Help centre / Guide 01
Start with one real place.
SANDR becomes useful through real observations and one problem followed through to a result. You do not need a perfect brief or a finished management setup – just enough truth to help with the next move.
01 / Guide 01
Begin with the venue you want to understand.
After your team is selected for the guided beta, use the invitation and onboarding instructions we send you to enter its workspace. Then add one real venue with a recognisable name, type, stage, and location.
Keep the venue separate from the brand: North Café, Vilnius is a place, and a brand may later connect several places. Your SANDR access role controls what you can see and change, and it is separate from your hospitality job title.
You can see the venue in Places, open it without an error, and start a check-in that is visibly scoped to that venue.
02 / Guide 01
Give SANDR one useful observation.
Write what happened, where, when, and what changed. Plain facts beat polished language, and separate observations are easier to review when they are on separate lines.
Friday, 08:00–09:00. One barista was absent. Drinks reached a ten-minute wait and two regulars left before ordering. The handover did not name a rush owner.
Names the window, the visible condition, the numbers, and the consequence.
“Staffing is bad.” It may be true, but it gives the team nothing specific to verify.
Choose the check-in depth that fits the facts you actually have: Light for one focused issue, Medium for a balanced read, or Full when a broader business review is genuinely useful.
03 / Guide 01
Review SANDR’s read before it becomes work.
SANDR organises the observation into proposed patterns, a report, and a practical next move. Confirm the venue, remove anything that does not match what happened, correct anything that looks too serious or not serious enough, and check that the proposed work answers the real issue.
A committed check-in becomes live workspace history. It is not demo content: you can return to its report, continue the plan, assign work, attach proof, and record whether the original condition improved.
- Keep only the patterns the facts support.
- Choose the smallest credible response.
- Give the first move an owner and a date.
- Define what completion proof should look like.
- Return after the next comparable service for the result.
04 / Guide 01
If setup is interrupted, continue – do not duplicate.
Sign back in with the same account and workspace. First-run setup resumes from the last step the server confirmed, including on another device.
If the venue was created before you left, choose Continue with [venue name] when it appears. That reconnects onboarding to the venue you already made without consuming another venue slot. Do not create a second copy to restart; if the existing venue is not offered, contact support with the workspace and venue names.
If the page closed while SANDR was reading the first observation, the flow returns to a safe check-in step so the read can be started again rather than hanging. If the first check-in already completed, SANDR reloads the saved result.
Open Settings → Help → Send feedback, or email gm@sandr-ai.com. Include the workspace and venue name plus the exact message shown. Never send a password or a verification code.
05 / Guide 01
Build a light routine after the first result.
A good beginning is one short manager check-in when a meaningful service or condition changes, followed by a deeper weekly review. Add Frontline facts, task updates, and proof as they happen instead of reconstructing the month from memory.
Lunch handover started 12 minutes late again. The new section map helped, but two tables still waited for drinks. Closing manager will confirm whether sections were assigned before tomorrow’s briefing.
Over time, repeated facts let SANDR separate an isolated incident from a pattern. Value grows when the team follows problems through to results, not when it creates more AI text.
06 / Guide 01
Know what is shared, what is metered, and what stays your judgement.
Business information stays inside the workspace and the access granted to each person. Share only what the work needs, and avoid passwords, card details, private keys, medical information, and unnecessary personal identifiers.
AI work uses the workspace’s shared SANDR Credits. Credits are a unit of AI work, not provider tokens, cash, or cryptocurrency. The composer shows the selected chat ceiling before a turn, realtime voice shows 21 credits per minute before it starts, and an uploaded file stays free until you explicitly choose the 21-credit Read with SANDR action. Final debits appear in Settings → Plan & usage.
Reading completed reports, editing existing work, completing ordinary tasks, and adding evidence before analysis do not use credits. If the balance reaches zero, new metered AI pauses while existing history and non-AI work stay available.
SANDR supports practical management decisions. She does not replace legal, employment, accounting, medical, or food-safety advice.
Keep going
Follow the next thread.
Each guide starts from a real part of the work and keeps the next move clear.