SynergyAI Workspace
One workspace connected to everything you already use
Connect Gmail, Slack, Notion, Drive and ~49 more, then describe a routine once — it runs on schedule and reports back.
Connected apps
Connect 53 apps, or your own MCP servers
Connect a service once and your agents and automations can read from it and act in it.
- You choose what agents touchPick which apps each automation stage may use.
- Connect the way you needOne-click OAuth, your own OAuth client, an API key, a cloud access key and secret, or a bot token.
- Inbox, calendar and drivesGmail and Slack feed the Unified Inbox; Google Calendar and your cloud drives open in the workspace.
- Anything else, by URLPaste a Model Context Protocol (MCP) server URL and your agents can call its tools.
In detail
All 53 services, by category
- Communication: Gmail, Outlook, Slack, Microsoft Teams, Discord, Zoom, Google Meet, Telegram, WhatsApp
- Productivity and knowledge: Notion, Google Docs, Google Sheets, Google Slides, Google Tasks, Google Calendar, SharePoint, Confluence, Calendly
- Files and developer: Google Drive, OneDrive, Dropbox, Box, AWS S3, GitHub, Supabase, Hostinger, AWS, Vercel
- Sales, marketing and support: HubSpot, Salesforce, Mailchimp, Resend, Intercom, Zendesk, Google Ads, Meta Ads
- Payments and finance: Stripe, PayPal, Razorpay, Shopify, QuickBooks
- Social: LinkedIn, Facebook, Instagram, Reddit, YouTube
- Design, web and analytics: Figma, Canva, Webflow, Google Analytics
- Research and data: Tavily, Firecrawl, CoinMarketCap
Connection health and MCP servers
- Each connection shows its health: Connected, Needs reauth, Setup incomplete or Not connected
- Refresh an MCP server's tool list, rename it, change its auth, or switch it off without deleting it
- MCP servers connect over Streamable HTTP, up to 256 KB per request
Automations
Describe a job once and it runs on schedule
Write it in plain English. It runs on a schedule in your timezone, or whenever another app calls its trigger URL, as one step or several in a row.
- Describe itSay what should happen and when, down to several times a day.
- Check the planSee its stages, the apps it needs, any setup still missing, and what it cannot do.
- Choose approvalApprove each result as a draft, or let it deliver automatically.
- Get the resultIn the workspace, by email, on Telegram or in Slack.
In detail
Chains, controls and templates
- Each stage has its own instruction, agent, allowed apps and fire-at time
- A stage can split into parallel workers; generated plans have up to 6 stages
- Stage conditions: always, skip if empty, only if contains, skip if contains
- If a stage fails: continue, stop, retry or ask you
- Run now, pause, pause for a set time, or pause everything with vacation mode
- One run at a time per automation, so retries cannot stack runs
- Run history shows each run's status, duration, cost, output and any error
- Use the template library, or save up to 24 templates of your own
- Attach a brand profile: voice, palette, taglines, do and don't lists, sample posts, audience and notes
Command Center
Hand off bigger work, and review the plan first
For work too big for one prompt and too specific for a template: send a brief, review the step-by-step plan, then let the agents run it.
- Edit or fork the planChange it before it runs, or fork a fresh copy.
- Answer its questionsWhen a mission hits a decision it cannot make for you, it pauses, then resumes where it stopped.
- Re-run one stepRetry a failed step without redoing the whole plan.
- Pick the modelChoose GPT, Claude, Gemini, Grok, DeepSeek, Qwen, Kimi or GLM models, or Auto.
- Work in many tabsUp to 20 open tabs, and recently closed ones can be restored.
- Attach, dictate, exportFiles up to 4 MB, voice dictation, and Markdown or plain-text export.
Unified Inbox
Gmail and Slack in one sorted list
Read Gmail email plus Slack direct messages and unread channel messages in one list, sorted into four priority tiers.
- Sort by fixed rulesUrgent, important, can wait or noise, decided by rules, not a model.
- See who owes whomNeeds reply lists what waits on you; Waiting lists what you are owed.
- Draft on requestPress Draft reply, edit it, then press Send. Sending is Gmail-only; copy a Slack draft out yourself.
- Write in your styleGreeting, sign-off, length, formality, warmth, emoji and a never-say list.
- Pin or mute sendersPinned senders stay important or above; muted ones clear on arrival.
- Clear in bulkClear, snooze or mark read many at once; one-click unsubscribe tells you when no unsubscribe link was found.
In detail
Views and sync
- Five views: Inbox, Needs reply, Waiting, Snoozed and Cleared
- Needs reply shows threads from 2 days old; Waiting from 3 days
- Background sync every 5 minutes, plus Sync now
- A "by a rule" badge marks automatic clears, and rule suggestions come from your own mail
- No-reply, daemon, list and notification senders are never replied to
- Steer a draft with your own instruction
Calendar and reminders
Book time and get reminded in plain words
Link Google Calendar, ask an agent to find or clear time, and set reminders just by typing them.
- See your calendarMonth grid, day agenda, multi-day events, and full event editing.
- Book time in plain words"Book three gym slots next week" or "clear my Friday afternoon".
- Set reminders three waysAt a set time, on a repeating schedule, or on an in-app event.
- Type /remind me"In 20 minutes", "every weekday" or "tomorrow", parsed in one go.
- Choose where they landWorkspace, Inbox, Email or Slack, as many as you tick.
- Approve suggested remindersAn hourly scan of your memories, psychological map, Mind chats, check-ins and inbox proposes birthdays and follow-ups; you accept or reject each.
In detail
Reminder details
- Snooze by 5m, 1h, tomorrow or 1w; dismiss with a reason, or mark it done
- Repeating reminders follow daylight saving in your time zone
- Birthday reminders arrive a day early by default
- The calendar command bar runs on Claude Haiku, with a live cost meter
Telegram delivers automation results, but not reminders.
Files
Your folders, versions and cloud drives together
Uploads, agent output and generated media live in project folders with version history, search and zip export, next to your connected drives.
- 5 GBstorage on Free
- 50 GBstorage on Pro
- 100 GBstorage on Studio
- 30 daysto restore from trash
- Browse your cloud drivesGoogle Drive, OneDrive, Dropbox and Box, browsable and writable in Files.
- Restore earlier versionsBring back any of a file's last 10 versions in one click.
- Search inside filesFull-text search across file names and contents.
In detail
Templates, limits and clean-up
- Start a project from a blank website, campaign kit or story bible template
- Each agent gets a home folder, and each run gets its own
- Open a project folder in Vibe Code; edits save back to it
- One storage pool covers uploads, generations and studio assets, with no per-file size cap
- Zip a folder (up to 20 MB and 400 files) or a whole media album
- AWS S3 connects, but cannot be browsed from Files
- Agent home folders can clear old working files after 7 to 3,650 days, or keep everything
- Preview a clean-up first; deliverables and starred files are always kept
Newsroom
Headlines from the topics you pick
Pick topic categories and your own keywords; headlines refresh in the background and open in an in-app reader.
- 32topic categories, up to 20 followed at once
- 30keywords of your own, each up to 80 characters
- Under 10 minold earns a story a Breaking badge
- 2 minbetween background refreshes
In detail
Keywords and freshness
- Headlines come from 173 RSS feeds
- Each keyword also runs its own Google News search, and non-English keywords work
- Seven base feeds keep keyword search broad: BBC, Wired, Vice, The Atlantic, Vox, Axios and NPR
- A story reported by several outlets shows once
- New badge under 30 minutes old; stories you have read on this device stop being flagged
- Refresh now, up to 3 times a minute
All 32 topic categories
Privacy, approval and cost stay in your hands
One account covers every studio — top up as you go, or choose a monthly Pro or Studio plan.
- Access tokens stay on the serverCalls to your connected apps are made server-side, never from your browser.
- Nothing posts without your consentAn automation cannot post to an app until you turn publish on.
- Replies are never sent for youA draft goes out only when you press Send.
- An estimate before a mission runsCommand Center shows a per-step and total range, labelled "Guidance, not a bill".
- A budget per automationSet a cost cap per run and a total budget.
- A failed draft is refundedIf no reply draft comes back, the charge is refunded.
Frequently asked questions
Can I connect Gmail and Slack to SynergyAI?
Yes. Both connect from the Inbox, and the Unified Inbox pulls Gmail email plus Slack direct messages and unread channel messages into one triaged list. Tokens stay on the server, and you can disconnect at any time.
Does SynergyAI reply to my emails automatically?
No. A reply is drafted only when you press Draft reply on a message. You read and edit it in the inbox, and it goes out only when you press Send. If no draft comes back, the charge is refunded. Sending is Gmail-only; copy a Slack draft out yourself.
How does the AI inbox decide what is urgent?
With fixed rules, not a model. Critical phrases score urgent from anyone. Softer urgency words score urgent from a non-bulk sender and important from a bulk one. Clear marketing becomes noise; newsletters you opted into become can wait. Drafts are written by a model; the scoring is not.
Can I add my own MCP server?
Yes. Paste a Model Context Protocol server URL — a company-internal server, a community server, or a service with no connector card — and its tools are discovered automatically. Auth can be none, bearer token, API key or HTTP basic. Each tool is set to always or never allowed, and every call is logged.
Can an AI automation run on a schedule in my timezone?
Yes. Describe it in plain English — hourly, every N hours, daily, weekdays, weekends, a named weekday or monthly on the 1st — even several times a day, such as 9:00 and 16:30. It uses the timezone you set, then your saved profile's, then the browser timezone recorded when you consented, then UTC.
Can I trigger an automation from Zapier or a webhook?
Yes. Every automation has a personal trigger URL that accepts GET or POST from Zapier, curl or any system. Data you POST, up to 32,768 characters, is passed to the first step. Each URL allows 10 fires per 60 seconds and can be rotated or revoked. A paused or vacation-mode automation, including one paused for a set time, refuses to run.
Can an automation post to my apps without my approval?
No. An automation cannot post to a connected app until you turn publish on. You can also have it hold each result as a draft to approve or reject before anything is delivered.
Which AI models can I pick for a workspace mission?
The Command Center brain picker offers GPT 6 Luna, GPT 6.1 Sol, GPT-6 Astra, Claude Haiku, Claude Sonnet, Claude Opus, Gemini Flash Lite, Gemini 3.5 Flash, Gemini Pro, Grok 4.5, DeepSeek V4 Pro, Qwen 3.7 Max, Kimi K3 and GLM 5.2, per mission or per agent thread. Auto uses each agent's own brain.
Start with one app and one automation
Connect your inbox or a drive, then describe a routine once and let it run on schedule.