Give every person a clear part in the work.

SANDR has four levels of access: Owner, Manager, Can view, and Frontline. A general manager, area manager, chef, or trainer is a job title. Give each person only the venues and access their work requires.

Use the lowest access that still lets the person do their job. Coverage may need several managers for opening, closing, leave, and specialist responsibilities – but extra access is not an invitation for everyone to spend the day in AI chat.

Owners manage members, authorised venues, billing, credit purchases, capped auto-reload, exports, and lasting workspace memory. They also set the team rules that make SANDR trustworthy.

1. Explain which venues each person can access, how Frontline privacy works, and how the shared balance works before inviting the team.
2. Review access when responsibilities change.
3. Check Plan & usage weekly and compare usage against useful output, not message volume.
4. Coach repeated or unnecessarily deep use before imposing limits.
5. Review the plan before adding venues, people, or scheduled work.

A venue manager captures observations, runs check-ins, reviews patterns, assigns work, keeps proof current, and returns for the result. A group leader may use the same access across authorised venues to recognise repeated problems and organise shared work.

• Use Quick for a current priority, a clarification, or a short handover.
• Use Context when venue history, reports, or connected work changes the decision.
• Reserve Deep for recurring, ambiguous, or consequential problems.
• Compare safely surfaced evidence across venues; never copy private Frontline material between them.
• Follow one problem through to a result before adding another layer of work.

A healthy manager routine is simple: capture what changed after a meaningful service, review SANDR’s read, assign one credible response, keep the owner and proof current, and record the result after the agreed review period.

Can view is read-focused access for a person who needs approved account information without owning the work. Use it for oversight, review, or specialist input where changing venue records and plans is not part of the person’s role.

Frontline is tied to an authorised venue. Contributors share factual observations, task progress, requested proof, and handover notes, and can use a protected private conversation when reflection would help.

Raw Frontline conversation is not an ordinary manager feed. When a contributor chooses to share something, they review a safe summary before it travels. Managers respond to the surfaced issue without trying to identify the person behind it. Authorised support, safety, or legally required review may still apply.

Ordinary participation does not create a personal bill. Reading, factual updates, task completion, proof, and manual handovers do not use credits. Optional AI responses use the shared workspace balance at the published rate. The selected chat ceiling is visible before a turn, realtime voice shows 21 credits per minute before starting, and file AI requires a separate 21-credit Read with SANDR confirmation after upload.

1. A Frontline contributor records a specific lunch-service observation.
2. The venue manager adds it to a check-in and reviews the proposed patterns.
3. The manager agrees one response with a person responsible, a date, what must happen first and the proof needed.
4. A group leader checks whether the facts show the same pattern elsewhere.
5. The owner gets involved only if access, capacity, plan, or spending controls need to change.
6. After the next comparable service, the manager attaches proof and records whether the original condition improved.

That final step is the difference between administration and learning. A task marked complete shows that work moved; a result shows whether the venue improved.

Venue-specific information belongs inside the correct place or venue-scoped conversation. Never paste passwords, payment details, medical information, or unnecessary employee identifiers into a note, file, or support request.

SANDR suggests what the facts may mean and what to do next; people remain responsible for checking the facts, removing weak matches, and approving important changes. The strongest workspace is not the one with the most AI activity. It is the one that turns a real observation into a reviewed response, proof, and a measured result.