Doti implementation cost: Slack plans, enterprise search and the work
Doti is no longer sold on its own. The cost question is now what Slack plan carries enterprise search, and what it takes to connect your sources safely.
Platform & ArchitectureMost people who ask what Doti costs to implement are really asking a different question: what does it cost to let staff ask a question in Slack and get an answer drawn from the systems the company already runs? Doti was an agentic enterprise search company. Salesforce bought it, and the acquisition record is short: "Salesforce completed its acquisition of Doti on December 1, 2025." The same announcement said Slack would be the conversational interface for the search capability the Doti team would help build.
That shapes the cost conversation in a useful way. You cannot order Doti, and nobody can sell you a Doti implementation in the old sense. What you can buy is a Slack subscription that includes enterprise search, and what you pay a partner for is the work of connecting sources, getting permissions right, proving the answers, and changing habits. This page prices both halves with the figures Slack publishes, and is plain about which numbers are not public.
If you want the background to the deal, read what happened to Doti. If you want to know how search should behave once it is live, the Doti and Slack enterprise search guide covers that. This page stays on money.
What you pay Slack, as at October 2026
Slack publishes its prices on its pricing page. As at October 2026 the page lists four plans, priced in US dollars per user per month. The important line for anyone thinking about Doti is in the Enterprise+ column, which describes enterprise search like this: "Search across connected apps, databases and systems with enterprise search."
| Plan | Price (USD per user per month) | Billing | Search capability listed |
|---|---|---|---|
| Free | $0 | Free forever | AI search over Slack content, 90 days of history |
| Pro | $8.75 monthly, or $7.25 | Monthly or annually | Basic AI, including AI search |
| Business+ | $18 monthly, or $15 | Monthly or annually | Advanced AI, including AI search |
| Enterprise+ | $45 | Annually, through sales | Enterprise search across connected apps |
Three things in that table matter more than the headline numbers.
First, enterprise search is listed only against Enterprise+. Slack's own help article on setting it up says the feature is available on the Enterprise+ subscription, and on Enterprise Grid with the legacy Slack AI add-on. If you are on Business+ today, the price of the capability that most resembles Doti is the step up to Enterprise+, multiplied across every seat in the organisation, not just the people who will search most.
Second, Enterprise+ is a sales conversation. The list price is printed, but the plan is bought through Slack's sales team, so discounting, term length and seat ramps are negotiated. Treat $45 as the starting point for a budget, not the final invoice.
Third, you may already own it. Salesforce's Sales Cloud pricing page lists Slack Business+ inside the Core edition and Slack Enterprise+ inside the Max edition, which is $550 USD per user per month billed annually. A sales team on Max may already hold the Slack plan that carries enterprise search, at least for those seats. Check the order form before anyone budgets for Slack twice.
What an enterprise search implementation is made of
Slack's enterprise search does not copy files into Slack. Org owners or org admins switch it on, add data sources, and then members connect their own accounts so their results include content from those places. Slack's help article lists the supported sources as Asana, Box, Confluence Cloud, Dropbox, GitHub, Google Drive, Google Mail, Jira Cloud, Microsoft Outlook, Microsoft Teams, OneDrive and SharePoint, Salesforce, and Seismic. It also notes that custom connectors can be built against internal sources with the Slack APIs. That list is the real scope document for most projects.
A sensible implementation breaks into six work packages.
Discovery and source inventory. Which of the supported sources do you actually use, who owns each one, and where does the knowledge staff ask about really live? This is a short piece of work, but it decides everything else. A team that keeps its policies in Confluence and its contracts in SharePoint needs two connectors done well, not thirteen done quickly.
Permission review. The help article is clear on the rule that makes this safe: "Search results, AI answers and Slackbot responses will only include source content that the person searching has permission to access." That is a strength and a warning. If a SharePoint site is shared with everyone by accident, enterprise search will faithfully surface it to everyone. The cost of an implementation is often set less by Slack than by how much sharing clean-up the source systems need first.
Connector configuration. Some sources need extra steps. Confluence Cloud asks for the subdomain, up to five of them. GitHub needs the organisation name and its own enterprise search app. Box has its own admin settings. The Salesforce source appears automatically once Slack and Salesforce are connected, and is managed from the Salesforce section of the admin dashboard. None of this is hard, but every source has its own administrator, and getting time from six administrators is where calendars slip.
Custom connectors. If key knowledge sits in an internal wiki, a document management system, or a database that is not on the list, someone has to build a connector with the Slack APIs. This is the one package that can turn a configuration project into a development project, and it should be quoted separately.
Testing with real questions. Write down the forty or fifty questions staff ask most, who is allowed to see each answer, and what a good answer looks like. Run them as different users. Run them again after every connector change. This is the only reliable evidence that search is helping rather than leaking.
Adoption and change. Search only pays back if people stop messaging colleagues for links. That takes channel guidance, a short training session per team, and a few weeks of champions answering "did you try asking Slack?" It is cheap compared with licences, and it is the package most often cut.
What drives the effort up or down
The figures you will be quoted vary with a handful of factors that are specific to search, rather than to Slack as a whole.
The number of sources is the obvious one, but the type matters more. Google Drive and OneDrive are usually quick because the sharing model is familiar. Mail sources raise privacy questions that take governance time even when the configuration is a few clicks. Code in GitHub raises a different question: do you want repository content appearing in answers at all?
Sharing hygiene is the hidden driver. An organisation with years of "anyone with the link" sharing will spend more on clean-up than on Slack configuration. That cost belongs in the business case, because it is real work, often by people outside the Slack project.
Identity is the third. Enterprise+ includes multiple SAML configurations and SCIM user management, and members connect their individual accounts to each source. Where identity providers, staff numbers and account names do not line up across systems, someone has to reconcile them before results are trustworthy.
The last driver is whether you separate traditional search from AI answers. The help article notes you can choose whether a source appears in traditional search results, AI answers, or both. A cautious rollout can start with links only and add AI answers per source later, which spreads the testing effort across phases.
How to get comparable quotes
When you ask partners to quote, give them the same brief, and make the brief concrete. A good request lists each source you intend to connect, the owner of each, the approximate number of users and spaces involved, whether any custom connector is needed, the identity provider, and the test questions you expect to pass before go-live.
Then ask each partner the same questions. Which sources do you assume are connected in phase one, and which are excluded? Who does the permission clean-up in the source systems, you or us? How many test questions are in scope, and how many test users? Is building a custom connector priced separately, and on what basis? What does hypercare cover, and for how long? What training is included, and for whom?
Comparing quotes that answer those questions is straightforward. Comparing quotes that only say "Slack enterprise search rollout" is impossible, because each partner has quietly assumed a different scope. If one quote is much lower, the usual reason is that it assumes clean sharing and no custom connectors.
Costs that sit outside the services quote
Several costs belong in the budget even though no partner will put them on the statement of work.
The plan step-up. Moving from Business+ to Enterprise+ applies to the organisation's seats, not to a pilot group, so model the full uplift before the pilot starts. Ask Slack how a pilot can be structured commercially before you sign anything.
Licence overlap. If you are replacing another search or knowledge tool, you will pay for both while the new search proves itself. Agree an exit date for the old tool before the project starts.
Source system licences. Some sources may need higher tiers or admin add-ons to expose the controls you need. Check each one with its owner.
Administrator time. Each source owner will spend days, not hours, on access review and connector set-up. That time is real even when it is not invoiced.
Ongoing ownership. Someone has to own the connector list, re-run the test questions after changes, and respond when a result appears that should not. Budget a slice of an administrator's role for this permanently.
Success plan and support. If you are buying Enterprise+ alongside Sales Cloud, check which support level each contract carries, because the two products may have different terms.
A sensible first phase
The cheapest serious first phase is narrow. Pick two sources where the knowledge is valuable and the sharing is already tidy, connect them, and test with a single department. Keep AI answers off for any source that has not passed the permission review. Measure how often people find what they need without asking a colleague.
That first phase gives you real evidence on three things you cannot know in advance: how much permission clean-up the remaining sources need, how much staff use search once it is there, and whether the jump to Enterprise+ is worth it for everyone or only for some. Keep a simple log during the pilot: the question asked, the source that answered it, and whether the answer was right and permitted. Twenty minutes a week with a department lead reading that log is the cheapest quality control the project will ever have, and it becomes the acceptance test for phase two. With that evidence, a second quote for the remaining sources will be tighter and more honest.
If you are weighing whether to build this yourself or bring in help, the guide to hiring a partner for Doti-style search sets out what to look for. The Doti licence cost page covers the subscription side in more depth, and the integration guide explains how the Salesforce source behaves once Slack and Salesforce are connected. If you are deciding whether to move from an existing search tool, see should I migrate from Doti.
The shortest honest summary for finance is this: there is no Doti invoice. There is a Slack plan, a set of sources, and a clean-up job, and each of those can be priced once you have named them.



