CUSTOMER ACCOUNT / ATTRIBUTED EVIDENCE
What this case establishes
This is a commercial publisher’s customer account. Roseram checked the source and wrote the analysis; it did not interview the customer, run the product, audit the figures or establish that AI caused the reported business result. The pilot below is a proposal, not a completed experiment.
Primary material reviewed 2026-10-12. Analysis and proposed evaluation by Roseram Editorial.- 01Owned operating handbook
- 02Location-aware answer
- 03Shift-lead escalation
Plex Coffee: the documented workflow
OpenAI’s October 2025 account describes Plex Coffee growing from one café to four and connecting operating knowledge to ChatGPT. Founders Philipp Cheng and Max Kamp report more than 50% fewer routine WhatsApp questions and faster onboarding. The source discusses procurement-agent experimentation; it does not establish autonomous supplier purchasing as a completed deployment.
The mechanism worth examining
A handbook becomes valuable when a new employee can find the right procedure at the moment of need. Retrieval should identify the location, current version and applicable equipment. A conversational answer can reduce searching, but it must not flatten a safety-critical instruction into an approximate paraphrase. Where the handbook does not answer the question, the assistant needs a clear route to a shift lead.
The implementation decision
Pilot with a short, approved set of ordinary operating questions and synthetic employee accounts. Assign an owner and review date to each procedure. Start with information retrieval rather than ordering stock or changing staff records. Restrict access where location-specific or employee information should not be visible to every user.
Where the workflow can fail
An old cleaning procedure can remain indexed after a manager changes it. Two locations may use different machines, making a generic answer misleading. The system can also answer a food-safety question too confidently. Test version conflicts and ask it to cite the applicable procedure. An uncertain or consequential question should reach the responsible person.
A practical evaluation, designed for this use case
Compare answers with the current handbook and the location manager’s reference responses. Track unanswered questions, correction time and the need for human escalation. A lower count of WhatsApp messages is useful only if staff still get correct help. Measure onboarding competence separately from how quickly someone completes an assistant session.
Primary sources
Figures and product details above are dated and attributed; they are not current guarantees.
- OpenAI: Plex Coffee primary account ↗The attributed workflow and reported result. Proposed pilots and decisions below are original Roseram analysis.
The decision matrix
Original evaluation criteria for this workflow. These are questions to test, not results measured by Roseram.
| Decision | Evidence to collect | Stop or revise when |
|---|---|---|
| Version | Current procedure and owner identifiable | Retired instructions remain authoritative |
| Location | Equipment and branch context checked | One café’s procedure is applied to another |
| Outcome | Correct help and escalation measured | Fewer messages conceal unresolved questions |
YOUR EVALUATION / NOT A REPORTED RESULT
A small pilot you can inspect
Use these steps as a planning checklist. Checking a box records your progress in this view; it does not run or validate an experiment.
Questions this case answers
Do fewer messages prove better onboarding?
Not alone. Check correct answers, unresolved questions and actual competence.
Was automatic purchasing established by this account?
No. The source describes procurement experimentation rather than a verified autonomous purchasing result.
Tools & capabilities referenced
Operators studying this story often evaluate adjacent AI products. Explore Roseram guides where available:
PRACTICAL NEXT STEPS
Take the evidence into your own project.
Calculate the whole project cost
Compare subscriptions, usage, launch work and renewals.
Test an existing-project workflow
Check import, source loading, preview and publication.
Verify the output you buy
Require evidence tied to a real customer task.
Wiki brief
Plex Coffee is examined here for café onboarding and accessible operating knowledge. The article answers “How can a growing café use AI to reduce repetitive onboarding questions?” using linked primary material and original evaluation design. It is not a customer testimonial for Roseram or an endorsement by Plex Coffee.
Operator takeaways
- Version the procedures.
- Include location context.
- Count unresolved questions too.
Related case studies
Plan → build → review with routed models and transparent usage.