Your department’s AI Training Profile
Wisteria’s AI Training Profile has two layers:
- Company baseline — written by the super_admin. Applies to every department.
- Department profile — written by you (the trainer for your department). Applies only to your department’s suggestions.
The two compose: when the AI evaluator looks at a candidate file for your department, it sees company baseline + your department’s profile.
For the concept, see AI Training Profile.
Where to edit it
Settings → AI Training → [Your Department tab]. The tab for your department is editable by you; tabs for other departments are read-only.
You can read every department’s profile — useful for understanding what context is informing the AI’s per-department judgments — but you can only edit your own.
What to fill in
Five plain-text fields:
1. Department mission
What does this department actually do? Two or three sentences.
Example (for a hospital’s Cardiology department):
Cardiology delivers diagnostic, interventional, and surgical cardiac care. Our team includes cardiologists, cardiac surgeons, cath-lab technicians, and recovery nurses. We coordinate closely with Emergency, Anaesthesia, and Imaging.
2. Department-specific training priorities
Beyond the company baseline, what’s especially relevant for this department?
Example:
ACLS recertification, sterile technique in the cath lab, post-procedure complications recognition, telemetry interpretation, medication reconciliation for cardiac meds.
3. Department language
Terms of art, acronyms, role names. Helps the AI evaluator recognise relevant content even when it uses jargon.
Example:
PCI, STEMI, NSTEMI, IABP, ECMO, TAVR, MIDCAB. We refer to “the cath lab” and “the OR” interchangeably with their formal names. RN positions are CVICU, step-down, or pre-op.
4. Department exclusions
What NOT to surface for this department, specifically. Adds to the company baseline’s exclusion list.
Example:
Cardiology-adjacent admin content (scheduling, billing, equipment procurement), Emergency Department–specific protocols (we get separate training for those), non-clinical staff onboarding.
5. Department notes
Anything else that should colour the AI’s framing — tone, culture, ongoing initiatives.
Example:
We’re rolling out a new TAVR programme this quarter, so any TAVR-related content is high-priority. Our department head emphasises “evidence over experience” — training should cite trials or guidelines where possible.
When to update
- After the first scan that includes your department — see what came back; tune.
- When your team’s scope changes — new procedure, new equipment, new compliance regime.
- Quarterly maintenance.
What happens if you leave it blank
The company baseline still applies. The AI evaluator falls back to generic reasoning for the department layer.
Workable, but you’ll get noisier results than if you’d given the AI context specific to your team. Even a few sentences in each field improves the signal-to-noise.
Reading other departments’ profiles
You can browse them as read-only. Useful when:
- A scan surfaces a file that seems intended for another department — checking their profile tells you whether their trainer would want it.
- You’re cross-pollinating: a peer department has a strong training programme and you want to see how they’ve framed their context.
Edit permissions
Only the trainer assigned to a given department can edit that department’s profile. If you’re a trainer in multiple departments (rare; Wisteria enforces one-department-per-user), you’d edit each separately.
Super admins can edit every department’s profile (they have global write access). Content managers and auditors can read but not edit.
Privacy
The profile text is sent to Anthropic’s API (Claude) along with each candidate file’s text. Anthropic doesn’t retain API data for training. Treat the profile as “context the AI needs,” not “everything secret about us.”