ChartPull uses artificial intelligence to turn your raw directory data into actionable insights about your organisation. Here is how it works, what each tier gets, and how we handle your data.
Every AI feature in ChartPull is powered by Claude Haiku from Anthropic. When you ask a question, generate a report, or request an insight, ChartPull sends a carefully structured prompt along with relevant employee data to the Claude API. The AI processes it and returns a structured response that ChartPull renders in the interface.
There is no training on your data. Each request is a one-time API call. Anthropic does not store or learn from the data sent through their API. Your employee information is used only to answer the specific question you asked, then discarded.
13 AI endpoints
ChartPull has 13 distinct AI-powered endpoints, each designed for a specific task: search, insights, person insights, data quality scoring, reports, predictions, weekly digest, audit analysis, dashboard narrative, onboarding playbook, natural language API, Slack bot responses, and org timeline analysis.All AI features — search, insights, reports, predictions, and the weekly digest — are included in the AI Features add-on, available as a per-employee annual subscription alongside your base license.
The Starter plan includes no AI features during normal use. However, every new workspace gets a 14-day trial that unlocks all AI features so you can evaluate them before choosing a plan.
The AI add-on unlocks the AI features you will use every day:
The AI add-on also includes administrative and strategic AI features:
Each AI endpoint has its own rate limit to ensure fair usage and keep response times fast. Limits are applied per API key or per user session.
| Endpoint | Limit | Tier |
|---|---|---|
| AI Search | 20 requests / 60 seconds | AI |
| Organisation Insights | 10 / 60s | AI |
| Person Insights | 10 / 60s | AI |
| Data Quality | 5 / 60s | AI |
| Executive Report | 3 / hour | AI |
| Predictions | 5 / 60s | AI |
| Weekly Digest | 1 / day | AI |
| Audit Analysis | 3 / hour | AI |
| Dashboard Narrative | 5 / 60s | AI |
| Natural Language API | 20 / 60s | Integrations |
| Slack Bot | 20 / 60s | Integrations |
Rate limit responses
If you exceed a rate limit, ChartPull returns a429 Too Many Requests response with a Retry-After header telling you how many seconds to wait. The UI will show a friendly message asking you to slow down.ChartPull takes two different approaches to personally identifiable information depending on the context:
When you use AI features within the ChartPull web application (search, insights, reports), your employee data is included in the AI prompt. This is necessary for accuracy — the AI needs to know names, titles, and departments to answer your questions correctly.
This data is sent to Anthropic’s Claude API via an encrypted connection. Anthropic does not store API inputs or outputs, and does not use them for training.