The short answer
Every major workplace platform now bundles AI search, but each one gates the part that searches across other apps, and the gates are not alike. Slack puts enterprise search on Enterprise+ only and publishes no price for that plan. Microsoft sells it as a paid Microsoft 365 Copilot add-on at $30 per user per month. Google sells Gemini Enterprise as a separate Google Cloud product with a published seat fee. Atlassian includes Rovo in eligible paid Cloud plans but meters usage with credits. The common thread is that bundled search is priced across your whole organization, while a dedicated tool is priced for the people who actually have the problem. That arithmetic, not feature parity, usually decides the comparison.
Last updated: August 2026.
If you are being asked why the company should buy a search product when Slack, Microsoft, Google and Atlassian all ship AI search already, this is the argument you need to be able to make honestly. The bundled options are real. Some of them are good. But the version that is included in the plan you are already paying for is almost never the version that searches across your other systems, and the difference between those two things is where every one of these evaluations actually turns.
What follows is what each platform gates it behind, checked against the vendors' own documentation, with the dates we checked. Prices in this category change quarterly, so treat every figure below as a starting point for your own verification rather than a fact with an indefinite shelf life.
What "bundled AI search" actually means
Almost every workplace platform now ships two different things under similar names, and the naming is doing a lot of work.
The first is AI search inside the platform: better retrieval, summaries and natural-language querying over the content that platform already holds. This is usually included in a mid-tier plan, it is genuinely useful, and it is not enterprise search. It only ever sees one system.
The second is search that reaches out into other systems, so a question asked in one place can be answered from content held somewhere else. This is the capability people mean when they say enterprise search, and it is the one behind the gate. Every platform below draws that line somewhere, and knowing exactly where saves a lot of wasted evaluation time.
What each platform gates enterprise search behind
| Platform | What is gated | Published price | Shape of the cost |
|---|---|---|---|
| Slack | Enterprise search, meaning search across connected apps, is on Enterprise+ only. Business+ does not include it. Disabled by default; only Org Owners and Org Admins can enable it | None. Slack shows "Contact sales for pricing" for Enterprise+ | Whole-organization tier migration |
| Microsoft | Microsoft 365 Copilot is a paid add-on. Copilot Chat is available at no cost but is grounded on the web, not on your tenant's content | $30 per user per month, annual. Microsoft 365 E7 bundles Copilot at $99 per user per month | Per-user add-on, on top of an existing license |
| Gemini Enterprise, formerly Agentspace, is a separate Google Cloud product. It is not a Workspace add-on | Google's own wording: "a monthly seat fee of $30 is billed at approximately $1 per day" | Seat subscription plus consumption | |
| Atlassian | Rovo is included in eligible paid Atlassian Cloud plans at no incremental license cost | No separate license fee. Chat and Agent requests draw on monthly credit allowances | Included, then metered by credits |
Slack details read from slack.com/pricing and Slack's enterprise search help documentation on August 16, 2026. Microsoft figures verified July 23, 2026. Google figures verified July 22, 2026. Verify anything you are about to budget for directly with the vendor.
Read down the last column and the pattern is hard to miss. Three of the four price this across your whole organization or your whole license base. None of them price it for the subset of people who actually cannot find things. That is the structural difference between bundled and dedicated search, and it is worth more attention than any feature checklist.
Why bundled search is priced per organization and dedicated search is priced per person
Bundled AI search exists to make the platform stickier. Its commercial job is to move accounts up a tier or to attach an add-on across a license base, which means the pricing naturally attaches to the whole estate. That is a perfectly rational thing for those vendors to do, and it is not a criticism. It just means the cost does not scale with the size of your problem.
Work an example. A 900-person company has a genuine cross-app search problem affecting roughly 60 people: support agents, a handful of solutions engineers, the finance team and every new hire during their first quarter. Moving the whole company to a top-tier plan prices 900 seats to serve 60. Even at a higher per-seat rate, a tool bought for the 60 is usually the cheaper line item, and it stays cheaper as the company grows because the denominator never changes.
The counter-case is real and worth stating. If you were already migrating to that top tier for compliance, data residency, administrative controls or eDiscovery, the search capability arrives as a benefit of a decision you had already made. In that situation, paying separately for a second product genuinely is waste. The question is never "is bundled search good", it is "was I going to buy this tier anyway".
When bundled AI search is the right answer
Three conditions, and they need to hold together rather than individually.
Your company genuinely runs on one platform. Not mostly. If a meaningful share of your people live in a different client, an answer that is only reachable from Slack or only reachable from Microsoft 365 solves half the problem at full price.
Your sources are on the supported list. Coverage is the whole game. Slack documents twelve connectors, which is a respectable list but a finite one, and it carries caps that catch out larger companies: five GitHub organizations, five Confluence Cloud subdomains, five Jira Cloud subdomains. Any group that grew by acquisition should count its Atlassian sites before assuming full coverage. We go through the full list and the caps on Slack enterprise search.
You were heading to that tier anyway. If the plan upgrade is already funded, take the search.
When a dedicated enterprise search tool is cheaper
The mirror image, in the order the situations usually appear.
When the answers live in systems the platform does not connect to. Help desks, HR systems, homegrown internal tools, data warehouses and wikis that are not Confluence are absent from most bundled connector lists. A search tool that cannot see the answer is not a search tool, however good its model is.
When only a subset of the company has the problem, for the arithmetic above.
When the people with the worst search problem are not in the platform at all. This is the one that gets missed most often. The roles that lose the most time to unfindable answers are usually support agents working a ticket queue, new hires who do not yet know your vocabulary, and finance staff checking a policy that changed. Support agents in particular spend their day inside a help desk, and any team running serious support operations will tell you that asking those agents to context-switch into a chat client to search is exactly the friction the tool was supposed to remove.
When you need the answer in more than one place. A dedicated index can serve a browser, a help desk, an onboarding portal and a chat client from the same source of truth. A bundled feature serves its own surface.
Does bundled AI search respect permissions?
Generally yes, and this is not a productive place to attack the bundled options. Slack's documentation states that search results, AI answers and Slackbot responses only include source content the person searching has permission to access, and because its search is federated it evaluates access on each query rather than from a synced snapshot. Microsoft's SharePoint agents honor permissions too. These vendors have the strongest possible incentive to get access control right in their own products.
The question worth asking any vendor, bundled or dedicated, is not whether permissions are supported but when they are checked. Query-time enforcement means a revoked permission takes effect on the next search. A scheduled access sync means there is a window during which revoked access still returns results, and the length of that window is a security property you are buying whether or not anyone puts it in the contract. Google's identity sync for Gemini Enterprise, for example, runs anywhere from every 30 minutes to every seven days depending on configuration, which defines exactly that window. We cover how to interrogate this properly on permission-aware AI search.
The three questions to ask before you upgrade a plan for search
How many people actually have this problem? Count them before you price anything. Most teams guess high, then discover the real number is a few departments and every new hire. That number is the denominator for the whole decision.
Which systems hold the answers people cannot find? Write the list from real examples, not from your app inventory. Then check it against the platform's published connector list, including any per-connector caps. If two of the top five systems are missing, the bundled option is out regardless of price.
Where will people be standing when they need the answer? If the honest answer is "in a help desk" or "in a browser" or "on a phone in the field", a search feature that lives inside a chat client will not get used, and unused software is the most expensive kind.
Can you run both?
Yes, and some companies deliberately do. They keep the platform's built-in search for quick in-flow lookups, where being right there in the conversation is most of the value, and run a dedicated index for the deeper cross-system questions and for the teams who work outside that platform. That is not indecision, it is two different jobs.
It only makes sense if you have counted. Running both means paying twice, and it is justified when the two populations are genuinely distinct: everyone gets the in-platform convenience, and the roles that live on hard questions get the tool that can actually answer them. If the populations overlap almost entirely, pick one.
Where to go next
If you are at the shortlist stage, the sixteen-product table on best enterprise search software compares coverage, citations and published price side by side, including the bundled options. For the platform-specific detail, we maintain pages on Slack enterprise search, Microsoft Copilot pricing, Gemini Enterprise pricing and SharePoint agents, each with the figures read from the vendor's own site and the date we read them.
And if the underlying problem is simply that documented answers are unfindable rather than that you need a particular vendor, start with internal search, which is the broader version of this question and the one most companies are actually trying to solve.