Keypup

Field ID: <code>surface_ref</code>

Description: The normalised surface the usage happened on, using a closed vocabulary shared across providers so they can be compared. Possible values are:

<code>IDE_COMPLETION</code> inline code completion.

<code>IDE_AGENT</code> agent mode in the IDE.

<code>CLOUD_AGENT</code> a remote or cloud agent run.

<code>CODE_REVIEW</code> pull request review.

<code>DESKTOP_APP</code> desktop application.

<code>AUTOMATION</code> a third-party automation.

<code>UNATTRIBUTED</code> reported by the provider without a surface type.

- <code>IDE_COMPLETION</code> inline code completion.
- <code>IDE_CHAT</code> chat in the IDE.
- <code>IDE_AGENT</code> agent mode in the IDE.
- <code>CLI</code> command line.
- <code>CLOUD_AGENT</code> a remote or cloud agent run.
- <code>CODE_REVIEW</code> pull request review.
- <code>WEB_CHAT</code> web chat.
- <code>DESKTOP_APP</code> desktop application.
- <code>AUTOMATION</code> a third-party automation.
- <code>UNATTRIBUTED</code> reported by the provider without a surface type.

Note: Empty on <code>TOTAL</code> and <code>CLIENT</code> records, which cover several surfaces at once.

Transformation logic: Each provider's own surface vocabulary is mapped onto this shared list. The provider's raw value is always preserved in <code>surface_name</code>, which is the other half of the same field.

The <code>Surface Ref</code> field is what makes AI usage comparable across providers. It answers where the AI was used, in the same vocabulary whatever tool produced the record.

Where AI is actually used: Filter to <code>breakdown = SURFACE</code> and group by <code>surface_ref</code> to see the split between completions, chat, agents, the command line and code review.

Measuring the shift to agents: Tracking <code>IDE_AGENT</code> and <code>CLOUD_AGENT</code> as a share of all interactions over time is the clearest signal of a team moving from autocomplete to agentic work.

Comparing tools fairly: Because the vocabulary is shared, chat usage in Copilot and chat usage in Cursor land on the same value and can be charted side by side.

Interpreting UNATTRIBUTED: This is not an error. The provider reported usage without saying where it came from, and it is kept rather than dropped.

When it is too coarse: Several provider surfaces map onto one value. Use <code>surface_name</code> when you need the provider's own distinction.

- Where AI is actually used: Filter to <code>breakdown = SURFACE</code> and group by <code>surface_ref</code> to see the split between completions, chat, agents, the command line and code review.
- Measuring the shift to agents: Tracking <code>IDE_AGENT</code> and <code>CLOUD_AGENT</code> as a share of all interactions over time is the clearest signal of a team moving from autocomplete to agentic work.
- Comparing tools fairly: Because the vocabulary is shared, chat usage in Copilot and chat usage in Cursor land on the same value and can be charted side by side.
- Interpreting UNATTRIBUTED: This is not an error. The provider reported usage without saying where it came from, and it is kept rather than dropped.
- When it is too coarse: Several provider surfaces map onto one value. Use <code>surface_name</code> when you need the provider's own distinction.

AI Usage > Surface Ref

No tickets or conversations found

Try selecting a different creation date range.

Tickets and conversations you've started through the messenger will appear here.

No tickets or conversations created by you

Try using different keywords or checking for typos.

Try selecting a different update date range.

Find answers and get help from Intercom Support and Community Experts

Conversations you've started through the messenger will appear here.

No conversations created by you

No conversations found

AI title

Title

This site employs cookies and other technologies that we and our third party vendors use to monitor and record personal information about you and your interactions with the site (including content viewed, cursor movements, screen recordings, and chat contents) for the purposes described in our Cookie Policy. By continuing to visit our site, you agree to our {websiteTermsLink}, {privacyPolicyLink} and {cookiePolicyLink}.

This site uses cookies and similar technologies ("cookies") as strictly necessary for site operation. We and our partners also would like to set additional cookies to enable site performance analytics, functionality, advertising and social media features. See our {cookiePolicyLink} for details. You can change your cookie preferences in our Cookie Settings.

We use cookies to make our site work and also for analytics and advertising purposes. You can enable or disable optional cookies as desired. See our {cookiePolicyLink} for more details.

Advertising cookies are set by our advertising partners to collect information about your use of the site, our communications, and other online services over time and with different browsers and devices. They use this information to show you ads online that they think will interest you and measure the ads' performance. Social media cookies are set by social media platforms to enable you to share content on those platforms, and are capable of tracking information about your activity across other online services for use as described in their privacy policies.

These cookies enable the website to provide enhanced functionality and personalisation. They may be set by us or by third party providers whose services we have added to our pages. If you do not allow these cookies then some or all of these services may not function properly.

These cookies are necessary for the website to function and cannot be switched off in our systems.

These cookies allow us to count visits and traffic sources so we can measure and improve the performance of our site. They help us to know which pages are the most and least popular and see how visitors move around the site.

You have the right to opt out of the sale of your personal information. See our {cookiePolicyLink} for more details about how we use your data.

Your Privacy Choices

We use cookies to enhance your experience. You can customize your cookie preferences below. See our {cookiePolicyLink} for more details.

Cookie Settings

Empty Help Center

Uh oh. That page doesn’t exist.

Home

Search results

Disappointed

Neutral

Smiley

Thinking...

Searching through sources...

Analyzing...

Tickets submitted through the messenger or by a support agent in your conversation will appear here.

No tickets created by you

No tickets found

Conversations and tickets in {topic}

Conversations and tickets in {topic} // {subtopic}

Volume, creators and topics for your company.

Insights

Top creators

AI topics

Track the progress of all tickets and conversations related to your company.

Customer Portal.

Track the progress of all tickets related to your company.

Tickets portal.

{assigneeName} needs more information from you

Tickets

No access to tickets portal

Search

Skip the search. Describe what you’re trying to do and get a grounded answer.

Browse articles and guides in this topic.

Create your workspace and invite the people you work with.

Set up your account

Build your first workflow and watch it run end to end.

Ship something

Connect the tools your team already uses to bring data in.

Connect your stack

A quick lap of the product so you know where everything lives.

{appName} in 60 seconds

This prototype demo is aware that you are reading “{title}”. It can point you back to the current article, but a production article-aware Fin contract is still follow-up work.

Prototype article-aware answer

Because you already asked about invitations, this prototype demo can add the follow-up: use the role selector before sending each invite to limit access to the teammate responsibilities.

Prototype role-limiting follow-up

This prototype demo suggests starting with your workspace profile, inviting the team members who need access, and then connecting the integrations you use.

Prototype getting-started answer

In this prototype demo, integrations are configured from workspace settings. Choose the integration, connect the account, and review the enabled permissions.

Prototype integrations answer

In this prototype demo, invite teammates from workspace settings, then choose the access each teammate needs before sending the invitation.

Prototype invite-team answer

This prototype demo does not have a scripted answer for that question yet. Try asking how to invite your team or open a source article. No production Fin API was called.

Prototype answer unavailable

I’ve got this page open. Ask me anything about it and I’ll answer from the docs.

GitHub Copilot	Calculated: mapped from the `feature` dimension, plus dedicated values for the agent app, code review and the cloud agent
Anthropic Claude (coming soon)	Calculated: mapped from `terminal_type` and `is_remote`. Remote work is `CLOUD_AGENT`, known IDE hosts are `IDE_AGENT`, everything else is `CLI`
Cursor (coming soon)	Calculated: mapped from the counter a record comes from, or from the event `kind`
OpenAI Platform (coming soon)	Calculated: `AUTOMATION` by default, since this is API traffic
OpenAI Codex (coming soon)	Calculated: mapped from `client_id`. Web is `CLOUD_AGENT`, CLI clients are `CLI`, IDE clients are `IDE_AGENT`

AI Usage > Surface Ref

App Mapping

Reporting Use Cases