Meet Kai
Your AI-powered colleague for understanding the Kaptio platform. Ask questions, explore configurations, and create documentation.
K What is Kai?
Kai is an AI assistant you interact with primarily through Slack. Unlike generic AI tools, Kai has deep context about Kaptio's platform, customer configurations, and the Outcome Model that structures our enablement content. Because Kai can access multiple data sources and execute code, there's a human-in-the-loop for anything that creates or modifies content.
Slack
Primary Interface
Tag @Kai in any channel to ask questions or request content. This is how most people interact with Kai.
Cursor
Execution EnvironmentKai runs in Cursor with workspace rules and access to data sources. Different operators have different access levels—see below.
🔐 How It Works: The Trust Model
Kai operates in two modes depending on what you ask. Looking up existing content is semi-autonomous—Kai can find and share it directly. Creating new content requires a plan that Ragnar or Ivan reviews before execution.
"What docs exist about pricing?" — Kai searches enablement.kaptio.com
New content, customer queries, or searches in Notion/Drive/Jira
🔑 What Kai Can Access
Kai's power comes from its access to Kaptio's entire ecosystem. With great power comes great responsibility—which is why we have access tiers.
Why restricted: Access to sensitive customer and commercial data requires careful handling.
- ✓ All code repositories
- ✓ Documentation & guides
- ✓ Internal tooling & configs
- ✗ No customer data
- ✗ No production GCP
Expanding soon: As workflows mature, more team members will get codebase-level access.
Core Capabilities
Research
Search code, docs, Notion, Google Drive
Create
Guides, showcases, configs, playbooks
Connect
Query Salesforce orgs via Core Health
Answer
Questions about configs, features, workflows
The Outcome Model
Kai doesn't just answer questions randomly—every piece of content aligns with a structured information architecture. This ensures consistency and makes content discoverable.
The Artifact Stack
Why this matters: When you ask Kai a question, it doesn't just give you an answer—it maps the response to this structure, ensuring content is connected and findable.
See It In Action
Here's what happens when you tag @Kai in Slack. The workflow differs based on whether you're asking about existing content or requesting something new.
@Kai What documentation exists about the booking wizard?
Found 2 resources about the booking wizard:
- 📖 Booking Wizard Configuration Guide
- 🎬 Booking Wizard Showcase
Scope: Semi-autonomous lookups search the Outcome Hub only. For broader searches (Notion, Drive, Jira, etc.), request via the create flow.
For Product Managers
How Karen, Rowan, Sigrún, and the PM team can use Kai
As a PM, you're constantly context-switching between customers, features, and documentation. Kai helps you get answers faster and create content that aligns with the Outcome Model.
Research Customer Configurations
"How does Tauck configure their package pricing?" — Kai can query customer Salesforce orgs via the Core Health API to show you real configurations.
"Kai, show me how Tauck's price rules are set up for their rail packages"
Understand How Features Work
Need to explain a feature to a customer? Kai can walk through the data model, field relationships, and business logic.
"Kai, explain how inventory reconciliation works between bookings and allocations"
Find Existing Documentation
Before creating new content, check what exists. Kai searches guides, outcomes, KB articles, and even Notion pages.
"Kai, what documentation exists about the booking wizard?"
Create Pitch Content
Need a showcase for a new feature? Kai can draft it using the standard template, complete with Kaptio branding.
"Kai, create a showcase for the new charter booking feature"
Frequently Asked Questions
How does Kai know where to categorize new content? ▼
When Kai proposes a plan for new content, the plan includes the categorization—which Layer, Solution, Outcome, and Feature the content maps to.
The operators (Ragnar and Ivan) use the Feature Registry as a guide. This registry documents how all our features and outcomes are organized, ensuring content lands in the right place and connects to related materials.
Example: A guide about "Charter Bookings" would be categorized under: Product Experience Layer → Quest Solution → Booking Management Outcome → Charter Bookings Feature
How long does it take to get new content created? ▼
It depends on the queue and complexity. Here's a rough guide:
- • Lookup existing content: Seconds to minutes (no approval needed)
- • Simple guide or config: Same day to next day
- • Complex showcase or playbook: 1-3 days depending on research needed
Kai acknowledges your request immediately and keeps you updated on progress.
Can I ask Kai about customer-specific configurations? ▼
Yes! Ragnar's Kai has access to all customer Salesforce orgs via Core Health. You can ask questions like:
- "How does Tauck configure their price rules?"
- "What custom fields does DuVine have on Itinerary?"
- "Show me Rocky's inventory allocation settings"
Note: Customer data queries go through Ragnar's Kai (full production access), not Ivan's.
What if Kai gets something wrong? ▼
We have multiple safeguards:
- 1. Plan review: Every content creation plan is reviewed by a human before execution
- 2. Jury review: Before publishing, we can trigger a "jury" that fact-checks the content
- 3. Iteration: You can always ask Kai to revise—just reply in the thread with feedback
If you spot an error in published content, tag @Kai with "quick fix: [description]" for fast corrections.
Can I edit what Kai creates? ▼
Iteration is encouraged. There are a few ways to refine Kai's output:
- • Before publishing: Review the draft and ask Kai to adjust tone, add sections, or restructure
- • After publishing: Use "quick fix: [description]" for small changes, or request a larger revision
- • Direct editing: For those with Cursor access, you can edit the content files directly
Will more people get Cursor access in the future? ▼
Yes, that's the plan. We're starting with restricted access to:
- • Refine the workflows and catch edge cases
- • Build up the rule library (Cursor rules) that guide Kai's behavior
- • Establish quality standards for different content types
As these mature, we'll expand codebase-level access (like Ivan has) to more team members. Full production access will remain restricted due to customer data sensitivity.
How is content quality ensured? EVOLVING ▼
We're implementing a multi-layer review process:
- 1. Plan review: Operator (Ragnar/Ivan) reviews and can provide feedback—Kai iterates until the plan is right
- 2. Jury review: Before publishing, a skeptical review can be triggered to fact-check content
- 3. Team review (coming soon): Content will be shared with subject matter experts for parallel review and approval
In progress: We're building a formal review workflow where team members can approve content in parallel to sharing. This ensures accuracy without slowing down delivery.
How It Works: The Enablement Pipeline
Kai follows a structured workflow defined by Cursor rules. This ensures consistent, high-quality content that's fact-checked before publishing.
📜 Powered by Cursor Rules
Kai's behavior is defined by .cursor/rules/ files in the kaptio-gallery repository. These rules give Kai context about:
The 9-phase workflow from signal to publish
Content types, schemas, and file locations
Fact-checking before any publish
Trigger Phrases
For Cursor Operators These phrases activate specific workflows when working with Kai in Cursor:
| Say this... | To do this... |
|---|---|
| "kai work queue" | Display pending @Kai mentions from Slack |
| "work on signal 1" | Start researching a specific signal |
| "acknowledge signal 1" | Send queue notification to Slack thread |
| "bring in the jury" | Trigger skeptical review before publish |
| "publish" / "ship it" | Commit, deploy, and notify Slack |
| "quick fix: [desc]" | Fast iteration on feedback (higher trust) |
Getting Started
Start in Slack
The easiest way to work with Kai
- 1 Go to any project channel, support channel, or DM
- 2 Tag
@Kaiwith your question or request - 3 Kai will acknowledge, look up relevant content, or queue for plan creation
Examples: "What docs exist about inventory?" (lookup) • "Can you create a guide for X?" (plan needed) • "How does Tauck configure pricing?" (research)
For Kai Operators
Executing plans in Cursor
Full production access — customer data, GCP, all APIs
Codebase access — repos, docs, internal tooling
- 1. Open the kaptio-gallery workspace in Cursor
- 2. Say "kai work queue" to see pending requests
- 3. Work through signals: research → plan → review → execute
As workflows mature, more team members will get codebase-level access like Ivan.
💡 Pro Tips
- • Be specific: "How does Tauck configure X?" beats "Tell me about pricing"
- • Provide context: Mention the customer, feature, or use case
- • Patience on creation: New content requires human review—Kai will acknowledge and queue
Coming Soon
This guide will expand with role-specific sections: