LokalBot

Product Introduction

1. Definition: LokalBot is a free, open-source desktop application for macOS, categorized as a local-first work memory and meeting intelligence tool. It functions as a personal knowledge base that operates primarily offline.
2. Core Value Proposition: It exists to solve the problem of digital information fragmentation by providing a unified, private search engine for your spoken and on-screen work. Its primary value is enabling users to instantly recall decisions, action items, and reference material from meeting transcripts and saved screen text without relying on cloud services or manual note-taking.

Main Features

1. Local Meeting Transcription & Search: LokalBot records your microphone and system audio on separate tracks to capture both sides of a conversation without requiring a bot to join the call. It then transcribes the audio locally using on-device models (with options for custom models like Ollama). The resulting transcript is fully searchable, with results linking directly to the timestamp in the recording for context.
2. Opt-In Screen Text Capture & Recall: Users can optionally save visible text from their screen. This feature is highly configurable, allowing exclusions for specific apps, websites, and private windows. Captured text is indexed for search, enabling users to find a webpage or document based on content they remember, even if they've forgotten the title. Screenshots are an additional, separate opt-in layer.
3. Local AI-Powered Productivity Tools: The app integrates several on-device AI functionalities. These include local dictation (triggered by holding ⌥+Space), local autocomplete for text input, and local "routines" that can automatically generate follow-up email drafts or work logs from meeting content, saving them to a user-designated folder for review.
4. Consolidated Daily Summary & Action Tracking: The "Today" view aggregates key information from your day, providing a summary of meetings and surfacing open action items without requiring you to re-read entire transcripts, effectively serving as an automated daily stand-up report.

Problems Solved

1. Pain Point: The "Where did I see that?" problem—the inability to quickly find information discussed in a meeting or viewed on screen hours or days prior, leading to wasted time and lost context.
2. Target Audience: Knowledge workers, remote team members, project managers, researchers, and consultants who participate in numerous calls and consume vast amounts of digital information daily. It is particularly relevant for privacy-conscious professionals in fields like legal, healthcare, and finance, as well as developers and open-source advocates.
3. Use Cases: A project manager needs to find the exact moment a deadline was agreed upon in a weekly sync. A researcher wants to relocate an academic paper they skimmed earlier based on a unique term. A consultant needs to compile action items from three client calls to draft a summary email. An individual preparing for a performance review wants to retrace their work and accomplishments from the past month.

Unique Advantages

1. Differentiation: Unlike cloud-based meeting assistants (e.g., Otter.ai, Fireflies.ai) or universal search tools, LokalBot’s default and primary mode is fully local. It does not send audio, transcripts, or screen data to external servers for core functionality, addressing major privacy and data sovereignty concerns. Its integration of screen text search with meeting search is also a unique hybrid approach.
2. Key Innovation: Its "enshittification-proof" local-first architecture is its core innovation. By processing sensitive data (audio recordings, screen content) exclusively on the user's Mac with optional, user-approved external model use, it inverts the standard SaaS model. The integration of separate, permission-gated modules for agents and tools provides extensibility without compromising the default privacy stance.

Frequently Asked Questions (FAQ)

1. Is LokalBot really free and private? Yes, LokalBot is completely free, open-source (GPL v3 licensed), and processes meeting recordings, transcriptions, and screen text locally on your Mac by default. No account or subscription is required, and your data does not leave your computer for core features.
2. How does LokalBot record meetings without a bot? It uses system-level audio capture to record your microphone and the audio output of your meeting application (e.g., Zoom, Teams) on separate tracks. This method does not require the app to join the call as a participant, but it is the user's responsibility to inform and obtain consent from all meeting participants before recording.
3. What screen data does LokalBot capture, and can I control it? By default, new installations use "activity-only tracking." Visible screen text capture is opt-in and configurable. You can exclude specific applications, websites, and private browser windows. Screenshots are a separate, additional opt-in feature. Captured data is automatically deleted after 14 days unless you explicitly save a moment.
4. Can I use my own AI models with LokalBot? Yes. While it includes built-in local models, LokalBot supports using alternative models for transcription, summarization, and autocomplete. You can configure it to use Ollama, Apple Intelligence, or connect to an OpenAI-compatible API, giving you control over the performance and privacy trade-offs for each function.
5. What are the system requirements for LokalBot? LokalBot requires a Mac with Apple Silicon (M1 chip or later) running macOS 15 (Sequoia) or later. The initial setup includes downloading necessary AI models, after which the core application can function entirely offline.