TollGateAI spend control · Sanrish Technologiesconnecting…
Daily spend
Metered spend per day, all teams and models
Spend by team
Bar = spend this budget window · marker = budget cap
Spend by model
Where the money goes across model tiers
Routing opportunity
Illustrative, token-normalized estimate from observed traffic
Recent requests
Latest traffic through the gateway
Create team
A team gets its own virtual keys and an enforced budget (daily, weekly, biweekly, monthly, or quarterly)
Teams
Budgets are enforced at the gateway. A capped team's requests are rejected until the window resets or the cap is raised
Loading…
Observe-mode spend report
The client-facing document: what was spent, by whom, on what — and what routing + caps would change. Print to PDF from the opened page.
Typical flow: run two weeks in observe mode (no caps), generate this report for the period, present it, then switch on budgets and routing.
Alert configuration
Set via environment (.env) — restart the dashboard to change
Loading…
Webhook
Slack-compatible {"text": …} POST on every alert
Alert history
Fired once per threshold crossing per team, most recent first
Loading…
Fallback configuration
When a model errors, traffic falls back to the next in the chain. Only models already in use are available as targets. Changes require a gateway restart to take effect.
Loading...
Model permissions
Restrict which models a department or project may call. A scope with no rule
may call any configured model; adding a rule turns it into a strict
allow-list. Enforced by the proxy before forwarding — a refused request is
rejected with model_not_permitted and costs nothing.
Loading…
Enforcement mode
Current global mode, as reported by the proxy. Set via ENFORCEMENT_MODE.
Loading…
Create user
Admin, team lead, or developer. Team leads and developers must be assigned to a team.