Choose how deeply SANDR participates.

SANDR Credits are a customer-facing unit of AI work. They are not provider tokens, money, cryptocurrency, or a personal employee allowance. The workspace shares one balance across authorised members and venues.

Chat responses, check-ins, scheduled AI, voice, optional AI file reading, and other clearly marked features use credits. Reading completed reports, viewing or completing tasks, adding facts before asking SANDR, managing members or venues, and ordinary work do not.

The composer shows the selected ceiling before a chat turn. Realtime voice shows and confirms its 21-credit-per-minute rate before starting. File upload and preview are free; the separate Read with SANDR action shows and confirms its fixed 21-credit rate before processing. Final debits appear in Settings → Plan & usage. Check the current rate card rather than an old screenshot.

Advisor

Managers keep facts, tasks, proof, and results current, then bring SANDR the decisions that need a second brain. Quick for small work, Context for a few venue-aware questions, Deep for rare high-value problems. A complete business record with low variable usage.

Focused manager

Brief manager check-ins around meaningful services, selected work that needs venue history, and a weekly review. Frontline contributes facts and proof; scheduled AI is limited to venues and days where it creates real value. A strong default for early adoption.

Day-to-day partner

SANDR carries the rhythm between shifts and meetings: daily facts, the overnight pass, a morning briefing, conversations tied to the right venue, Frontline participation, files, and regular result reviews. More credits, and much less information lost.

1. Performance – Quick is a small no-tool ceiling; Context reads selected sources and takes safe actions; Deep supports complex multi-step work; Auto picks the lightest sufficient level.
2. Context – include only the sources that can change the answer. The switches cost nothing by themselves, but the resulting work may.
3. Rhythm – enable scheduled work only for active venues where the briefing is actually read.
4. Owner spending controls – credit packs and capped auto-reload are deliberate owner actions, and auto-reload is off by default.

Avoid a blunt message quota. A short generic question can be wasteful, and one deep investigation can be excellent value. Coach the work and its result first, then adjust controls if a pattern persists.

When available credits reach zero, new metered AI work pauses. Completed reports, history, tasks, evidence, and non-AI work stay available, and there is no automatic excess bill simply because the included allowance ran out.

An owner can buy a one-time credit pack, or enable capped auto-reload where the plan allows it. If the wallet is empty, the message SANDR shows says so plainly and points the owner to Plan & usage rather than offering a retry that will fail again.

• Included credits refresh monthly and do not roll over.
• Purchased credits remain available for the period shown at checkout.
• A failed auto-reload adds no credits.
• Reserved credits return if managed AI does not start or usage cannot be reconciled.
• Credits do not unlock a feature or capacity the plan does not include.

Plans determine active venue and member capacity as well as included credits. Adding a venue does not automatically add credits or people, and extra credits do not unlock a higher plan’s features. Preview planned public-launch limits and prices on Plans and pricing.

Applications for the selected 60-day guided beta are open now. Participating teams receive onboarding, practical support and 2,000 promotional credits without a monthly subscription fee. Wider self-guided plans will open later; the launch date is not yet confirmed. Apply for the guided beta if you would like to take part.

Weekly while the team is getting started, and monthly once the rhythm is stable, the owner should review available, reserved, and used credits; the forecast; recent usage by feature; purchases; renewal; and auto-reload status. The current usage view does not provide a venue-by-venue breakdown.

1. Which AI work produced a decision, plan, reusable brief, or measured result?
2. Where did repeated prompts point to missing context, training, or a product gap?
3. Are scheduled runs being read and acted on?
4. Is one venue or person using depth without a proportionate decision?
5. Should the adoption style, context defaults, or monthly cap change?

Managers can use the shared balance available to their role. Only the workspace owner controls purchases and billing settings.