About Guru Credits

Overview

Teams use Guru to search across their knowledge, get answers, and have Knowledge Agents do work on their behalf: drafting and updating content, sending announcements, running scheduled automations, and answering questions inside Slack, Microsoft Teams, and other tools through the API and MCP.

Guru credits are how we measure that AI work. Every workspace receives a shared pool of credits, and that pool is available to everyone in your organization and across every credit-based feature. Credits let your team expand AI usage as you automate more, without a separate add-on for every new capability.

What is a Guru credit?

A Guru credit represents a unit of work performed by Guru. Credits are consumed when Guru's AI does something on your behalf, such as searching connected sources, generating an answer, or when a Knowledge Agent creates or updates content. Credits are always explicitly tied to the specific work performed by Guru.

Most individual AI actions consume 1 credit. More complex workflows, like a Knowledge Agent automation or a Research session, consume credits based on the underlying actions they perform.

Work that a person does manually in Guru does not consume credits. Writing a card, browsing content, organizing folders, and editing skills by hand are all examples.

Credits are not tokens

Tokens are the unit of measurement of information passing back and forth with an LLM such as Anthropic's Claude or OpenAI's GPT models. Tokens are separate from and unrelated to Guru credits. While much of Guru's work is performed via Knowledge Agents using inference with LLMs, Guru's pricing model is based on credit consumption tied to specific work performed, without additional or separate token costs.

Why credits?

Guru is a modern enterprise AI knowledge platform. It answers questions wherever people work, keeps content accurate, drafts and publishes on a schedule, and powers other tools through the API and MCP. Credits are how that work is measured, and they are designed to make expanding AI usage simple.

• You pay for work, not potential. Credits track the AI work Guru actually performs, rather than charging for capacity that sits unused.
• One pool, shared across the organization. Credits are not assigned to individual users. A heavy user can draw on capacity that lighter users leave unused, and admins never need to reallocate credits between people.
• One model for every surface. Whether your team uses Guru in the app, in Chat, in Slack or Teams, or through the API and MCP inside tools like Claude or Cursor, the same credits apply. There is no separate rate for any channel.
• New capabilities without new add-ons. As Guru adds AI features, they draw from the same pool. You can adopt new capabilities as they ship instead of negotiating each one separately.
• Manual work stays free. Everything a person does by hand in Guru is included with their seat. Credits only apply when AI does the work.

How credit consumption is calculated

Guru calculates credit consumption based on the AI actions performed. A single action, such as one search or one generated answer, consumes 1 credit. A multi-step interaction consumes credits for each underlying action.

Credits are consumed wherever Guru's AI does work: in the Guru app, in Chat, through the API, through MCP, and in Slack and Microsoft Teams. The rate for an action is the same regardless of where it happens.

A few worked examples:

• Asking a question that performs one search and generates one answer: 2 credits
• Asking a Knowledge Agent to create a card, where it searches, drafts, and publishes: 3 credits
• A Research session that reads many documents and takes several actions: variable, based on the work performed

Features that consume credits

Search and answers

Knowledge Agent actions

Each action a Knowledge Agent performs on your behalf consumes 1 credit. This includes:

• Creating, updating, moving, or archiving a card
• Listing cards from a collection
• Creating, updating, getting, listing, publishing, or discarding a draft
• Listing, creating, or updating a collection
• Listing, creating, updating, moving, or deleting a folder
• Viewing or retrieving a card or source during a chat
• Sending an announcement
• Posting to Slack, Teams, or email
• Creating a ticket

Automations

Scheduled and complex Knowledge Agent automations are charged for the actions they perform, exactly as if the same steps were run manually in Chat. There is no flat per-automation fee. Pausing an automation stops future credit consumption. Failed runs generally do not consume credits, though a partial amount may be consumed if processing began before the failure.

Operator Mode

When an agent creates, edits, or reads a skill, updates or reads the system prompt, or inspects an answer in Operator Mode, each action consumes 1 credit. Editing skills manually in the Skills interface is free.

Quality Agent

Content used in answers is not re-reviewed within seven days.

Guardrails

Guardrails, such as Jailbreak Detection, are off by default. When a guardrail is enabled on a Knowledge Agent, each user input it evaluates adds 1 credit to that request, across Chat, Search, API, MCP, and Slack. For example, a request with one guardrail, one search, and one answer consumes 3 credits.

API and MCP

This applies whether you use MCP from within Guru or to make Knowledge Agents available inside tools like Claude, ChatGPT, Gemini, or Cursor.

Slack and Microsoft Teams

Slack and Teams use the same rates as the rest of Guru. A search is 1 credit, a generated answer is 1 credit, and each Knowledge Agent action is 1 credit. There is no separate rate for these channels.

Data Loss Prevention (DLP) masking

Connecting and syncing sources does not consume credits on its own. If DLP masking is configured for a source, each masked record processed during sync (new or updated) counts as a DLP event, at 1 credit per 10 events. Sources without DLP masking incur no charge. A re-index of a masked source can generate DLP charges.

What does not consume credits

• Manually writing or editing cards
• Manually adding comments
• Manually creating Pages
• Viewing, browsing, and reading content in the Guru app
• Manually organizing collections and folders
• Manually editing skills in the Skills interface
• Paused automations
• Content migration during implementation
• Connecting and syncing sources (without DLP masking)
• Source administration, such as editing templates, fields, and facets, or inspecting sync details
• Dataset metadata calls, such as listing datasets or reading a schema

The simple rule: manual work is included with your seats. The same work performed by an AI agent consumes credits.

Monitoring credit usage

Guru tracks credit consumption by event and summarizes it at the team level. Admins receive notifications as the workspace approaches its allocation and continue to receive notifications as usage increases, so there is always time to plan ahead.

Managing credit consumption

Admins have several ways to control how credits are used:

• Pause automations that are no longer needed
• Use Knowledge Agent permissions to control who can create and manage automations (Manage Own Automations and Manage All Automations)
• Remove a user's access to a Knowledge Agent, which automatically pauses that user's automations on it
• Disable optional guardrails when they are not needed
• Manage which users and agents have access to Knowledge Agents and sources

Need more credits?

If your team's AI usage grows beyond your current allocation, you can add credits at any time. Self-serve customers can add credits through Guru Billing. Annual customers can work with their Guru account team. For details on allocations and pricing, see your agreement or contact your account team.

Frequently asked questions

Is a regular Guru search really billable?
Yes. Search in Guru is an AI-powered search across your Guru content and connected sources, with your permissions applied, so it consumes 1 credit. Browsing and reading content directly in the app does not.

Do credits get charged per user?
No. Credits are a shared organizational pool. Individual users do not have their own balances, and admins do not need to allocate credits to people.

Does using Guru in Slack or Teams cost more?
No. Slack and Teams use the same per-action rates as the rest of Guru.

Does viewing an answer someone else already generated cost credits?
No. Once Guru has generated an answer or a Knowledge Agent has created content, reading it in the app does not consume additional credits.

Can I control or limit credit usage?
Yes. Admins can pause automations, restrict who can create them, disable optional guardrails, and manage access to Knowledge Agents and sources. See "Managing credit consumption" above.

---