Compass is currently in early access. Features and availability are subject to change.
Compass is one of Chameleon's AI agents, designed to automatically understand how your end-users behave inside your product — and turn that understanding into actionable intelligence your team can use. No additional analytics setup, no custom event tracking, no engineering work required.
Compass sits alongside Chameleon's other AI agents: Copilot (experience creation and analysis), Ranger (account management and optimization), and Prism (1:1 personalization, coming soon).
What does Compass do?
Compass currently enables four core capabilities (with more in active development):
Read what a user did in a session
Get a plain-English summary of any user's recent session — what they clicked, what they explored, what features they engaged with, and where they ran into trouble.
Review usage across a company
Roll up individual sessions into a company-level view of adoption patterns, feature engagement, and user lifecycle stages. Understand how a customer's team is actually using your product, not just whether they've logged in.
Target users based on actual behavior
Create smarter Chameleon audiences using Compass-derived data (based on features used or not used, use case, engagement level, or lifecycle stage) without relying on manually configured events or properties passed from your backend.
Identify friction in your product
Ask Copilot about cases where users repeatedly fail, hesitate, or abandon tasks. These friction signals surface as structured, actionable findings your product team can investigate and act on, including well-written issue drafts ready to file in your product management tool (e.g. Linear).
How does Compass work?
Compass runs a continuous pipeline in three stages:
1. Session tracking
When one of your end-users interacts with your product, Compass captures their behavioral events in the background: clicks, navigation paths, feature interactions, and page visits.
2. AI processing
At the end of each session, an AI model analyzes the raw activity and produces a structured summary: what the user accomplished, where they struggled, which features they engaged with, and any friction signals detected. Each friction signal is classified by type (usability issue, broken interaction, repeated failed attempts, etc.) and severity.
3. Rollups and intelligence
Session summaries accumulate over time into a user-level profile (including lifecycle stage, feature adoption depth, and engagement patterns) and a company-level view of how a customer's team uses your product as a whole. This intelligence is then accessible through Copilot and used to power smarter audience targeting.
Where does Compass intelligence surface?
All Compass intelligence is accessible through Copilot, Chameleon's AI assistant. You can ask Copilot questions like:
"What did [user] do in their last session?"
"Which features has [company]'s team adopted?"
"Where are users running into friction in the onboarding flow?"
"Write up a product issue for the top friction pattern this week."
"Give me context on [user] — their lifecycle stage, what they've used, where they're stuck."
Beyond Copilot, Compass data powers:
Audience targeting — Build Chameleon audiences based on Compass-derived behavioral properties, without needing to manually pass custom events or attributes from your backend
Friction reporting — Review automatically identified friction candidates across your product, with structured findings and suggested issue writeups ready for your PM workflow
What data does Compass collect?
Compass collects:
Page URLs visited during the session
Click actions — button and link text, action type, ARIA labels
DOM element details (element tag, ID, and attributes)
Navigation paths and feature interaction sequences
Timestamps per event
Chameleon Profile ID and Account ID (the identifiers already configured in your Chameleon installation)
Compass does not collect:
User-entered text or form field contents
Passwords, payment information, or any sensitive input
Any data beyond click and navigation events
Whether Compass processes personally identifiable information depends on what your team sends to Chameleon via the snippet installation.
Chameleon does not independently collect names, email addresses, or other personal identifiers — but if your identify call passes those as custom properties (e.g. email, name, role), they are part of the user profile that Compass processes.
It's also worth noting that "Personal Data" is defined more broadly in some privacy frameworks (e.g. under GDPR, IP addresses may qualify as Personal Data depending on jurisdiction and context). If your organization has specific data classification requirements, we recommend reviewing your existing Chameleon identify calls to understand what profile data is in scope before enabling Compass.
Data privacy and security
Subprocessors
Compass uses Chameleon's existing subprocessor infrastructure, including OpenAI for AI processing. All subprocessors are disclosed here and operate under our existing GDPR, CCPA, and SOC 2 compliance frameworks.
AI processing
Behavioral activity (clicks, navigation, page URLs) and user profile data are sent to AI models solely to generate session summaries and user intelligence. This data is not used to train AI models. Chameleon processes behavioral data in a manner similar to an analytics provider.
Data storage
All data is stored within Chameleon's own infrastructure. Raw session events are retained in accordance with Chameleon's standard data retention policies. Processed summaries and user profiles persist as part of your Chameleon account data.
Data controller / processor relationship
Chameleon acts as the data processor; your organization remains the data controller. You control what user attributes are passed to Chameleon via your SDK integration.
Frequently asked questions
Does enabling Compass require any engineering work?
No. Compass is enabled at the account level by Chameleon. Your existing Chameleon installation is all that's required — no new code, no new events, no SDK changes.
How long until I see insights after enabling?
Typically within one week. Compass needs a meaningful volume of sessions before patterns and summaries become useful.
Will Compass affect my users' experience or product performance?
No. Session tracking runs silently in the background and has no impact on performance or the end-user experience.
Does Compass replace my existing analytics tool?
It can, but it's primary function is to focus on powering more value from Chameleon. Compass is optimized for understanding user behavior in the context of in-app experience delivery, adoption, and friction detection. It is not a general-purpose analytics replacement, so you cannot use it to create charts or funnels etc.
Do I need to set up event tracking or custom properties for Compass to work?
No. Compass generates its own behavioral signal from session activity. That said, any custom user properties you already pass via your identify call (e.g. plan, role, lifecycle stage) are used to enrich the intelligence Compass produces — so the more context your integration provides, the richer the output.
What if I have data residency or security requirements?
Compass operates within Chameleon's existing data infrastructure and DPA. Reach out to your CSM or email the security team to discuss specific requirements before enabling.
I'm not sure Compass will be useful for my use case. How should I evaluate it?
Compass delivers the most value for teams who want to understand user friction and adoption without building out custom event tracking. The best starting point is usually something like wanting the answer to: "What do we wish we knew about why users churn during onboarding?"
How do I enable Compass?
Compass is currently in early access and available by invitation. The enrollment process is:
Express interest to your CSM or email us
Chameleon enables session tracking on your account — no code changes are required
Insights begin appearing in your Chameleon UI within approximately one week as sessions accumulate
You'll be invited to a shared Slack channel for direct feedback and access to the team