Understand credits, modes & usage metering
Kai Hub meters every AI action against a single credit wallet. This article explains how your balance is made up, how the two execution modes differ, exactly what consumes credits, and how to read the Usage tab. Kai Hub is admin-only.
The Overview credit-balance card
The hero card on Overview shows your Available credits as a big number, with a sub-line reading {plan} · {emps} employees · {grantPerEmp} cr/emp/mo pooled — for example, Professional · 500 employees · 15 cr/emp/mo pooled.
A segmented pool meter breaks the wallet into four legend segments:
- Used
- Grant left (subscription pool)
- Reserve (employee reserve left)
- Top-up (never expires)
A note reads Grant resets on 1 Aug · top-up credits never expire · ~{d} days to depletion at current burn. Buttons let you ⚡ Top up credits, and a status chip shows Auto-recharge on or Auto-recharge off.
For the worked example — plan Professional, 500 employees, grant 15 cr/emp/mo gives a monthly grant of 7,500 cr; the top-up balance is 2,400 cr and the employee reserve is 20%.
Low-balance banner and burn-down
When less than 50% of credits remain, a banner appears: ⚠️ {X}% of credits left — at the current burn (~{n} cr/day) the wallet lasts ~{d} more days. with a Top up now button.
The Overview KPI row shows five figures: Total AI calls, Credits burned (which links to Usage), Est. value (₹, = credits 1:1), Features used, and Active users. Two charts sit below: Credit burn-down · this cycle (plotted against the employee reserve floor) and Daily AI calls. Estimated value is ₹1 per credit.
Kai Swift vs Kai Genius
Every action runs in one of two modes:
- Kai Swift — "fast, low-cost models for everyday actions".
- Kai Genius — "frontier models with extended tool use, reserved for the three features where depth pays for itself."
Employee flows are always Swift.
What consumes credits
Each feature bills at a fixed price in the mode shown:
Kai Copilot - Copilot message · admin — Swift — 2 cr - Copilot message · Genius — Genius — 7 cr - Copilot message · employee — Swift — 0.5 cr
Workforce Intelligence - Insight card — Genius — 15 cr - Cockpit refresh — Genius — 50 cr - Ad-hoc question — Genius — 10 cr
ATS · Hiring - Resume parse — Swift — 5 cr - JD Generator — Swift — 3 cr - Candidate scoring — Swift — 1 cr - Interview questions — Swift — 3 cr - Candidate insights — Swift — 5 cr - Feedback synthesis — Swift — 3 cr - AI search — Swift — 1 cr - Auto-Pilot (batch of 25) — Swift — 25 cr
Letters - KredAi letter draft — Swift — 2 cr
Payroll - Auto-Pilot run — Genius — 100 cr
Performance - Review draft / summary — Swift — 2 cr — shipped OFF by default
The Usage tab
Usage by module
The Usage by module card has an ⇩ Export button (toast: Usage CSV exported). Columns are Module · Calls · Credits · % of burn · Mode split (⚡ Swift % / ◆ Genius %) · Trend. Modules listed are Kai Copilot, Workforce Intelligence, ATS · Hiring, Letters, Payroll, and Performance.
Rows are clickable. Selecting one opens a drill-down, ↳ {Module} · by feature, showing each feature with its mode badge, a bar, and {credits} cr · {calls} × {price} cr.
Usage by user
The Usage by user card has a Search user… box. Columns are User · Department · Role · Calls · Credits · Top feature. When nothing matches, it shows No users match.
Selecting a row opens a per-user modal with the user's avatar, {dept} · {role} · this cycle, total credits and calls, a By feature breakdown, and a cap chip — Cap: {n} cr/mo, Unlimited exception, or No individual cap. Buttons let you Export (toast: Usage report for {name} exported) and Set cap.
When to contact support
If your Credits burned figure or a module's % of burn looks wrong for the selected date range, contact support with the module name and date range so we can check the metering.