Help and docsLiving documentation
The in-app help centres are the source of truth.
Every product ships its own manual inside the app, and every changelog entry links to the help article it changed. This page is the map: how each product’s help is organised, how to get started, and where to look when something looks wrong.
LevelUp help
Help & Learning inside the app: 14 categories, 60 articles, 5 guided tours, in-app search, and an AI answer box. The Help Center is also the assistant’s knowledge base, so “how do I run a weekly review?” gets the page and the keystroke.
| Category | What it covers | Articles |
|---|---|---|
| Getting Started | First steps, onboarding, and orientation | 6 |
| Tasks & Projects | Manage tasks, projects, and deadlines | 4 |
| Calendar & Events | Scheduling, views, and time-blocking | 1 |
| Notes & Journal | Capture, organise, and review your thoughts | 3 |
| Contacts & CRM | Manage relationships and enrichment | 1 |
| Settings & Integrations | Connect accounts, configure preferences | 4 |
| AI & Automation | AI keys, chat assistant, AI-powered features | 5 |
| Appearance & Themes | Theme presets, custom colours, scheduled profiles | 1 |
| Reports & Widgets | Build custom dashboards and schedule emails | 1 |
| Command Center | Cross-tool PM dashboard, Standup, Programs, briefings | 7 |
| Sales Pipeline | Opportunities, stages, forecast, pipeline reports | 4 |
| Ideas & Thinking | Ideas lifecycle, Mind Maps, Focus, GTD, Knowledge Graph | 7 |
| Your Workspace | Show only what you use: pages, sidebars, sections, and team layouts | 7 |
| Routines & Mastery | Daily, weekly and monthly rhythms that make the app pay off | 9 |
Quick start
- Open My Day to see today’s focus.
- Press Ctrl/⌘ N to quick-capture a task from anywhere, or C for the capture modal: task, note, journal or idea. Use #tag inline and a ! prefix for high priority.
- Use Ctrl/⌘ K to search across everything and run actions.
- Press ? to open the Help Center and the shortcuts overlay anytime.
Your first week: five habits
- Capture immediately, sort later. Do not stop to pick a project or a due date. Tidy up on the Process page.
- Start the day on Planner. Decide once what today looks like.
- Link things as you go. The links are what let the AI answer with real context.
- Let it read your own work. Ctrl/⌘ K searches everything; the ✨ menu answers from your own notes and cites which ones it used.
- Trim it to fit. Hide what is noise: Customize this page, or Settings → Workspace. After the first week, turn one thing back on.
Keyboard shortcuts
| Shortcut | Action |
|---|---|
| Ctrl/⌘ + K | Command palette: search and actions |
| Ctrl/⌘ + J | AI chat assistant |
| Ctrl/⌘ + N | New task (quick add) |
| Ctrl/⌘ + / | Ask LevelUp |
| ? | Shortcuts overlay |
| Esc | Close any modal, drawer or overlay |
| Ctrl/⌘ + S / B / I / U | Save, bold, italic, underline in editors |
| ← / → | Previous / next image in the lightbox |
Velocity help
Help is built in: hover help on every rail item, an in-app Help Center with an operator’s manual and Ask AI over every article, guided tours, and a sample-data workspace for onboarding. The manual holds the full routines, repeatable processes, every page explained, the data model and troubleshooting.
The data model, in one paragraph
Velocity separates the who from the deal. Accounts and Contacts are durable records that persist across many deals. The Opportunity is the only thing that travels the pipeline and closes Won or Lost. Won turns the account into a Customer with a health score, renewal stage and QBR history. Leads are inbound or engaged people awaiting qualification; converting one creates the account, contact and opportunity in one step.
The one control to learn first
Every autonomous feature is on the Autonomy Center with the same dial: Off, Approve, Auto. Start every autopilot in Approve. It does the work and stops, so you can read exactly what it would have sent. Promote a dial to Auto only when you find yourself approving everything unchanged. Dials are admin-only.
How teams run it
| Cadence | What to do |
|---|---|
| Daily, about ten minutes | Work the Home attention panel top to bottom, read the engine events, prep today’s meetings, close the day with touches logged and drafts approved. |
| Weekly, about forty-five minutes | Engine Performance and each campaign’s step performance, campaign health and rejections, sequences, conversations, data health, pipeline and forecast, leads older than two days, inbound, dials and saved reports. |
| Monthly | Budgets and credit cycles, recalculate scoring, review the ICP, prune, promote one dial from Approve to Auto, deliverability, prompts, customer health, hygiene, access, backup. |
Roles
Super admin, admin, manager and rep. Autonomy dials, sending accounts, keys and budgets are admin-only. Every assistant action is in the audit log.
Publish help
Inside the app: the owner’s guide (the whole machine on one page, a fifteen-minute read), a Help Center with role-based paths and searchable answers by category, hover help on every rail entry and control, Elsie’s short tours, and a morning digest on the days it matters. The assistant knows all of it: ask “how do I…” or “why is … held?” and it gives the page and the reason.
Set a workspace up, once
- Give it a brain. Publish Admin → API keys: an AI provider key, the matching default model, and a live web search key (Tavily or Serper) so it can source claims.
- Connect where it publishes. Settings → Connections: social accounts, a mailbox, your website (WordPress or HTML export), analytics.
- Teach it your voice. Brand → Tone & motifs: the seven Motifs, topics, guardrails, brand kit.
- Tell it what you actually do. Brand context: differentiators, products, brand documents. Nothing here is AI-generated.
- Set the clock. Settings → Schedule: timezone and posting slots.
- Choose your gates. Settings → People: require approval; queue on approval.
- Set the autonomy dials. Settings → Automation. Start low.
- Wire up measurement. Publish Admin → Analytics: Search Console and GA4.
- Invite the team. Settings → People.
Repeatable processes
| Process | Path |
|---|---|
| An article, end to end | Ideas (approve) → the sweep drafts it → auto-review fills SEO, renders images, sources claims → the Inbox shows anything it could not fix → you answer, verify or approve → final approval → the publish day → Publish shows the live link → social variants follow |
| A held article you want out anyway (admin) | Inbox or Review → the held-article card → an optional reason → Advance anyway. Who, when and why are recorded. |
| A held image | Inbox → Images that need your eye → Approve, or pick or upload instead. Auto-review already spent up to three renders on it. |
| A claim with no source | Inbox → Claims with no source → verify with a URL that genuinely supports the sentence, or drop the claim. |
| Publishing without WordPress | Publish → Download HTML → add it to the site → Mark as published with the live URL. Upload the Open Graph image to the site for social previews. |
| A new competitor | Research → Intel → add by handle or keyword → outliers appear → Bookmark → Make it an idea, or ask the assistant about the video. |
| Changing the cadence | Settings → Automation for the weekly article target, the publish day and social posts per week; Settings → Schedule for the slots. |
| Running the loop by hand | Settings → Automation → Run cycle now, then watch Drafts and the Inbox. |
When something looks wrong
| Symptom | Cause and fix |
|---|---|
| Output reads generic or mentions mock | No working AI key for this workspace. Publish Admin → API keys, and match the model. |
| Nothing is being drafted | The Approved column is empty, the weekly target or daily budget is reached, drafting is on manual, or global pause is on. Settings → Automation says which. |
| An article is held with nothing to act on | The Inbox card names the failing check. A flagged claim needs a live-search key; give it one sweep, then verify or remove the sentence yourself. |
| At final approval but never publishes | Not the publish day yet, or no WordPress (Download HTML and Mark as published), or publishing is not on auto. |
| Social posts never send | No slots or timezone, a post awaiting approval, approved but never queued, or a broken account (red on the provider’s own verdict). |
| Measure shows dashes | Analytics not connected, or the Search Console property is missing the service account. Publish Admin → Analytics runs the live probe that says which. |
| A button does nothing after an update | A tab held open across a deployment. Reload. |
Producer help
Producer’s help is the app itself: Setup checks the machine and fixes what it can with one click, the Claude panel shows every run as it happens, and one button sends a failed check back to Claude to fix. A built-in self-test exercises the whole chain on a sample recording.
Setup checklist
- Claude Code found and signed in, using your own Anthropic subscription. Install and sign-in actions open a terminal and walk you through it.
- FFmpeg and the HyperFrames engine, both bundled.
- The render browser, a one-time download of about 160 MB.
- The production skills Claude needs, installed into your Claude skills folder.
- A transcription engine: whisper.cpp installed automatically (about 490 MB for the model), or Python Whisper detected.
- The narration voice model, about 330 MB on first use, cached locally.
- A projects folder.
Requirements
Windows x64, per-user install, no admin rights. A Claude Pro, Max or Team subscription with Claude Code installed. About 1.6 GB on disk for the app plus the models above. Rendering is CPU-bound: a three-minute video takes about four minutes on a fast laptop.
Troubleshooting
| Symptom | What to do |
|---|---|
| Windows SmartScreen warns on download or first launch | Version 0.1.0 is not yet code-signed. The warning is expected; a signed build is on the changelog’s list. |
| The check fails after a build | One button sends the log back to Claude to fix. It runs lint and the check again until they pass. |
| A line was edited after recording | Narration changes are script edits. The app tells you which lines to re-voice; re-voice only those. |
| Rebuilding from the script | The app warns first, because manual Studio edits would be overwritten. |
| A figure in the script is wrong | Check it against the frames in the Script step and fix the line by hand, or re-draft. Every number must be visible in the footage. |
Docs are living
Every entry links to an article
The in-app help is the source of truth
Removals are documented too
Something the docs did not answer?
Post it as a problem. Missing help is a problem too, and it gets the same ten-day decision.
