DPA) pages published.
Improved
- Landing page performance upgraded with enhanced scroll animations, lazy loading, and skeleton screens.
- Opened public signups, removing the waitlist in favor of a dedicated registration flow.
- Upgraded background processing infrastructure for faster, more reliable task execution.
- Updated application time zone to Asia/Yangon for accurate timestamping.
- Applied security and performance best practices across the platform.
v0.9
Mobile, fast, ready to ship
April 24, 2026The last release before launch is the one where all the thousand small things get fixed. Safe areas, keyboard handling, iOS audio quirks, viewport math, the stuff that makes a web app feel inevitable on a phone instead of feeling like a desktop product in drag.
Improved
- Cross-browser audio initialization, a dummy AudioContext is unlocked on the first user gesture, so iOS never clips the opening seconds of a recording.
- Safe-area-aware layouts for iPhone notches and Android gesture bars, applied consistently across the recording UI and dashboard shell.
- Recording screens now use 100dvh instead of 100vh so dynamic browser chrome on mobile never clips the controls.
- Popovers, dialogs, and the AI chat panel re-size to the viewport on small screens instead of spilling off-screen.
- On mobile, recording uses mic-only capture; system audio was unavailable on iOS and Android anyway and now no longer pretends otherwise.
Fixed
- Signup and login inputs no longer trigger iOS zoom on focus, every input hits the 16px minimum.
- The keyboard no longer hides the send button in the AI chat panel on mobile.
- Landing-page layout glitches on Safari have been resolved.
v0.8
Notifications, speed, and hardening
April 23, 2026Two weeks out from launch, we turned our attention to the quieter parts of a SaaS product: how fast the landing page feels on a cold first visit, how we reach users between their meetings, and how we keep their data out of the wrong hands.
Added
- SMTP-based email notification system, transcription completion, analysis completion, and processing failure templates.
- Per-category notification preferences with RFC 8058 one-click unsubscribe headers.
- Admin-only broadcast channel for product updates, changelog digests, and marketing announcements.
- Welcome email on signup with onboarding next steps.
- Robust token-bucket rate limiting across eight user-facing API endpoints.
- Third-party authentication token encryption at rest.
- Payment webhook deduplication via a processed-events ledger.
- Prompt-injection defense wrapping user transcript content in explicit untrusted-content delimiters before it reaches the model.
Improved
- Landing-page LCP improved by roughly four to five seconds, the logo was re-encoded to under five percent of its original weight, background images now route through an image optimizer, and the canvas waveform was replaced with lightweight CSS bars.
- Meeting-analysis prompts were tuned and internal token ceilings were removed so long meetings get the summaries they deserve.
- Background polling is paused in hidden tabs to save battery and API quota.
- API ownership checks added to nine user-facing routes as defense in depth on top of database-level security.
Fixed
- Cron secret moved out of the URL query string and into a request header so it no longer ends up in access logs.
- A full RLS policy audit closed residual permissive policies that had accumulated during the build.
v0.7
Plans, payments, and folders
April 19, 2026Real commerce meets real organization. This release introduced the pricing system that will power v1.0, the payment integration behind it, and the folder hierarchy users needed long before they admitted it.
Added
- Three pricing tiers: a $1 Paid Trial for 45 minutes, Basic at $9.99/mo for 600 minutes, and Pro at $17.99/mo for 1,200 minutes, Basic and Pro also available annual with four months free.
- Top-up packages that never expire, $1.99 for 60 minutes, $4.99 for 300 minutes, and $11.99 for 900 minutes.
- Full Checkout and Customer Portal integration, with per-user quota enforcement and a complete usage ledger.
- Meeting folders with drag-and-drop hierarchy and realtime sync across tabs and devices.
- Auto-filing, new recordings land in whichever folder was active when the meeting started.
- An Unfiled virtual bucket for everything that hasn't been categorized yet.
- Per-plan re-analysis limits, one on Trial and Basic, two on Pro, and Ask Loka unlocked on Pro.
Improved
- Usage-limit-exceeded flows now surface an in-app upgrade dialog instead of silently blocking the user.
- The usage ledger was hardened against double-refunds when a meeting is deleted mid-processing.
Removed
- Legacy per-month usage controls, retired in favor of the new plan + top-up system.
v0.6
Dynamic batching & the public waitlist
April 12 – 17, 2026Growing interest forced us to think about cost and capture at the same time, which produced the public waitlist landing page and an admin-controlled batching mode that trades a little latency for a dramatically lower transcription bill.
Added
- Animated waitlist landing page with an orb background and optimized hero media, positioned as the public entry point ahead of launch.
- Dynamic batching for transcription, a small wait in exchange for a materially lower cost on longer meetings.
- Pre-capture transcription mode selection, so behavior is decided before a meeting starts instead of negotiated afterward.
Improved
- Transcription fallback chain: primary model first, secondary automatically for languages or edge cases, batch fallback if both time out.
- Stuck-meeting guards and client-side polling windows were widened for dynamic-batching meetings to match our transcription provider's published SLA ceiling.
v0.5
Loka becomes Loka
April 10, 2026The product had been shipping under a placeholder name. This release cemented the brand we'll carry into launch, Loka, at lokanote.com, and tightened the transcription models underneath at the same time.
Changed
- Complete rebrand from Knot to Loka across the UI, marketing copy, email templates, and documentation.
- Domain migration to lokanote.com with updated metadata and open graph assets.
- Logo, favicon, and brand imagery re-cut for the new identity.
Improved
- Transcription model defaults upgraded with an automatic fallback.
- Language-specific routing added so short Thai and Burmese clips always use BatchRecognize, which produced materially better results than the sync path.
v0.4
Onboarding & UX polish
April 9, 2026A four-batch audit pass through the entire product, pipeline correctness, API ownership, UI robustness, and UX polish, each surfaced a list of small sharp edges that we spent the day rounding off.
Added
- A six-step onboarding tutorial with a spotlight overlay for first-time users.
- A preloader that tracks onboarding state across sessions so the tutorial never replays.
- A live mic-level probe and meter on the recording setup screen.
- A feature request backlog and a bug report system with up-to-four screenshot uploads.
- Deletion confirmations and stricter form validation across the app (advanced validation libraries).
Fixed
- API ownership check regressions surfaced by the audit were closed.
- Dialog focus traps, empty-state flicker, and a handful of other UI robustness issues identified during the batch pass.
v0.3
Real-time transcription & AI chat
March 27 – April 4, 2026The biggest architectural shift of the build. We dropped the meeting-bot model entirely and replaced it with direct in-browser recording, real-time transcription, and a conversational AI panel that turns every meeting into a queryable document. This is where Loka started to feel like its own product.
Added
- Direct browser recording with resumable chunked uploads and integrity verification at every step.
- Real-time AI transcription with an automatic fallback for edge cases.
- Audio device selection with microphone input plus optional system audio capture where the browser supports it.
- Multi-language transcription with automatic detection, eleven languages supported at v1.0.
- Manual transcript upload and paste flows for meetings captured off-platform.
- An AI chat panel that lets you ask questions of a finished meeting and get transcript-grounded answers.
- Meeting templates driven by a Database-backed schema with strict data validation on every analysis output.
- Audio playback with variable playback rate and timeline scrubbing synced to the transcript.
Removed
- The entire meeting bot infrastructure was retired in favor of direct capture. Simpler, cheaper, and no uninvited participants.
v0.2
Meeting analysis & dashboard core
March 19 – 22, 2026The first version that felt like a product. Meetings were uploaded, transcribed, and returned as structured summaries, the first cut of the end-to-end loop our users still rely on today.
Added
- AI-generated meeting summaries, action items, and key decisions.
- A meeting timeline view with playback synchronized to the transcript.
- Per-account plan and usage visualization, colorized badges and percentage meters at a glance.
Removed
- Experimental data scraping and ad reporting scaffolding that had been prototyped earlier, out of scope for a focused v1.0.
v0.1
First commit
March 17, 2026The foundation. A modern full-stack web architecture with per-user row-level security wired through every table from the first migration. Everything else was built on top of this.
Added
- Full-stack architecture with server-side rendering and component library.
- Authentication with email/password signup and login.
- Per-user row-level security policies across every user-facing table.
- Initial recording pipeline and speech-to-text integration groundwork.
- Multi-tenant data model with user_id scoping enforced on every write.
Want updates in your inbox?
Every future release gets a short digest sent to anyone opted in to product updates, unsubscribe any time, one click.