Govern your AI assistant with precision

The ChatCME Admin Console lets accredited CME providers ingest program materials, define learning objectives, configure and QA the chat experience, and analyze outcomes. Editors keep full control over scope, safety, and objective alignment.

Request a demo

Who uses the Admin Console

Editors and program teams at accredited CME providers

Common roles: Program Director, Medical Director, CME Editor, QA Reviewer, Analyst

Access: Admin, Editor, and Analyst permissions with audit trails

What you can do

Ingest your materials

Upload slides, videos, transcripts, and PDFs in minutes, supporting 50+ languages.

Define learning objectives

Set clear objectives per activity and map them to content sections.

Run seed‑prompt QA

Test source grounding and citation behavior before going live.

Align every message

Map learner messages to objectives with confidence scores and editor review.

Analyze outcomes

Track themes across programs, sites, and conferences with detailed analytics.

View analytics dashboard →

Export CSV

Access message‑level data with objective alignment and citation fields.

See export options →

Core workflow

1

Create program and ingest content

Upload slides, videos, transcripts, and PDFs. The system parses slide titles, sections, and speaker notes when available, aligns transcripts to slides or timestamps, and prepares an index.

2

Use editorial AI tools

Generate a concise program description and five diverse starter questions. Editors always review before publish.

3

Define learning objectives

Set clear objectives for the activity. Link objectives to key slides and segments to support alignment and reporting.

4

Optional: Configure the assistant

Defaults are ready out of the box. Adjust tone, depth, citations, safety controls, and disclaimers only when needed.

5

Run seed‑prompt QA

Test source grounding and citation behavior, capture reviewer notes, and approve.

6

Monitor and refine

Track outcomes, themes, and objective coverage. Review unaligned messages and update objectives or content.

Ingestion and processing

  • •Formats: PPT, PDF, videos, transcripts, long‑form PDFs and articles
  • •Structure detection: slide headers, sections, speaker notes when available
  • •Transcript alignment: link citations to slide numbers or timestamps
  • •Languages: 50+ languages supported for ingestion and for answer output
  • •Versioning: reprocess updated files and keep a diffed history

Learning objectives

  • •Define and manage objectives per activity
  • •Map objectives to content so answers and alignment point to the correct slides or timestamps
  • •Objective alignment for every message: each learner message is classified to one or more objectives with a confidence score
  • •Editor review and override: accept, relabel, or mark ambiguous; changes are logged with user and time

Objective analytics surfaced in Admin

  • –Alignment rate across a date range
  • –Coverage by objective with trends
  • –Unaligned message queue to refine objectives or content

Configuration and guardrails

Defaults are production‑ready. Configuration is optional per program and used when you need fine‑grained control.

  • •Scope: restrict retrieval to the current activity
  • •Safety: avoid off‑label speculation, encourage guideline‑aligned language
  • •Citations: set density and formatting; require sources for factual statements
  • •Style: tone, depth, bullets vs narrative, summary toggle
  • •Disclaimers: consistent language you control, surfaced contextually

Editorial AI tools

Program Description Generator

Produces a concise markdown preview of scope and themes without revealing conclusions. Editors can tweak before publish.

Seed‑Question Generator

Creates five diverse, practice‑relevant starters that guide learners to explore the content. Drug classes only, length‑controlled, JSON output for the UI.

Human in the loop: nothing ships without editor approval.

Preview and QA

  • •Seed‑prompt runs: verify citations, scope, and tone
  • •Source inspection: check slide and transcript references for each answer
  • •Reviewer notes and statuses: acceptable or needs revision
  • •Re‑test after updates: run the suite again after reprocessing

Embedding and theming

See technical architecture →

  • •One snippet embed: copy, paste, and go
  • •Brand controls: color, type, light or dark
  • •Containers: inline, modal, or dedicated section
  • •SSO support: pass tokens to run within your authenticated session
  • •Kiosk mode: conference‑ready with session persistence and a PIN to exit

Analytics inside Admin

View full analytics dashboard →

Participation and engagement

Unique learners, sessions, questions per session, median dwell time

Evidence use

Citation opens, reference interactions, slide and section heatmaps

Theme mining

Clustered question patterns mapped to learning objectives

Objective coverage

Alignment rate, messages per objective, session‑level trends

Unaligned review

Identify content gaps or refine objectives

Portfolio rollups

Compare activities, sites, and conferences

Chat‑level Audit (Quality Review)

Review individual learner sessions for quality and accuracy. View full conversations with citations, track interactions through a visual timeline, and identify patterns to improve content—all with pseudonymized identifiers and provider‑only access.

Conversation view

  • •Full chat history with timestamps for user and assistant messages
  • •Per‑answer citations including slide numbers or video timecodes with quick open for source verification
  • •Auto‑generated titles and session context

Timeline view (visual)

A compact, horizontal Timeline that surfaces interaction density and event chips across the session. Filter by event type, jump to the corresponding message, and scan engagement at a glance.

Interactive timeline showing learner engagement and interactions

Quality workflows

  • •Identify answers with lower votes or frequently revisited citations
  • •Spot cold spots and hotspots in your content
  • •Feed findings back into QA and content updates

Exports and privacy

  • •Export per‑message CSV with objective alignment and citation fields
  • •Analytics are de‑identified outside the provider's team
  • •No learner‑level data is exposed to supporters
  • •Retention follows your organizational policy

CSV export fields

Your CSV includes message‑level data for analysis and archiving.

Identifiers

  • message_id
  • session_id
  • program_id
  • user_hash (pseudonymized)

Objective alignment

  • objective_id
  • objective_label
  • alignment_label
  • alignment_score (0–1)

Metadata

  • timestamp
  • slide_number
  • video_timecode

Sources

  • citation_ids[]

Roles and governance

  • •Roles: Admin, Editor, Analyst
  • •RBAC: permissions at org, app, and program level
  • •Approvals: gates for publication with reviewer identity and time
  • •Audit trails: configuration changes, reprocessing events, reviews, and overrides are logged

Security, privacy, and independence

Security & trust details →

Independence by design

Supporters receive aggregated, de‑identified insights only and have no editorial control

See what supporters receive →

Privacy

No learner‑level reporting to supporters; PHI is not required for typical CME use

Security

Encryption in transit and at rest, least privilege access, audit logs

Compliance posture

HIPAA safeguards under BAA when applicable, SOC 2 aligned controls

Optional module: measured learning outcomes

When you need to claim learning outcomes directly, we offer an optional module that we can discuss with you.

  • •Pre and post knowledge checks tied to objectives
  • •Competence scale to track confidence in key tasks
  • •Commitment to change with a 30 to 60 day follow‑up
  • •Transparent methods including N, confidence intervals, and effect sizes

Frequently asked questions

Request a demo