EN
UI Features
A tour of Memo's user interface and its capabilities.
Navigation
The app uses a persistent sidebar layout. On mobile, the sidebar collapses into a hamburger menu. Navigation items:
- Chat — Main conversational interface (agent mode and web-search toggles now live directly in the chat top bar, alongside the existing model switcher)
- Model Store — Browse and download models (Discover tab with hardware-aware setup recommendations)
- WhatsApp — Bridge inbox and AI-assisted replies
- Calendar — Events and reminders
- Routines — Scheduled automations
- Developer — Developer API Gateway (Anthropic-compatible local endpoint)
- Swarm — Memo Swarm (beta, requires Beta Features enabled)
- Settings — Configure Memo
Chat Screen
The main interface for conversations with Memo.
| Element | Description |
|---|---|
| Sidebar | Lists all conversation sessions. Tap to switch, swipe to delete. New chat button at the top. |
| Top Bar | Shows current session name, an agent mode toggle and a web-search toggle side by side, orchestra mode toggle, and status indicator (connected, processing, error). |
| Message List | Scrollable chat history with user and assistant messages. Messages support Markdown rendering and code blocks with syntax highlighting. |
| Input Area | Multi-line text field with send button. Supports Shift+Enter for newlines and Enter to send. |
| STT Button | Speech-to-text input. Press and hold to record, release to transcribe and send. |
Agent Screen
A dedicated workspace for running autonomous agents:
- Tool Panel: Shows available tools the agent can use, with toggle switches to enable/disable per tool.
- Output Console: Real-time stream of agent actions, tool invocations, and results.
- Progress Bar: Visual indicator of multi-step task completion.
- Interrupt Button: Manually stop a running agent at any time.
Model Store
Browse, download, and manage LLM models, with a Discover tab for finding new ones:
- Catalog View: Grid of available models with name, size, parameter count, and description.
- Search & Filter: Discover's filters (Tools/Vision/Code/Embedding/Size) combine with OR instead of AND, grouped into multi-select dropdowns with an "N filters active · clear" indicator.
- Download Queue: Multiple concurrent downloads (no longer blocked one-at-a-time), with combined progress shown in the engine status bar.
- Installed Models: List of locally available models with quick-switch capability.
- Capability Badges: Tool-calling and code-capability badges are computed from the model's actual chat template and tags, not a hardcoded list of "known" families.
- Context Size Guard: Local model context size is read from the model file's real maximum and the slider can no longer be pushed past it — a free-text field used to allow crashing the engine with an unrealistic value.
- Accurate Branding: The logo shown for a model reflects who actually made it, even for a requantized/repackaged upload.
Settings Dialog
Organized into tabbed sections:
| Tab | Contents |
|---|---|
| General | Language, theme, startup behavior, data directory, Minimal Mode (with independently re-enable-able persona/disclosures/proactive-learning switches) |
| Identity | User name, assistant name, conversation style, system role |
| Memory | Embedding model, persist directory, auto-embed, Top-K, similarity thresholds |
| Model Parameters | Temperature, repeat penalty, Top-P, GPU layers, context size |
| API Providers | Provider selection (9 providers), API keys, base URLs, model overrides |
| Orchestra | Conductor model, agent roles, max parallel agents, timeout |
| Cloud Sync | Toggle, passphrase, sync interval, last sync timestamp, force sync |
| Remote Access | Toggle, API port, Web UI port, network binding, access token |
| Beta Features | Master toggle unlocking Tailscale and Memo Swarm |
| Stats | Usage Stats — KPI cards (requests, tokens, avg tokens/sec, most-used model), 30-day token usage chart, per-model breakdown |
| Report Bug | Prefills a GitHub issue, optional last-10-error attachment, opens in your browser for you to review and submit |
Setup Wizard
First-run experience that guides new users through:
- Welcome screen with privacy notice
- Identity configuration
- Hardware-aware model recommendation (reads RAM/GPU, suggests a matching chat + memory model pair, one button starts both downloading while setup continues)
- Provider setup (optional)
- Cloud sync setup (optional)
- Completion and first chat
Theme System
Light and dark themes with system-following auto mode:
- Light: Clean white background with indigo accents
- Dark: Dark gray background with softer indigo accents
- Auto: Follows OS-level light/dark preference
- Custom accent color selection coming in a future release