Platform controls

Platform controls is the master switchboard for the whole platform. From here a PlatformAdministrator turns the five production loops on or off, toggles cost-bearing features, sets the global platform mode, and reads recent AI spend per loop. Open it from Settings → Platform controls.

Platform-wide, every account

These switches affect every account, not one. Disabling a loop stops that activity across the platform; in-flight productions finish, but no new work of that kind starts. Read the tooltips in the page before flipping anything.

🎬 Platform controls tour

Short video coming soon. The full written walkthrough is right below — everything the video will cover.

Prefer to read? Open the step-by-step transcript

1. Open Settings → Platform controls.
2. Toggle loops 1–5 and the feature flags — each shows its cost note and dependencies.
3. Use platform mode → Idle for a maintenance window or emergency stop.
4. Check the cost summary to see recent AI spend per loop.

The five loops

Loop	What it does	Cost	Needs
Loop 1 — Production	Runs the ContentInventoryAgent (~every 30 min) to start new productions	AI cost per production (text, image, audio, video)	—
Loop 2 — Distribution	Publishes finished productions to connected channels	AI cost for text assembly + captions	—
Loop 3 — Engagement	Polls channels for comments/likes/interactions	Ongoing AI cost even with no new content	Loop 2
Loop 4 — Performance	Computes settled performance scores + optimisation insights	AI cost for optimisation generation	— (best with Loop 3)
Loop 5 — Promotion	Evaluates and submits paid promotion campaigns	AI cost for campaign planning	Loop 2

Note

The same loop switches exist per account in Account settings. Platform controls is the global master; an account can only run a loop that’s on at both levels.

Feature flags

Flag	What it does	Cost	Needs
Engagement sentiment analysis	Classifies the sentiment of each incoming interaction	Per interaction	Loop 3
Engagement auto-response	Generates and posts AI replies to interactions	Per reply	Loop 3 + sentiment
Market intelligence	Weekly scans of market feeds to propose content angles	No direct AI cost (structured feeds)	Loop 4
Optimisation agent	Weekly content-optimisation recommendations from performance data	Per optimisation run	Loop 4
Benchmarking	Live AI calls to benchmark model response quality	AI cost — enable only for deliberate evaluation	—

Platform mode

The platform mode toggle switches between Active and Idle:

1. Active — normal operation; loops and features run as configured.

2. Idle — overrides all loop switches and feature flags; only always-on functions run. Switching to Idle asks you to confirm — “This will pause all production pipeline processing. Active productions will be held until the platform is reactivated.” Use it for maintenance windows or an emergency stop.

Infrastructure tier & markup

• Infrastructure tier (POC / Small / Standard / Production) is an informational signal that records the declared Azure resource tier — it doesn’t provision Azure resources itself (those are set in infrastructure-as-code). The platform uses it to detect tier mismatches.
• Customer markup % (0–1000) is the percentage added to JARAI’s underlying AI cost to produce the customer-facing cost. It’s applied per AI call and snapshotted at the time of the charge — changing it never rewrites historical charges. See Finance & P&L for how markup flows into margin.

Reading the cost summary

The page shows recent AI spend per loop (last several days, all accounts) plus per-feature call counts, so you can see what a loop or feature is actually costing before you leave it on. It also records who changed the config last and when.

Tip

For “is anything broken?” use Operations & reliability and the Alert operations / Production quality dashboards. Platform controls is for steering cost and capacity, not diagnosing failures.