The Lab Manager's Day: Orchestrating Dozens of Cases on Cympha

The queue, the assignments and the activity log in this cookbook are the daily tools of dental lab case management software, seen from the lab manager's desk.

A busy dental lab runs on the lab manager's attention. Who's working on what. Which doctor is waiting for an answer. Which case is overdue because the doctor never sent the CBCT. Which designer is at capacity and which one has bandwidth. In the old shop, that lived in a whiteboard, a spreadsheet, and the manager's head.

Here's the same job on Cympha — the same questions, answered without leaving one tab.

Morning — the queue

You sign in. Cympha's left sidebar has Case Queue near the top — "Review incoming cases and assign designers". Click it and every case the lab has in flight is in one table: patient, category, status, designer, created.

Above the table sit a patient-name search and a status filter. The filter lists the real case statuses present in your queue — Submitted, Intake Review, Designing, Awaiting Approval, In Manufacturing, Shipping, Delivered — plus All statuses. Each row carries a Customer chat button, a Designer chat button and an Assign button; clicking the row itself opens the case detail page.

That's the morning standup, minus the standup.

Assigning work to a designer

A newly submitted case arrives at Submitted, held by you. Press Assign on its row: the Assign Designer dialog opens with a Select a designer… picker that lists only approved members of your own lab. Confirm and the case moves to Designing, and the designer finds it under My Assignments. If they run Cympha Agent Manager with Cloud Monitoring switched on, the case folder — files plus a freshly generated lab slip — lands on their machine on the agent's next check, about twenty seconds later.

A few hours later a designer calls in sick. Open one of their cases: the Assignment panel has an Assign button that opens a dialog titled Assign Case — Select a team member to assign as the current worker on this case. Pick someone else and the handoff happens. The new designer's Cloud Monitor pulls the files, and the case's event log records the reassignment with actor and timestamp.

What we don't have yet, and what's coming: a per-designer workload column on the team page (current active case count, recent throughput) so you can assign by capacity at a glance instead of by memory. Today, you assign and watch; in the next iteration, the queue will surface load.

The activity log — auditable history per case

Every case carries an Activity Log: an append-only event history covering 22 event types — creation, status changes, worker assignment and reassignment, files added and removed, labels added and removed, comments, emails to the doctor, approval requested, design approved, QC approved, case branching, and the consultation events. Nothing in it is edited and nothing is deleted. Entries read as whole sentences with actor and timestamp — "Maria moved status from Designing → Awaiting Approval", "Tom uploaded upper.stl" — and the page updates live as new events land.

When a doctor asks "why is this case taking so long," you open the activity log and read it back to them. When a designer says "I never got the CBCT," you check whether the file event is in the log. When a suggestion was auto-applied under your lab's policy, the status change or label it produced sits in the log like any other event, with the actor and the exact time.

There's no separate ticketing system. The case is the ticket; the log is the audit trail.

AI proposals — letting the busywork happen on its own

Open any active case and lab personnel see a Suggested actions tray — Cympha's AI Coworkers reading the case as it moves. Each card carries a written reason, a confidence percentage, and exactly two buttons: Apply and Dismiss. Doctors never see the tray, and it renders nothing at all when there is nothing worth suggesting; a quiet case shows no AI.

The tray knows five kinds and no others: Change status, Add label, Remove label, Request customer approval, Reassign worker. Nothing outside that list can be suggested — there is no "missing files" suggestion and no pricing suggestion — and a suggestion can neither invent a person nor bypass a rule: Apply runs the identical checks your own click would run, and a refusal comes back in the workflow's own words.

For each proposal, you can hit Apply and the action happens, or hit Dismiss and it goes away. Or — and this is the part that scales — you can decide once per lab whether each kind auto-applies.

On the My Lab page, the AI assistance section holds a master toggle — AI suggestions enabled for this lab — and, under Per-kind rules, one row per kind. For each one you set:

Auto-apply on/off — should the AI just do this without asking?
Min confidence (0–1) — only auto-apply above that score.
Daily cap (per kind, this lab) — never auto-apply more than that many times a day.

A fresh rule starts at a confidence floor of 0.85 and a cap of 100, and the screen ranks the kinds in its own words: Add label and Remove label are "Low-risk"; Request customer approval "Triggers a customer email — only turn on if you trust the AI to time these."; Change case status is "High-risk — flips the case's lifecycle stage. Recommend leaving for human review." A lab that never opens the screen gets suggestions for review only — nothing is ever applied unattended. The point isn't "let the AI run the lab" — it's "let the AI handle the labels and the obvious flags, and keep the irreversible decisions for the humans." That is where the peace of mind lives: a written floor, a written ceiling, and a Reset that revokes one kind without touching the others.

What we don't have yet: a cross-lab proposals dashboard where you see every pending proposal across all cases in one batched review queue. Today, proposals are surfaced on the case where they apply (which keeps the context tight); the batched dashboard is the next step for managers who'd rather burn through 30 proposals in one sitting than open 30 cases.

Talking to the doctor without leaving the case

Each case has a Chat thread between your lab and the doctor's clinic. It's per-case, not per-relationship — questions about Mrs. Lopez's crown stay on Mrs. Lopez's case, not buried in a long thread covering twelve patients.

Internally, there's also a lab-side channel for talking to your designer about the same case without the doctor seeing. Same UI; the chat panel switches channels at the top.

Messages are real-time on the recipient's side. No "did they get my email" anxiety.

Operations — services, onboarding, billing

The lab's catalog of work lives on the My Lab page, in the Services card: which case types you accept, what fields each one asks the doctor to fill in, what they cost. The doctor sees these when they submit; you don't customise per-doctor.

If you're new to Cympha, Onboard My Lab opens a four-step wizard — Lab Info, Introduction, Services, Payment — ending on Complete Onboarding. Until you finish it, the sidebar keeps a red dot on My Lab, and your lab stays invisible to doctors searching for it.

Each case has a price field and a payment status (paid / unpaid / refunded). On a per-case basis, you can see whether the doctor has settled. What's on the next-quarter roadmap: a payment-status column in the queue table so you can sort the whole month's work by "unpaid, oldest first" and chase efficiently. Today, payment status is per case; soon it's a queue dimension.

What the platform takes off your plate

The traditional lab manager's day is a string of low-value coordination work:

Find out what came in overnight (check three inboxes, the WhatsApp chat, and the scanner export drives).
Decide who works on what (read a whiteboard, ask designers what they're up to).
Field doctor questions (look up the case, find the file, type the reply).
Chase missing inputs (email the doctor, wait, email again).
Update statuses (move sticky notes on the whiteboard).
Reconcile who-did-what for payroll/payouts.

Cympha collapses (1) into a queue that's always current; (2) into an assign dialog; (3) into a per-case chat; (4) into suggestions that surface on the case itself as events happen. (5) is narrower than it sounds, deliberately: when a designer's aligner review file (.gvw) arrives through Cympha Agent Manager on a case already in Designing, the case flips itself to Awaiting Approval with the right step label and emails the doctor. The same file dragged into the browser does not auto-advance — the lab vets it first. And (6) is already built rather than pending: every paid case writes one ledger row, My Earnings rolls it up as Lifetime earned, Pending payout, Paid out and Refunded, and a payout batch is cut once a month covering every case whose settlement window has elapsed.

The job that's left is the job a lab manager actually wanted to do in the first place: pick the right designer for the case, set quality standards, talk to doctors when something's interesting. That is the People dimension of a Living Case earning its keep — every case knows who holds it, who held it before, and who it goes to next, so a Virtual Lab runs on the record rather than on one person's memory. The doctor feels it from their side too: they stop chasing status by email, because the case answers for itself. Less coordination, more management — and a lab you would happily walk a doctor through.

Cookbook #3 in our series. Next: a deep dive into Cympha Cloud — file sharing, collaboration links, and how a clinic, a lab, and a remote consultant collaborate on one set of files. Open Cympha at app.cympha.com.