Astra MotionsDocumentation
Astra Motions is a platform for teams who ship. It combines an AI assistant that works with your real accounts (Aura), deployable AI employees (agents), and a document-reading scanner (Invoice OCR) — all under one credit wallet.
What you can do
- Chat with Aura — draft, summarize, extract, and answer questions. Connect Gmail, Google Calendar, and Notion so Aura can read, and write with your approval.
- Build AI agents — describe an employee in plain language, attach reference documents, and deploy it as a public chat widget on any website.
- Scan invoices — upload an invoice and get the vendor, date, totals, and line items extracted into an editable form.
- Track everything — every request draws from your credit wallet with a full ledger, and the Usage page shows your balance and activity.
Quick start
From sign-up to your first Aura message in about a minute.
- Sign in with Google — the Login button in the top-right of the landing page.
- Complete onboarding — choose a username and add your company name (phone is optional). This happens once; you land in Aura afterwards.
- Type anything in Aura — try "draft an email to my team about next week" or "summarize this text: …".
- Check your credits — open the Usage item in the sidebar to see your wallet balance and monthly allowance.
Every request is billed against your wallet at a flat per-request rate for the model used (see Models). If a request costs more than estimated, the extra is charged after the fact; if it costs less, the difference is refunded.
Aura chat
Aura is your personal AI assistant inside the workspace. It keeps the conversation in a session (recent chats are listed in the sidebar) and answers in plain text.
Everyday requests
- Draft an email, a message, or a document
- Summarize pasted text into overview + bullet points
- Extract dates, names, numbers, and key terms
- One-off questions and explanations
Connected apps
Once you link an account (see Connected apps), Aura can work with real data:
- Gmail — read and search your inbox, and draft emails.
- Google Calendar — check upcoming events and availability, and prepare event creations, updates, and cancellations.
- Notion — search pages and databases, read content, and prepare pages, rows, and updates.
AI agents
Agents are AI employees you can build once and reuse — internal assistants that answer from your documents, or public widgets anyone can talk to.
Create from a description
- Open the New Agent tab and describe the employee in plain language — for example "an agent that answers customer questions using our policies".
- Astra Motions drafts the configuration: name, role, instructions, tone, and model. You can edit any field or start from one of the template cards.
- Save to create the agent, or go straight to its workspace.
Reference documents
Attach PDF, DOCX, or TXT files (up to 10 MB each) from the agent workspace. The knowledge-base pipeline extracts the text and the agent answers using it, citing the source documents. Uploaded files live in your organization's private knowledge base.
Manage
- My Agents — list all agents in your organization and open their workspaces.
- Workspace — chat with the agent to refine its behavior, review its instructions, and manage its files and deployment.
Deploy & embed
Pro plans can publish an agent publicly. Visitors get a chat-widget bubble on your site that talks to the agent — answered from its instructions and reference documents.
Publish
- Open the agent workspace and toggle Public access in the Deploy panel.
- The agent receives a public id like
pk_agent_xxxxxxxxxxxx. Turning it off unpublishes instantly.
Embed on any site
Add one line before the closing tag of your page:
<script src="https://YOUR-SITE/embed/pk_agent_xxxxxxxxxxxx.js" async></script>The widget renders a fixed bottom-right chat bubble, posts to /api/public/chat, and needs no other dependencies. The script is cached for 5 minutes, so rotating the public id also invalidates old caches.
Invoice OCR
Upload an invoice image or PDF and the scanner extracts the structured fields into an editable form you can review and correct.
- Vendor, date, and totals
- Line items and amounts
- Payment and tax fields where present
Start from the landing page upload bar or the dedicated /invoice-ocr page. Files are limited to 10 MB and the feature requires a signed-in account.
Credits & billing
Every AI request — Aura, agents, and invoice scanning — draws from your credit wallet at a flat per-request rate. There is no token metering on your side: the model's listed credit cost is what a request costs.
- Reserve → settle — the estimated cost is reserved before the call; afterwards the difference is charged or refunded automatically.
- Full ledger — every movement (purchase, usage, refund, bonus) is recorded and visible on the Usage page.
- Free — starts with 1,000 credits.
- Pro — 3,000 credits every month, refreshed on your billing-cycle date.
- Business — 25,000 pooled credits each month for the whole organization, shared by all seats.
- Overage packs — top up with 5,000 credits for $39 whenever you need more.
Plans & pricing
| Plan | Price | Includes |
|---|---|---|
| Free | $0 / month | 1,000 starting credits · Gemma 4 (cloud) + GPT-OSS free models · Aura chat · AI agents · Invoice OCR |
| Pro | $29 / month | Everything in Free · 3,000 credits monthly · 10 premium models via OpenRouter · public agent embeds · Gmail / Calendar / Notion connections |
| Business | Custom | Everything in Pro · 10 seats · 25,000 pooled credits monthly · full model catalog · overage credit packs · priority support |
See the live pricing page for the current copy.
Models
Choose the model for your agents and Aura in Settings. Free accounts see the two free models; Pro unlocks the full catalog. Cost is credits per request.
| Model | Tier | Credits / request |
|---|---|---|
| Gemma 4 31B (cloud) | Free | 1 |
| GPT-OSS 120B (free) | Free | 1 |
| DeepSeek V4 Flash | Pro | 1 |
| Gemini 2.5 Flash-Lite | Pro | 1 |
| Gemini 2.5 Flash | Pro | 2 |
| MiniMax M3 | Pro | 2 |
| GLM 5.2 (Z.ai) | Pro | 4 |
| OpenAI o3 | Pro | 6 |
| Gemini 2.5 Pro | Pro | 6 |
| Claude Sonnet 4.6 | Pro | 10 |
| Claude Opus 4.8 | Pro | 18 |
| GPT-5.5 | Pro | 20 |
Settings & profile
Open the account menu (top-right) to reach:
- Settings — theme (light, dark, or system) and the default AI model used for Aura chat. Your model choice follows you across devices.
- Edit profile — update your name, username, phone, and company name. Aura greets you by name and uses your company as context.
Connected apps
Aura's connectors let it work with your real accounts. Open the apps panel from Aura to connect:
- Gmail — read, search, and draft email.
- Google Calendar — read events and availability.
- Notion — read pages and databases.
Connections use standard OAuth through Composio. You can disconnect any app at any time — Aura immediately loses access to it.
Limits & safety
Abuse-prevention limits protect the platform (and your credits) from runaway traffic:
| Surface | Limit |
|---|---|
| Aura chat | 30 requests / min / user |
| Agent generation | 20 requests / min / IP |
| Agent refinement | 30 requests / min / user |
| Invoice OCR | 10 uploads / min / user |
| Public embed chat | 20 requests / min / IP |
| File uploads (documents, invoices) | 10 MB per file |
| Chat message / agent prompt length | 4–16 KB |
When a limit is hit you get a clean 429 response (or a friendly error in the UI) rather than a silent failure, and oversized uploads are rejected before they're processed.