One workspace, two ways to work
Talk it through, or hand it off.
Mystical is a chat that remembers what you told it and an agent that goes and does the work itself — switching between them is one click, and both draw on the same memory, connectors, and files.
Chat
A conversation that remembers where you left off
Ask a question, drop in a file, or start writing — Chat carries memory across sessions, searches the web when it needs current information, and hands back real deliverables instead of only prose.
- Memory that carries facts and preferences across conversations
- Web search grounded in current information, not a training cutoff
- Documents, spreadsheets, slide decks, and small interactive apps it writes and you open
- Code that actually runs in a sandbox, for real answers instead of guesses
- Thinking modes for the questions worth slowing down for
Turn last quarter's numbers into a one-pager I can send the team.
Q3 summary.docx
Document · 1 page
Cospace
Give it a goal. It plans, works, and reports back.
Cospace is the same interface pointed at longer, autonomous work: describe what you need, and it breaks the goal into steps, works through them one at a time over however long it takes, and keeps going even after you've moved on.
- A visible plan up front, then a live step-by-step work block while it runs
- Pauses for your approval before any sensitive connector action
- Produces the same kinds of artifacts Chat does
- Runs survive a closed tab — the work keeps going on our side
- Scheduled tasks turn any run into one that repeats and emails you the result
Read the last four weeks of support tickets
Drafting the summary document
Wants to send the summary to #support on Slack
Underneath both
The parts that make it feel like one workspace
Six pieces shared by Chat and Cospace, so a fact you taught it once or a connector you approved once works everywhere, not just in the product you happened to be using.
Connectors
Gmail, Google Drive, and Calendar today, plus any MCP server you point Mystical at. Every tool call it wants to make is approved — once for good, or every single time.
Memory
Durable facts about you and your work, carried across conversations and shared across whichever assistant you're using.
Projects
Files, instructions, and shared members in one place, with every chat inside drawing on the same context.
Library
Every document, sheet, deck, and mini-app either product has produced, kept in one place you can find again.
Skills
Reusable instructions — written by you, or by the assistant when you ask it to remember how you like something done.
Scheduled tasks
Turn a Cospace run into something that repeats — every morning, every Monday — and get the result by email instead of checking back.
For developers
One protocol, two directions
The developer platform speaks the Model Context Protocol both ways: publish a server so it shows up as a connector inside Mystical, or take a URL that points any MCP-aware assistant at your own Mystical memory.
Publish a connector
Register an MCP server and, once reviewed, it's usable by anyone in Chat or Cospace.
Connect your assistant
Point Claude, ChatGPT, or anything else that speaks MCP at your own Mystical memory.
Pricing
Start free. Upgrade when you need more.
Chat and Cospace draw on one allowance that refills every five hours, so a heavy afternoon doesn't cost you the rest of the day.
Free
Try it out
$0
- 64K context window
- 64K tokens every 5 hours
- 5 min of code execution every 5 hours
- DeepSeek V4.1 Flash
- No thinking mode
- Chat only, no Cospace
Go
For everyday chat
$7/ month
- 128K context window
- 128K tokens every 5 hours
- 15 min of code execution every 5 hours
- DeepSeek V4.1 Flash, GLM 5.3 and DeepSeek V4 Pro
- Thinking modes
- Chat only, no Cospace
Plus
PopularChat and Cospace
$20/ month
- 256K context window
- 256K tokens every 5 hours, shared by Chat and Cospace
- 1 hr of code execution every 5 hours
- DeepSeek V4.1 Flash, GLM 5.3 and DeepSeek V4 Pro
- Thinking modes
- Cospace for long, autonomous tasks
Max
For heavy, long-running work
$50/ month
- 256K context window
- 750K tokens every 5 hours, shared by Chat and Cospace
- 3 hr of code execution every 5 hours
- DeepSeek V4.1 Flash, GLM 5.3 and DeepSeek V4 Pro
- Thinking modes
- Cospace for long, autonomous tasks