Questions buyers ask before they install

Straight answers to what support leads and their IT teams ask us first. Everything here describes what the product does today. When something changes, this page changes with it.

Are you a safe bet?

What happens if Slack changes its rules or blocks the app?

Slack is one way in, not the only one. The same inbox takes email from your Gmail support address, chats from a widget on your website, and requests through a REST API. A workspace can run on Gmail alone: without Slack, your team's alerts arrive by email from your support mailbox. If Slack ever cut the app off, your tickets, customers, help docs and history stay in Scaled Support and keep working over email.

You are a small company. What if you disappear?

Scaled Support is built and run by its founder. We say so up front rather than hide it. Your data is never locked in: an admin can download everything in the workspace at any time, free, as one zip of CSV and JSON files (Settings → Data & AI → Download everything). You can also export any ticket as a PDF, and read your tickets, customers and replies through the REST API.

What does it cost, and can a small team grow into it?

One plan per workspace, from $49 a month, with unlimited teammates and every feature included. You choose its size by the AI credits you need each month ($1 buys 10 credits, and a credit writes about 6 replies), from $49 up to $1,999, or Enterprise beyond that. Extra credits come in $25 packs. When credits run out the AI pauses instead of charging you more. There are no seat minimums and no annual contract. Any price change is announced 90 days ahead and never applies to a month you already paid for, you cancel in one click from your settings, and the free full download keeps working for 30 days after you cancel. These apply from the day paid plans start; the whole pricing promise is on the home page.

What can it see?

Which Slack permissions does it ask for, and why?

Only a Slack admin can install it. The app itself asks to:

• see your channel list and which channels are shared with other companies (Slack Connect), so you can pick which ones to watch
• join the public channels you pick, and read messages in channels it is a member of, so it can turn them into tickets
• post replies and status reactions in those threads, under your own bot name and logo
• send the teammate who owns a ticket a direct message, if you switch that on (never used to message customers)
• look up people's names and email addresses, so it can tell your teammates from your customers and match customers to accounts

Two more are optional and not used yet: reading your workspace's name and icon, and noticing a new Slack Connect invitation. You can install without them.

Separately, each teammate can choose to connect their own Slack so the replies they write or approve post under their own name and picture. That asks for one permission only: to post messages as them. Nobody has to, and it never lets the app read anything as them. Every permission and the reason for it is listed on our security page.

Does it read all our channels or our DMs?

No. It never reads direct messages. It only reads channels it is a member of: public channels you choose (by a naming rule or by hand), and private channels only if someone invites it. It does not read old history; it starts from the moment it joins. For channels you don't watch, it keeps only the channel name, so it can show you the list; messages from them are dropped and their text is not kept.

Which Gmail permissions does it ask for, and why?

Three: read (to see new support emails), send (to reply from your support address) and modify (to label emails it has handled, so your inbox stays tidy). It only reads mail that arrives in the inbox after you connect it, plus your team's own replies in threads it is already tracking, so it knows a person has answered. It skips spam, newsletters, mailing lists, auto-replies and mail from your own company.

Can we switch it off?

Yes, at any time. Pressing Disconnect on Slack or Gmail in the dashboard, or uninstalling the Slack app, stops it straight away and deletes the stored access token. Your tickets and history stay until you delete them.

Where does our data go?

Where is our data stored?

Slack and Gmail access tokens are encrypted (AES-256-GCM). Every table is locked to your workspace by row-level security, so one company can never read another's data. The full list of sub-processors is on our security page.

What do the AI providers see?

To draft a reply, the AI receives the customer's question and the earlier messages in that thread, the passages from your help docs that match it, and up to three earlier answers your team gave the same customer. The question's text is also turned into a search index entry to find those passages. AI processing runs outside the EU, usually in the United States. Every AI request is sent with zero data retention and data collection refused, so it only goes to providers that process it and keep no copy. Nothing is used to train AI models.

Do you store attachments?

No. We keep the file's name and type so the ticket shows that something was attached. The file itself is never downloaded.

How long do you keep data, and can we delete it?

Tickets and their messages are deleted automatically 365 days after their last activity, by a nightly job. Admins can shorten this to as little as 30 days. An admin can also delete any ticket at any time, and its messages are deleted with it. Deleting a whole workspace is not self-serve yet: email support@scaledsupport.com and we will do it for you. More detail is on our security page.

How do you handle personal data?

Customer messages are stored as they were written, because the bot needs them to answer and your team needs them for history. Names, email addresses and Slack ids are kept so each message has a sender. Nothing is masked automatically. Please don't use Scaled Support for card numbers or health records: we are not PCI or HIPAA certified.

Do you have SOC 2?

No, not yet. We will start a SOC 2 audit when customers ask for it, and we will say so on this page with a date the day it starts. Until then we are happy to answer your security questionnaire directly.

What will the bot say to my customers?

Will the bot reply on its own?

Not until an admin turns it on. Automatic replies are off for every workspace to start with, and every reply waits for a teammate's approval. Meanwhile the bot practises: for each type of question, the dashboard shows how many replies would have gone out on their own and how much your team changed them. Once automatic replies are on, the bot sends one on its own only when all of these are true:

• its send score is at least 85% (you choose the bar). The score mostly measures how closely your help docs match the question and how recently those docs were checked; the AI's own guess counts for only 15%. A reply with no help-doc passage behind it, or only a loosely related one, scores zero.
• the customer's tier and the question's type are ones you allowed
• the question is not about deleting data, an upset customer, or something outside your docs
• if it touches a risky topic (refunds, cancellations, pricing, security, account ownership or contract terms), you switched that topic on and the send score is at least 95% (you choose this number per topic too). Risky topics are checked in the customer's message and in the bot's own draft.
• it does not need to ask the customer for more information
• every link in the reply comes from your own docs
• no teammate has replied in that thread, and the customer has not asked for a person

Every risky topic starts switched off, and the bot never promises a refund, credit or discount. Questions sorted as access or security always wait for a person, whatever the settings.

What exactly does the bot post in our customer's channel?

Only inside the thread of the customer's message, never as a new top-level post:

• small status reactions on the customer's message: 👀 seen, ⏳ working, ✅ answered, 🧑‍💻 a teammate has it
• the reply itself: under your bot name and logo when the bot sent it alone, or under the teammate's own name when they connected their Slack
• when the customer asks for a person: one short message saying a person will reply
• outside your opening hours, if you set them: a short away message

It never posts "ticket created" messages and never @mentions anyone in the customer's thread. A "Thanks, we're on it" message is off unless you switch it on. Small talk such as "hi", "thanks" or 👍 gets no reaction and no ticket, and a thank-you does not reopen an answered ticket. The teammate who owns the ticket is told in your private review channel, or by a direct message if you switch that on.

Will our customers know they are talking to AI?

Yes. Replies the bot sent without a teammate's check carry a short line saying they are automated, and a second line telling the customer to reply "person" to reach someone on your team. If a customer asks whether they are talking to AI, the bot says so and never claims to be a person. Replies a teammate approved or wrote carry no label, because they are the teammate's.

What happens when a teammate steps in?

The bot steps back for good. Once someone from your team has replied in a thread, from Slack, Gmail or the dashboard, or the customer has asked for a person, the bot never answers there on its own again. It can still write drafts, which wait for your team. The first time a customer asks for a person, the bot replies once to say the team has it and a person will reply.

What if it gets an answer wrong and nobody notices?

Every reply, automatic or approved, is kept on the ticket with the help-doc passages it used, and every action the bot takes is written to an audit log your admins can read. The Ally AI replies page lists everything the bot sent on its own, with a Mark as wrong button and a breakdown of why it said each one, plus rejected drafts and customers who came back within 48 hours. A weekly digest of the same goes to your review channel every Monday. A ticket only counts as resolved once 48 hours pass with no word from the customer.

Our help docs are a mess. Is it still useful?

Yes. Ticketing, grouping a customer's quick burst of messages into one ticket, assigning an owner and response-time tracking all work with no docs at all. With thin docs the bot drafts less and sends nothing on its own, so a weak wiki costs you drafts, not wrong answers. The Gaps page lists the questions your docs could not answer, grouped and counted, so you know what to write first.

Will it work at our size?

Does it scale past a handful of customer channels?

It watches as many Slack Connect channels as you share, and joins new ones automatically when their name matches your rule. All of them land in one inbox you can filter by customer, tier, priority, owner and status. New workspaces have a daily limit of 300 AI calls, which we raise on request.

Slack search is poor. How do we find past answers?

Each customer's tickets are kept together, so their history is one filter away, and the answers your team gave a customer are reused when the bot drafts their next reply. Ask AI answers questions across your tickets and docs. There is no full-text search box across tickets yet.

Will it flood our Slack with notifications?

No. It never posts ticket-created messages in customer channels, and internal notes go to one private review channel you choose. Without Slack, those notes come by email to the teammate who owns the ticket.