AI Assistant for Slack and Microsoft Teams: Cost, Set-up and Security
Short answer: You can buy built-in AI or build your own. Slack includes AI features in its Business+ plan at $15 per user a month (annual). Microsoft 365 Copilot is a $30 per user a month add-on for enterprise plans. A custom assistant that answers from your own documents inside Slack or Teams typically costs $5,500–$14,500 to build.
That build range comes from our AI cost estimator. Answer eight questions and it prices your own scope and timeline in about two minutes.
Two routes: built-in AI or a custom assistant
Most teams start by asking whether the AI that comes with their chat tool is enough. It often is for summaries, search and writing help. A custom assistant makes sense when you need answers from systems the built-in AI cannot reach, your own rules, or actions in your other tools.
| Option | Price (USD) | What you get |
|---|---|---|
| Slack Pro | $7.25 per user a month (annual), $8.75 monthly | Thread and channel summaries, huddle notes, AI assistant apps |
| Slack Business+ | $15 per user a month (annual), $18 monthly | Pro features plus Slackbot as a personal AI agent, AI search, daily recaps, file summaries, translations and AI steps in Workflow Builder |
| Slack Enterprise+ | $45 per user a month (annual) | Business+ features plus enterprise search |
| Microsoft 365 Copilot (enterprise add-on) | $30 per user a month paid yearly, or $31.50 paid monthly on an annual commitment | Copilot across Microsoft 365 apps, including Teams; needs a qualifying Microsoft 365 plan |
| Microsoft 365 Copilot Business (add-on) | $18 per user a month paid yearly (offer price to 31 Dec 2026; list $21) | For eligible Microsoft 365 Business plans |
| Microsoft Copilot Chat | No additional cost with an eligible Microsoft 365 subscription | Chat with web data and file uploads (limits apply), with IT controls |
| Custom assistant (RixDigi estimator) | $5,500–$14,500 one-off build, plus usage | Answers from your own sources, your rules, and optional actions in your tools |
Slack prices include the whole Slack plan, not only the AI. Microsoft prices are for the Copilot add-on on top of Microsoft 365.
What seat licences add up to
Per-user pricing is simple, but it grows with headcount. For a 50-person team, Microsoft 365 Copilot at $30 a month comes to $18,000 a year. Upgrading 50 people to Slack Business+ at $15 a month comes to $9,000 a year, though that also buys the rest of the plan.
A custom assistant has a different shape. You pay once to build it, then for AI usage and optional support. Our estimator puts support at 15–25% of the build price a year, with a minimum of $200 a month. Language model costs for internal questions are often small, because each answer uses only a few thousand tokens. If the assistant must answer from your own documents, our RAG development cost guide shows what that layer costs to build and run. For an assistant that also acts in your tools, our AI agent development cost guide covers the extra build effort.
The real question is not which is cheaper. It is whether the built-in AI can see the information your team actually asks about.
When a custom assistant is worth it
- Your knowledge lives outside the chat tool. For example, in a help centre, a CRM, an ERP, a ticketing system or a shared drive the built-in AI does not index.
- You need fixed answers. HR, IT or compliance teams often want answers only from approved policies, with citations.
- You want actions, not only answers. Raise a ticket, look up an order, book leave or update a record from a Slack or Teams message.
- You use both Slack and Teams. One assistant can serve both, so you keep a single set of sources and rules.
- Only some staff need it. A support or sales team of 15 may not justify AI seats for the whole company.
What a custom Slack or Teams assistant costs
We ran three common scopes through the estimator's own pricing rules.
| Scope | What it includes | Build price | Timeline |
|---|---|---|---|
| Knowledge assistant in one workspace | Slack or Teams, 1–2 connected tools (for example Google Drive or SharePoint), up to a few hundred files, standard business data | $5,500–$14,500 | 5–10 weeks |
| Assistant that takes actions | An AI agent in Slack or Teams that reads and acts in 1–2 tools, such as a help desk or CRM, with some documents | $7,000–$17,000 | 6–10 weeks |
| Company-wide, Slack and Teams | Both channels, 3–5 connected systems, thousands of documents, 1,000–20,000 questions a month, personal data | $12,000–$30,000 | 9–14 weeks |
Urgent delivery in under three weeks adds 25%. AI usage and hosting fees are extra and depend on volume.
There is also a low-code middle route on Microsoft. Copilot Studio is priced at $200 a month for a pack of 25,000 Copilot Credits, and it needs an Azure subscription to run agents. It suits simple agents that stay inside the Microsoft stack.
How set-up works in Slack
Slack gives developers dedicated surfaces for AI apps. According to its developer docs, these include a split-view container, a top navigation entry point, app threads, text streaming and suggested prompts. The assistant feels native: staff open it beside a channel and ask questions there.
Slack also offers an MCP server, which lets AI apps search channels, send messages and manage canvases. A Slackbot MCP client can connect your own MCP server's tools to Slackbot. This is useful if you want Slackbot itself to call your internal systems.
One rule matters for budgeting. Since 29 May 2025, new commercially distributed apps outside the Slack Marketplace face tight limits. They get one request a minute on conversations.history and conversations.replies, with at most 15 messages per call. Internal apps built by a company for its own workspace keep 50+ requests a minute and up to 1,000 messages. So build your assistant as an internal app in your own workspace, unless you plan to list it on the Marketplace.
How set-up works in Microsoft Teams
Microsoft's docs say the Teams AI library has been renamed Teams SDK. It is generally available for JavaScript and C#, with Python in developer preview. Developers use the Teams SDK or Bot Framework SDK with the Microsoft 365 Agents Toolkit (previously Teams Toolkit).
A Teams bot can run as a personal chat, in a group chat or in a channel. Most knowledge assistants start as a personal chat, then add channel use once answers are reliable.
Azure Bot Service, which connects bots to channels, is described by Microsoft as free to use, with charges for the resources you consume. Check the Azure pricing calculator for your region before you commit.
Security checklist
An assistant inside your chat tool can see a lot. Agree these points before the build starts:
- Permissions follow the source. The assistant should only answer from documents and channels the person asking can already open.
- Least-privilege scopes. Request only the Slack scopes or Microsoft Graph permissions the assistant needs, and review them with IT.
- Data handling. Decide which AI provider processes messages, where data is stored, and whether prompts and answers are logged.
- No training on your data. Confirm in the provider's terms that your business data is not used to train shared models.
- Sensitive data. Keep HR cases, health and payment data out of scope, or add masking and stricter controls.
- Audit and review. Keep logs of questions, sources used and actions taken, and review a sample each month.
- Human approval for actions. For anything that changes a record or sends a message, start with a confirm step.
If your data is regulated, the build takes longer and costs more. The estimator adds 15% for personal or financial data and 30% for regulated data or private hosting.
A typical rollout plan
- Week 1: scope. Pick one team and 30–50 real questions they ask today. List the sources that hold the answers.
- Weeks 2–4: build. Connect the sources, set up search and permissions, and create the Slack or Teams app.
- Weeks 4–6: test. Run the question set, fix gaps, and check that the assistant refuses questions it cannot answer.
- Pilot. Release to one team, collect feedback in a channel, and track how many answers people accept.
- Scale. Add teams, sources or actions once the pilot answers well.
Our internal workspace assistants service follows this plan. See the wider AI agents services page for assistants that also take actions. Unsure which you need? Read our AI agents vs chatbots guide. When you are ready, book a call.
Frequently asked questions
How much does an AI assistant for Slack cost?
Built-in AI comes with Slack Pro ($7.25 per user a month, annual) and more with Business+ ($15). A custom assistant on your own documents typically costs $5,500–$14,500 to build, plus usage and optional support.
How much does Microsoft 365 Copilot cost per user?
The enterprise add-on is $30 per user a month paid yearly, or $31.50 paid monthly on an annual commitment. Copilot Business is $18 per user a month paid yearly until 31 December 2026, against a list price of $21.
Can one AI assistant work in both Slack and Teams?
Yes. One back end can serve a Slack app and a Teams app, sharing the same sources and rules. Our estimator prices a company-wide build across both at $12,000–$30,000.
Is it safe to connect an AI assistant to Slack or Teams?
It can be, with the right design. Keep permissions tied to the source, request minimal scopes, log questions and answers, and confirm your AI provider does not train on your data.
How long does it take to build a Slack or Teams AI assistant?
A knowledge assistant in one workspace typically takes 5–10 weeks. Adding actions takes 6–10 weeks, and a company-wide build across both tools about 9–14 weeks.
Sources
- RixDigi: AI cost estimator (pricing rules, 5 November 2026)
- Slack: pricing and plan comparison
- Slack developer docs: AI apps and agents
- Slack developer docs: rate limit changes for non-Marketplace apps
- Microsoft: Microsoft 365 Copilot for enterprise
- Microsoft: Microsoft 365 Copilot for business
- Microsoft: Copilot Studio pricing
- Microsoft Learn: bots in Microsoft Teams
- Microsoft Azure: Azure Bot Service pricing