Models chosen per task
We match each task to the right Claude model — a fast, economical model for high-volume analysis, a balanced model for everyday work, and the most capable model for complex reasoning — and move to newer models as they ship.
Capability · AI
AI is inside the work, not above it.
We don’t add AI because every landing page has to mention it. We add AI where it actually helps ministry admins — coaching managers on review feedback, catching payroll errors before they ship, and summarizing long review datasets.
Where AI lives
Woven through the platform, not bolted on.
TMR: Leadership Reports
AI analyzes each manager’s review feedback — specificity, actionability, score justification — and generates coaching recommendations with priority. Editable by HR, PDF-exportable, emailable.
Learn more →TMR: Comment analysis
AI analyzes individual review comments or batches of comments. Feedback quality score, scoring accuracy against comments, manager profile labels (Mentor / Effective / Developing / Emerging / Needs Coaching).
Learn more →Payroll: Anomaly detection
AI flags pay runs that diverge from historical patterns for this employee or this period. Catches data-entry errors before the pay run is submitted.
Learn more →Employee 360: Whole-person analysis
AI assembles a composite health and retention score per employee and writes the strengths, concerns, and issues behind it — surfacing who needs attention before it’s obvious.
Learn more →RMS / ChMS: Optional engagement signal
Churches that connect a ChMS can optionally include giving engagement as one signal in the health and retention scores — visible, tunable, and entirely controlled by your admins.
Learn more → How we use AI responsibly
What we think is worth mentioning.
Batch API for cost
Organization-wide analysis uses the Anthropic Batch API — 50% lower cost than real-time. Results applied asynchronously without blocking UI.
Human in the loop
Every AI-generated coaching recommendation and comment analysis is reviewable and editable by HR before it reaches a manager or goes into a report.
Prompt transparency
The prompt templates that drive our AI features are reviewable and versioned. Not a black box.
Query logs
Every AI call is logged with prompt, response, token counts, and cost. Queryable for debugging or audit.
Cost tracking
AI spend per organization is tracked. We can answer "how much did AI cost us last month?" with numbers.