Help.

The Windows installer (Maxiful-x64.msi) is code-signed by PolyTrader Ltd with a Certum OV certificate, so Windows installs it like any verified app, with no unknown-publisher prompt. It needs Windows 10 or later.

A word of caution

A valid signature confirms who published the file, but good habits still matter. Only install the file you downloaded from your own Maxiful account, never from an email attachment, a forum, or a re-hosted link. If you want to be thorough, verify the file’s checksum against the one shown next to your download button (see “Verify your download” below).

The macOS app arrives as a disk image, Maxiful-arm64.dmg, signed and notarized by PolyTrader Ltd with an Apple Developer ID, so macOS opens it like any verified app, with no Gatekeeper override needed. It runs on Apple Silicon and Intel Macs with macOS 13 or later; the image carries both builds.

Open the disk image and drag Maxiful into Applications, then open it from Applications or Launchpad. Because the app is notarized, macOS opens it without a security prompt.

Only install the file you downloaded from your own Maxiful account, never from an email attachment, a forum, or a re-hosted link. If you want to be thorough, verify the file’s checksum against the one shown next to your download button before you open it.

Every release is published with its SHA-256 checksum recorded at release time. The checksum for the current build is shown right next to the download button on your account page. Comparing it against the file you actually downloaded confirms the installer arrived intact and untampered, a sensible integrity check even for a signed installer. All downloads are delivered over HTTPS from maxiful.ai only.

Windows

Open PowerShell and run:

Get-FileHash "$env:USERPROFILE\Downloads\Maxiful-x64.msi" -Algorithm SHA256

macOS

Open Terminal and run:

shasum -a 256 ~/Downloads/Maxiful-arm64.dmg

Compare the value

Letter case doesn’t matter. PowerShell prints the hash in upper case. If the values match, you’re good to install. If they don’t, delete the file, download again from your account page, and tell us at [email protected]. Don't install a file that fails verification.

Welcome. Your agent is running. Here's what to do right now to make the most of your first session.

Step 1: Open your messaging app

Your agent lives in Telegram or Slack: whichever you chose during setup. Open it now. You should see your agent listed as a contact or in a group. If you set up Telegram, look for the bot you named during the wizard.

Step 2: Say hello

Type a simple message. "Hi" works fine. Your agent will introduce itself, confirm it's ready, and tell you what it's connected to (calendar, email, and any other services you authorised during setup).

Step 3: Give it something real to do

Don't overthink this. Try one of these to get started:

• "Summarise my emails from this week"
• "What's on my calendar tomorrow?"
• "Draft a follow-up email to [name] about [topic]"
• "Find me three competitors in [your industry]"

Step 4: Check Mission Control

Open Mission Control in your browser (the link was in your welcome email). This is your dashboard. You can see your agent's health, connected services, and usage. You don't need to use it daily, but it's good to know where it is.

One thing to remember

Your agent runs on your {{device}}. If your {{device}} is asleep or turned off, your agent is paused. For best results, keep your {{device}} on and connected to Wi-Fi when you're working, or even overnight, so your agent can handle background tasks. More on this in The one rule: keep your {{device}} on.

Your agent understands natural language. You don't need special commands or a particular format. But a few habits make conversations go much more smoothly.

What works well

• Be specific about what you want. "Write a short thank-you email to Markus after our call today. We discussed the Q3 proposal" is better than "write an email."
• Give context. The more your agent knows, the better the result. Don't assume it remembers context from a conversation yesterday unless you remind it.
• Ask for a draft first. For anything you'll send to clients, say "draft this first, don't send yet." Your agent will show you the output for approval.
• Be direct about deadlines. "I need this before 3pm" works. Your agent takes timing seriously.

What doesn't work as well

• Very vague requests. "Help me with my business" doesn't give your agent enough to act on. Try to name a specific task.
• Assuming it remembers everything. Your agent has a strong memory, but it helps to re-state important context in long gaps between conversations.
• Multi-step tasks in one sentence. Break complex projects into steps: "First, summarise the attached PDF. Then suggest three follow-up questions I could send."

Approval requests

For actions that can't be undone, sending an email, deleting a file, booking something, your agent will ask for your confirmation before proceeding. Just reply "yes" or "go ahead." If you want to change something, say what to change and it'll revise first.

A good mindset

Think of your agent like a capable new team member who's great at their job but doesn't know your specific preferences yet. Be clear, be specific, and give feedback when something isn't right. It gets better the more you use it.

Your Maxiful agent is a general-purpose AI assistant running continuously on your {{device}}. It is connected to your tools, it takes actions, and it works while you are busy with other things.

Communication & email

• Triage your inbox and surface what matters
• Draft and send emails (with your approval)
• Write follow-ups, proposals, and client responses
• Handle meeting invites and scheduling

Research & information

• Search the web and summarise findings
• Monitor competitors, news, or industry topics
• Read and summarise documents you share
• Answer questions with up-to-date information

Documents & content

• Write reports, proposals, and presentations
• Edit and improve documents you're working on
• Translate text between languages
• Create structured summaries from messy notes

Files & organisation

• Organise files and folders on your {{device}}
• Find specific files when you can't remember where they are
• Rename, move, and sort documents automatically

Calendar & admin

• Check your schedule and flag conflicts
• Prepare briefings before important meetings
• Track tasks and follow up on open items

What grows over time

New skills are added regularly and update automatically. You don't have to do anything. Business plan customers can also request custom skills built around their specific workflows. Your agent learns your preferences through use, so it gets more accurate and useful the longer you work together.

Your agent runs on your computer. That is what keeps it private, and it means your computer needs to be on and connected to do its job.

What keeping your computer on actually means

Your computer needs to be:

• Powered on (not shut down)
• Not in sleep mode (or using sleep settings that allow background activity)
• Connected to Wi-Fi

When these three conditions are met, your agent is active and can handle tasks, process emails, and respond to your messages.

Sleep mode: what to do

By default, computers can go to sleep after a period of inactivity. This pauses your agent. We recommend adjusting your power settings so it stays active when connected to power:

This is especially important if you want your agent to work overnight or process tasks while you're away from your desk.

What happens if it sleeps anyway

Nothing is lost. When your computer wakes up, your agent resumes automatically, usually within a minute or two. Any messages you sent while it was asleep will be processed when it comes back online. You'll see a response as soon as it's caught up.

Travelling or away for a few days?

If you'll be away from your computer for a few days, you have two options: leave your computer on (plugged in, display off is fine), or accept that your agent will be paused until you return. Your data and configuration stay on your computer, downtime does not delete anything. As always, keep your own backups of anything you cannot afford to lose.

Wi-Fi interruptions

Brief Wi-Fi drops are handled gracefully, your agent will retry and continue when the connection comes back. If your internet is down for an extended period, your agent can still work on local tasks (file organisation, drafting documents) but won't be able to search the web or access email until connectivity returns.

Telegram is the most reliable way to communicate with your agent. Setup takes about five minutes and only needs to be done once.

What you'll need

• A Telegram account (free, download from telegram.org or your app store)
• Your agent's Telegram bot name (shown in the Maxiful setup wizard)

How it works

Your agent runs as a Telegram bot. During setup, the wizard creates a private bot just for you, nobody else can message it. You talk to your agent by messaging this bot, just like chatting with a friend.

Setting up via the wizard

If you're in the initial setup wizard (Maxiful desktop app), the Telegram step will guide you through this automatically:

1. The wizard will give you a bot name (something like @YourAgentBot)
2. Search for that name in Telegram and open the chat
3. Send the message /start to activate the bot
4. Return to the wizard. It will detect the connection and confirm

Finding your chat ID (if asked)

Occasionally the setup will ask for your Telegram chat ID. To find it:

• Open Telegram and search for @userinfobot
• Send any message to it
• It will reply with your chat ID, a number like 123456789
• Copy and paste that number into the setup field

After setup

Once Telegram is connected, every message you send to your bot goes directly to your agent on your {{device}}. Responses typically arrive within a few seconds. You can message from your phone, tablet, or desktop, Telegram syncs across all your devices automatically.

Connecting Google is a single step that covers your whole Google Workspace, Gmail included. This is what that means and how to do it.

What access your agent gets

When you connect Google, your agent gets access to six services in one go: Gmail, Calendar, Drive, Contacts, Sheets, and Docs. In practice, it mostly uses:

• Read your emails: to summarise, triage, and draft replies
• Send emails on your behalf: always with your approval first
• Read your calendar: to see your schedule and flag conflicts
• Create and modify calendar events: with your confirmation

It can also search Drive, read and update Sheets, export or draft Docs, and look up Contacts whenever you ask it to, no separate connection needed for those.

How to connect (first time)

1. Open Mission Control in your browser
2. Go to Settings → Connected Services
3. Click Connect Google
4. A Google sign-in window will open, choose your account and click Allow
5. You'll be returned to Mission Control, and Google will show as connected

Reconnecting after a disconnection

Google connections occasionally expire, especially after a period of inactivity or a password change. If your agent tells you it can't access your email, follow the same steps above, clicking Connect Google again will refresh the connection without losing any settings.

Multiple Google accounts

You can connect more than one Google account if you use separate addresses for work and personal email. Each account is listed separately in Mission Control and can be enabled or disabled independently.

Revoking access

You can disconnect Google at any time from Mission Control → Settings → Connected Services. You can also revoke access from your Google account settings directly (Security → Third-party apps with account access).

Google Workspace isn't required. Your agent can also work with email and calendar accounts outside Google.

Any email provider

Your agent can connect to any email account that supports IMAP and SMTP, which covers the vast majority of providers: iCloud Mail, Outlook/Microsoft 365, Fastmail, Zoho, a work Exchange account, and most others. Just tell your agent you'd like to connect an email account and give it the address; it'll walk you through the rest (your host's IMAP/SMTP settings and an app password where your provider requires one).

Calendars beyond Google

For calendars, your agent supports CalDAV, the standard most calendar providers speak, including iCloud, Fastmail, and Nextcloud, alongside Google Calendar itself. Tell your agent which calendar you want connected and it will guide you through the provider-specific setup.

Using both at once

You're not limited to one email or one calendar. Connect your Google Workspace account for one address and an IMAP/SMTP or CalDAV account for another, your agent keeps track of each separately.

You can speak your messages instead of typing them. Hold the microphone button in the message box, say what you need, and let go: the words arrive as text for you to check before you send.

Where it works

Voice input is part of the desktop app on macOS and Windows. It is input only. Your agents answer in text, the way they always have.

Where the transcription happens

Your speech is turned into text on your own computer. On macOS the app uses the speech recognition built into the system. On Windows it uses a local speech model that runs on the machine. Either way, the audio is not sent to us and not sent to a cloud transcription service.

Speaking a message

1. Open a chat with any agent
2. Press and hold the microphone button in the message box
3. Speak while the button is held
4. Let go: the transcript drops into the message box
5. Fix anything that came out wrong, then send

Choosing the language

Open the small menu next to the microphone to pick the language you're speaking. English, German, and Swedish are available. On Windows, a language you haven't used yet offers an Install option that downloads its speech model once; after that it works offline like the others. On macOS the languages come from the system, so if one is missing, add it in macOS Settings and it will appear in the menu.

The microphone button isn't there

• Permission. The first time you use it, your computer asks for microphone access, and on macOS for speech recognition too. If you declined, grant it in your system settings and restart the app.
• An older version. Voice input arrived in a recent desktop release. Check for updates, then restart the app.
• No microphone. If your computer has no input device available, the control stays hidden.

Nothing was transcribed

Hold the button for the whole sentence: letting go ends the recording. Very short bursts, a noisy room, or speaking a different language from the one selected in the menu are the usual causes of an empty or garbled result.

Every plan comes with the full team, a Chief of Staff plus specialists, each focused on a different area of your work, and you can add more specialist agents on any plan.

What multiple agents means

Each agent is a separate assistant with its own:

• Name and personality (you can customise both)
• Connected services (e.g. one for email, one for research)
• Skills and focus areas
• Telegram contact

For example, a small business owner might have one agent managing client communications, another handling research and competitor monitoring, and a third managing internal admin.

Adding a new agent

1. Open Mission Control in your browser
2. Go to Agents → Add Agent
3. Give your new agent a name and choose its focus area
4. Connect the messaging channel (Telegram)
5. Configure its skills and connected services
6. Click Launch: your new agent will start within a few minutes

What if I'm on Starter?

The Starter plan includes the same full team as every other plan, and you can add more specialist agents on any plan. Upgrading to Professional or Business gives you more AI capacity and faster support, not more agents, and you can change plan at any time from Mission Control → Account → Upgrade Plan. The change takes effect immediately, and you'll be charged the prorated difference for the current billing period.

Changing an existing agent

To change an agent's name, skills, or connected services, go to Mission Control → Agents → [agent name] → Settings. Most changes take effect immediately. If you're changing the messaging channel (for example, to a new Telegram bot), you'll need to complete a brief reconnection step.

Your agent receives updates automatically. Here's what to expect and how updates work.

How Updates Work

Maxiful pushes updates through the gateway service running on your {{device}}. Most updates happen silently in the background, your agent restarts briefly and continues where it left off.

What to Expect

• Brief pause (30–90 seconds): your agent may be unresponsive during the update. Messages sent during this window are queued and processed afterward.
• No data loss: your files, memory, and configuration are preserved. Updates only change the agent's capabilities and fix bugs.
• Changelog notification: after significant updates, your agent may message you with a summary of what changed.

Types of Updates

• Skill updates: new capabilities or improved versions of existing skills. These are the most common.
• Gateway updates: improvements to the core service that runs your agent. Less frequent, slightly longer restart.
• Security patches: applied as soon as available. These are never delayed.

Can I Control When Updates Happen?

Updates are automatic to ensure security patches are applied promptly. You'll always be notified of significant changes. If an update causes any issues, contact support immediately.

See What Changed

Visit our changelog for a full history of updates, improvements, and fixes.

How to move your agent to a new {{device}} without losing anything.

Before You Start

Make sure you have:

• Your new {{device}} set up and connected to the internet
• Your Maxiful account email (used for login)
• Access to your old {{device}} (recommended, but not required)

Step 1: Install on the New Computer

1. Go to maxiful.ai/account and sign in
2. Download the installer for your new {{device}}
3. Run the installer. It will set up the gateway and configure your agent

Step 2: Sign In

During setup, sign in with the same account. Your subscription, preferences, and agent identity transfer automatically.

Step 3: Reconnect Services

Some integrations (Google, Telegram) may need reconnection on the new {{device}}:

• Telegram: your bot token transfers automatically. No action needed.
• Google services: you may need to re-authorize. See our Google connection guide.
• Local files: if your agent worked with files on your old {{device}}, those files don't transfer. Move them manually or via cloud storage.

Step 4: Verify

Message your agent on Telegram. If it responds, you're set. The first response may take a minute while the agent initializes.

What About the Old Computer?

Your agent only runs on one {{device}} at a time. Once you've verified the new installation works:

• The old installation will stop receiving heartbeats automatically
• You can uninstall from the old {{device}} if you like, but it's not required

What Transfers Automatically

• Your subscription and account
• Your agent's identity and personality
• Your preferences and instructions
• Your Telegram bot connection

What Doesn't Transfer

• Local files on the old {{device}}
• Browser sessions and cookies
• Locally cached data

If you run into any issues during migration, contact support, we'll walk you through it.

Telegram is the most reliable channel for talking to your agent. This article covers setup confirmation, common issues, and how to reconnect.

Confirming Telegram is working

After setup, send your agent a message like "Hello" or "Are you there?". It should respond within 10–15 seconds. If you see a response, Telegram is working correctly.

Agent not responding on Telegram

If your agent stops responding, work through these steps:

• Check your {{device}} is on. Your agent runs on your {{device}}, if it's asleep or shut down, it can't respond. Wake it up and try again.
• Check your Wi-Fi. Your agent needs internet access to receive and send messages.
• Wait 2 minutes. If your {{device}} just woke from sleep, the agent needs a moment to reconnect.
• Check Mission Control. Open Mission Control in your browser. It will show if your agent is online, offline, or showing any errors.

Bot shows as offline or "Bot stopped"

This occasionally happens after a restart or system update. To fix it:

1. Open Mission Control → Agents → [your agent] → Status
2. Click Restart Agent
3. Wait about 60 seconds
4. Send a message in Telegram, your agent should respond

Reconnecting your Telegram bot

If the bot connection was broken (e.g. after creating a new bot token), go to Mission Control → Settings → Connected Services → Telegram → Reconnect. Follow the same steps as initial setup: find your bot in Telegram and send /start.

Multiple devices

You can message your agent from any device where you have Telegram installed, phone, tablet, or computer. All messages go to the same agent, and conversations are synced automatically.

Your agent is designed to act. For actions that can't easily be undone, it is designed to ask for your permission first. This is called an approval gate.

What triggers an approval request

Your agent will ask before it:

• Sends an email on your behalf
• Deletes or moves files permanently
• Books or modifies calendar events
• Makes any purchase or financial action
• Accesses a new service it hasn't used before

What an approval looks like

Your agent will send you a message in Telegram describing what it's about to do and asking you to confirm. For example:

"I've drafted a reply to Markus. Here it is: [draft]. Shall I send it? Reply yes to send, or tell me what to change."

How to respond

• Approve: Reply "yes", "go ahead", "send it", or similar
• Edit first: Tell your agent what to change, "make it shorter" or "add a line about the deadline", and it will revise and ask again
• Cancel: Reply "cancel", "skip this", or "don't send" and the action is dropped

Adjusting approval gates

If your agent is asking for approval too often for something routine (like sending a specific type of email), you can adjust this in Mission Control → Agent Settings → Approvals. You can set certain action types to auto-approve, or tighten the gates for anything sensitive.

What if I don't respond?

If your agent is waiting for approval and doesn't hear back, it will hold the action and remind you once. After a set period (usually 24 hours), it will cancel the pending action and let you know. Actions behind an approval gate wait for your confirmation; action types you have set to auto-approve run without it.

Your agents can message each other. When a job needs two of them, your chief of staff hands out the work and the specialists settle the details between themselves, so you aren't carrying messages back and forth. This is switched on for new installations.

What it looks like in chat

An exchange between two agents appears in your thread, folded under the name of the agent that was brought in. Open the fold to read what was asked and what came back. Your chief of staff still reports the outcome to you in plain language, so you can ignore the detail unless you want it.

A channel for each agent

Chat lists every agent as its own channel, alongside one for the whole team. Message a specialist directly when you know who you need, or message your chief of staff and let it delegate. Unread counts show you where something new is waiting.

Adding an agent

Use the + above the channel list and describe the job you need doing. The new agent names itself and joins the team.

Can they talk in circles?

No. Exchanges between agents are loop-guarded, so a conversation can't bounce back and forth indefinitely between two of them.

Does this change what needs my approval?

No. Approval gates work exactly as before. Agents talk to each other freely, but anything that reaches the outside world, sending an email, publishing, spending money, still waits for your yes.

Your agent runs on your {{device}}. If your {{device}} is asleep, shut down, or disconnected from the internet, your agent cannot respond.

What to check first

1. Wake your {{device}} if it was sleeping
2. Make sure it is connected to the internet
3. Open Mission Control and confirm your agent shows as online
4. Wait 60 to 90 seconds after wake or restart, then try again

If your agent still is not responding after that, restart it from Mission Control or contact support.

Google connections occasionally expire. When this happens, your agent will tell you it can't access your email or calendar, and you'll see a warning in Mission Control. It's easy to fix.

Why this happens

Google connections expire after a period of inactivity, or when your Google password changes. It's a security feature, nothing has gone wrong with your agent or your account.

How to reconnect

1. Open Mission Control in your browser
2. Go to Settings → Connected Services
3. You'll see a warning icon next to Google, click Reconnect
4. A Google sign-in window will open
5. Sign in to your Google account and click Allow when asked to grant access
6. You'll be returned to Mission Control, Google will show as connected again

After reconnecting

Your agent resumes full access immediately. Any tasks it couldn't complete while Google was disconnected (like reading new emails) will be processed right away. You don't lose any configuration or history.

Preventing disconnections

The most common cause of Google disconnections is changing your Google password. If you change your password, reconnect Google in Mission Control straight away to avoid any interruption. Your agent will also notify you proactively when a reconnection is needed.

Agents make mistakes. What to do when yours does, and how it learns from corrections.

Why Mistakes Happen

Your agent uses large language models to understand requests and take action. These models are remarkably capable, but they can:

• Misunderstand context: especially with ambiguous requests or unfamiliar topics.
• Make confident-sounding errors: AI can be wrong while sounding certain.
• Miss nuance: sarcasm, implied context, or domain-specific jargon can trip it up.

This is normal. Even the best AI systems have limitations.

What to Do

1. Tell your agent directly. Say something like: "That's not right, I meant X" or "Don't do that again." Your agent understands corrections in plain language.
2. Be specific. Instead of "that's wrong", try "the meeting is on Thursday, not Wednesday", specific corrections stick better.
3. Check important outputs. For anything critical (emails to clients, financial data, scheduling), review before it goes out. Your agent can draft, but you should approve.

How Your Agent Improves

Your agent has memory. When you correct it, that correction is stored and referenced in future interactions. Over time, your agent gets better at:

• Understanding your preferences
• Avoiding repeated mistakes
• Matching your communication style
• Knowing what needs your approval vs. what it can handle independently

Setting Boundaries

You can tell your agent things like:

• "Always ask before sending emails on my behalf"
• "Never schedule meetings before 10am"
• "Double-check any numbers before including them in reports"

These instructions become part of your agent's operating rules.

Our Commitment

We're transparent about AI limitations because trust matters more than hype. Your agent is a tool that gets better with use, not a magic solution. The more you work with it, the more useful it becomes.

Your agent usually responds within a few seconds. If responses are taking longer than usual, here's what's likely happening and what (if anything) you need to do.

What's normal

• Simple questions: 3–10 seconds
• Drafting a long document: 15–30 seconds
• Web research: 20–45 seconds (your agent is actually searching and reading)
• Complex multi-step tasks: 1–3 minutes

If your agent is working on something complex, it may send a "Working on it..." message to let you know it's still going.

Causes of slower-than-usual responses

• High load: If you've asked for something research-heavy or are running multiple tasks, responses will take longer. This is normal.
• Slow internet: Your agent uses the internet for research and AI processing. A slow connection will slow everything down.
• Computer running low on resources: If many apps are open and your {{device}} is working hard, your agent may run slower. Closing unused apps can help.
• AI service delays: Occasionally, the cloud AI services your agent relies on experience brief slowdowns. This is rare and resolves on its own.

When to worry

If your agent takes more than 5 minutes to respond to a simple question, or doesn't respond at all, it may have stopped. Check Mission Control to see if it's still online, and restart it if needed. See Agent not responding for the full checklist.

If your {{device}} restarts, whether planned or after a power cut, your agent restarts too. Here is what to expect.

What happens after a restart

When your {{device}} boots up, Maxiful starts in the background and brings your agent back automatically. You usually do not need to do anything. The process takes about 2 to 3 minutes from login to your agent being fully online.

If it does not come back automatically

Occasionally, especially after an operating system update, startup needs a nudge:

Messages sent while your {{device}} was off

Any messages you sent while your {{device}} was restarting will be queued and processed as soon as your agent reconnects. No messages are lost.

Your Maxiful subscription covers everything you need to run your AI agent, one straightforward bill, nothing extra.

What's included in every plan

• Your AI agent: running 24/7 on your {{device}}, fully managed by us
• AI access included: the AI that powers your agent is provisioned and paid for by Maxiful. You don't need a separate account anywhere.
• All skills included: email management, research, calendar, documents, file organisation, and more. New skills added regularly, updated automatically.
• Mission Control dashboard: monitor your agent's health, manage settings, view usage
• Updates and maintenance: we keep your agent running, updated, and secure
• Support: included at the response level for your plan

Plans at a glance

• Starter: €49/month: your full AI team, email support (48-hour response)
• Professional: €129/month: more AI capacity, email + chat support (24-hour response)
• Business: €299/month: maximum AI capacity, priority support (4-hour response on weekdays)

One bill, everything included

You pay Maxiful monthly. That's it. There's no separate bill for AI usage, no per-message fees, and no hidden costs. We manage everything on the backend so you don't have to think about it.

When does billing happen?

Your subscription renews automatically on the same date each month, the date you first subscribed. You'll receive an email receipt after each payment. You can view your billing history at any time in Mission Control → Account → Billing.

About AI capacity

Every plan includes AI reasoning: no separate AI subscription, no per-question charges. Each tier comes with a monthly AI budget, and our model proxy records usage metadata (which model was called and token counts, never content) to keep usage within your plan's included budget. If your agents regularly need more capacity, a higher tier gives you a larger AI budget and faster support.

You can upgrade or downgrade your plan at any time from Mission Control. Changes are handled automatically with fair proration.

Upgrading your plan

To upgrade (e.g. Starter → Professional or Professional → Business):

1. Open Mission Control → Account → Plan
2. Click Upgrade Plan and choose your new tier
3. Confirm the change, your new plan activates immediately

You'll be charged a prorated amount for the remainder of the current billing period, only for the days at the new rate. Your next full month will be billed at the new plan price.

Downgrading your plan

To downgrade, follow the same steps and select a lower tier. Downgrades take effect at your next billing date, you keep your current plan's features until the end of the period you've already paid for.

Downgrading to a lower plan reduces your included AI capacity and support level, not the number of agents, your full team stays intact on every plan.

Proration explained

Proration means you're only charged for what you use. If you upgrade halfway through the month, you pay for half a month at the old price and half at the new, automatically calculated and shown before you confirm.

When changes take effect

• Upgrades: Immediately
• Downgrades: Next billing date
• Increased AI capacity (after upgrade): Available immediately

If a payment fails, here's what happens and how to fix it quickly.

What happens when a payment fails

If your card declines, we'll:

1. Send you an email notification immediately
2. Retry the payment automatically after 3 days
3. Retry once more after another 3 days

During this period, your agent continues running. We don't suspend immediately on a failed payment. You have time to fix the issue without any service interruption.

Grace period

You have a 7-day grace period after the first failed payment to update your payment details. If the payment hasn't been resolved after 7 days, your subscription will be paused until payment is updated.

Updating your payment method

1. Open Mission Control → Account → Billing
2. Click Update Payment Method
3. Enter your new card details, stored securely via Stripe
4. Click Save: your next payment attempt will use the new card

Alternatively, click the link in the payment failure email. It goes directly to the payment update page.

Charges I don't recognise

If you see an unexpected charge, check your billing history in Mission Control → Account → Billing. If something still doesn't make sense, email [email protected] with the date and amount. We'll investigate and respond promptly.

You can cancel your subscription at any time. No phone calls, no hoops, no retention scripts.

How to cancel

1. Open Mission Control → Account → Plan
2. Click Cancel Subscription
3. Confirm your cancellation

Alternatively, email [email protected] and we'll process it immediately.

What happens to your agent

Your subscription remains active until the end of the billing period you've paid for. After that, your agent stops running and is removed from your {{device}}. Your files and data are not deleted, they're stored on your {{device}} and they stay there. You keep everything you created with your agent.

What we keep and what we delete

• We delete: Your account, billing information, and any server-side account data
• Stays on your {{device}}: All agent files, conversation history, and anything your agent created or organised
• Your Google/Telegram connections: These are revoked from our side. You can also revoke them manually from those services if you prefer

Changed your mind?

If you cancel and want to come back, you can resubscribe at any time at maxiful.ai. Your previous configuration won't be available (we deleted it), but setup is straightforward and most of your data, your files, documents, and anything saved on your {{device}}, will still be there.

Privacy isn't a policy document with us. It's how the product is built. Here's exactly where your data lives.

Stays on your {{device}} (always)

• Your files and documents: anything your agent reads, creates, or organises
• Your conversation history: every message between you and your agent
• Your agent's memory: preferences, context, and anything it's learned about how you work
• Email and calendar data: fetched and processed locally, not stored on our servers

What goes to the cloud (briefly)

When your agent needs to think, to answer a question, write a document, or process a request, it sends a prompt through Maxiful's model proxy to a cloud AI service. This prompt is:

• Routed via our model proxy, which authenticates the request and enforces your plan's spend limits, the proxy does not store the content of the prompt
• Processed for inference under no-training terms; the content is not stored by us
• Never used to train AI models (we use services with strict no-training agreements)
• Transmitted over encrypted connections

What Maxiful stores

On our servers, we store only what's needed to run your account:

• Your name and email address
• Your subscription and billing information
• Agent health and uptime status (anonymous technical data)
• Usage metadata from your agent's AI calls, which model was called and how many tokens it used, never the content. Our model proxy records this to enforce your plan's AI budget and power your usage dashboard; these events are buffered on our infrastructure for at most 7 days

We do not store your conversations, files, emails, calendar events, or agent memory.

Third-party services

We use a small number of trusted third-party services to run the business: Stripe for payments, and standard infrastructure providers for hosting our account management systems. These providers do not have access to your agent's local data or conversations.

Short answer: very little. By design.

What we can see

• Your account information: name, email, subscription tier
• Your billing history: payments, invoices, plan changes
• Agent health signals: whether your agent is online, offline, or experiencing errors (anonymous technical status, not content)
• Usage metadata: which AI model your agent called and how many tokens each call used, recorded by our model proxy so we can enforce your plan's AI spend limits. Token counts only: never the content of those calls

What we cannot see

• Your conversations: messages between you and your agent are stored on your {{device}}, not our servers
• Your files: documents, emails, calendar data, anything your agent touches lives locally
• Your agent's memory: all context and preferences are stored on your device
• The content of your prompts: prompts travel from your {{device}} through Maxiful's model proxy to the AI provider. The proxy authenticates each request and enforces spend limits; it records only usage metadata (model name and token counts) and does not store the content of your prompts or the AI's responses

What about support access?

If you contact support and we need to diagnose a problem on your {{device}}, we will ask for your explicit permission before connecting. You can approve, limit, or deny this at any time. Support sessions are time-limited and logged. You can see and revoke access whenever you like from Mission Control.

Our internal policy

No Maxiful employee accesses customer systems without explicit, documented consent. This is policy, and it applies to everyone on the team including founders.

You have full rights over your data under GDPR, and we make it easy to exercise them.

Your rights at a glance

• Right to access: Request a copy of everything we hold about you
• Right to deletion: Ask us to delete your account and all associated data
• Right to correction: Ask us to correct inaccurate information
• Right to portability: Export your data in a standard format
• Right to object: Object to any processing you haven't consented to

Exercising your rights

Email [email protected] with the subject line "Data Request" and describe what you need. We'll respond within 30 days (usually much faster). For common requests like account deletion, you can also initiate this directly from Mission Control → Account → Privacy.

Your data on your {{device}}

Remember: the majority of your data, conversations, files, agent memory, lives on your {{device}}, not ours. You already have full access and control. When you cancel or request deletion from us, we delete your account data from our servers; the data on your {{device}} stays yours.

Data we hold about you

You can request a complete export at any time from Mission Control → Account → Privacy → Download My Data. This export covers everything we hold on our servers, account info, billing history, and account-level settings.

Questions?

Our privacy policy is at maxiful.ai/privacy. For specific questions about how we handle your data, email [email protected]: a real person will respond.

Your agent can read emails, manage files, and take actions on your behalf. This is how it is designed to do only what you intend.

Approval gates

For high-impact actions, sending emails, deleting files, making bookings, your agent is designed to ask for your confirmation before proceeding, unless you have set that action type to auto-approve. You see exactly what it's about to do, and you choose whether to proceed. See Approvals and confirmations for details.

Skill verification

Every capability your agent has (we call these "skills") is tested and reviewed before we make it available, and is given scoped permissions, limited to what it needs. A research skill can search the web without touching your files; an email skill can read your inbox without access to your banking. Skills from third-party marketplace publishers are subject to automated checks but remain the publisher's responsibility, so review a listing before installing.

Sandbox isolation

Where your operating system supports it, your agent runs in a sandboxed environment on your {{device}} that limits what skills can reach. Sandboxing reduces risk but is not a guarantee that an action is impossible, so keep approval gates on for anything you would not want run without review.

What your agent is designed not to do

By design, and with the sandbox and permission settings you configure, your agent should not:

• work with files or folders outside the areas you've approved;
• connect to services you haven't authorised;
• take high-impact actions you've kept behind an approval gate without your confirmation; or
• send your data anywhere other than the AI services needed to process your requests.

No automated safeguard is perfect. Review outputs before you rely on them, keep approval gates on for consequential actions, and keep backups.

Keeping it secure

Keep your {{device}} and your Maxiful app up to date, security updates are included in your subscription and installed automatically the next time your device checks for updates. If you ever suspect something unusual, contact support immediately at [email protected].

There's always a real person on the other end. Here's how to reach us and what to expect.

Email support

The best way to reach us for most issues:

[email protected]

Describe your issue and include your account email. We'll respond within the time for your plan.

Response times by plan

• Starter: Email support, response within 48 hours (weekdays)
• Professional: Email + chat support, response within 24 hours (weekdays)
• Business: Priority support, response within 4 hours (Monday–Friday, 09:00–18:00 CET)

Support hours

Our support team works Monday–Friday, 09:00–18:00 CET. Messages sent outside these hours will be picked up on the next working day (counted toward your plan's response time from when we open).

Self-service (fastest for common issues)

Many common issues can be resolved immediately without contacting us:

• Mission Control: check your agent's status, restart it, reconnect services
• This Help Centre: step-by-step guides for the most common situations
• Your agent, ask it "are you working properly?" or "check your connection", it can often diagnose and fix issues itself

Most issues can be resolved in a few minutes without waiting for a reply. Try these steps first. They fix the majority of problems we hear about.

Agent not responding?

1. Is your {{device}} on and awake? (Most common cause)
2. Is your {{device}} connected to Wi-Fi?
3. Check Mission Control: is your agent shown as online?
4. Try restarting your agent from Mission Control (takes about 60 seconds)

Google or calendar issues?

1. Go to Mission Control → Settings → Connected Services
2. Check if Google shows a warning icon
3. Click Reconnect and sign in again. This fixes most email/calendar issues

Messaging not working (Telegram)?

1. Check your internet connection on your phone
2. Check Mission Control: is the messaging channel shown as connected?
3. Try sending /start to your bot again (Telegram)

Billing question?

• Your billing history and invoices are in Mission Control → Account → Billing
• You can update your payment method there too

Still stuck?

If none of the above helps, we're here. Email us at [email protected]: see the next article for what to include so we can help faster.

The more context you give us upfront, the faster we can help. Here's what makes a great support message.

Always include

• Your account email: so we can find your account immediately
• What's happening: describe the problem in your own words. "My agent stopped responding" is great. "It's broken" leaves us guessing.
• When it started: "this morning", "after I updated my computer", "since yesterday afternoon"
• What you've already tried: saves us suggesting steps you've already done

Helpful extras

• Which messaging channel: Telegram or Slack
• Your computer model and operating system version, found in your system information settings
• A screenshot of Mission Control: especially if it shows an error or warning
• Any error message you saw: exact text if possible, or a screenshot

What you don't need to include

Don't include your password, your Google credentials, or any sensitive personal information. We'll never ask for these. If resolving your issue requires access to your computer, we'll request your permission through a formal support session, not via email.

Good example

"Hi, my agent stopped responding in Telegram about 3 hours ago. It was working fine this morning. I've checked Mission Control and it shows the agent as online, but messages aren't getting through. My computer is on and connected to the internet. I tried restarting the agent from Mission Control but it still doesn't respond. I'm on Starter plan. Account email: [email]"

That gives us everything we need to start investigating straight away.