Agents

Agents is your AI workspace for program questions, structured insight reports, and proposed actions you approve before anything changes in HealthTasks. Open Agents from the sidebar.

Overview

Institution administrators use Agents to:

Chat with a conversational agent about your program data
Run canned Insight Reports for performance, curriculum mapping, competencies, course objectives, and program outcomes
Ask the agent to propose changes to placements, compliance templates, or curriculum mappings—then review and approve them in the chat

Agents replaces the older AI Insights page. Old AI Insights bookmarks redirect here. Usage is pooled monthly for your institution and managed on Billing → Agents.

Agents requires an eligible Intelligence plan. It is not available on Free Trial or Vision-only setups.

Conversational agent

Start a New chat and ask natural-language questions about classrooms, students, curriculum, or program performance. Threads persist so you can return later.

Open the + button next to the message box. The menu splits Skills and Context:

Skills — pre-built playbooks (see below) that add a badge on the composer
Context — pin a classroom when the job is scoped to one class

Add optional details in the message box, then send.

Common thread actions:

Search, pin, rename, or delete chats
Share a thread with other admins at your institution (shared viewers can read the thread but cannot send messages or approve changes)
Narrow scope (for example classrooms or students) when you want a focused answer

The agent can query your institution’s data, cite HealthTasks help articles, and build charts in the chat. Treat analytical answers as decision support—verify important conclusions against Reports and source records before accreditation or leadership presentations.

Skills (guided workflows)

Skills are one-click playbooks in the composer + menu. They tell the agent exactly how to run a multi-step job without you pasting a long prompt.

Skill	What it does
Automated placements	Match unassigned preceptors and students to nearby eligible affiliated sites. Prefers existing schedule tracks with open capacity; can propose new tracks when needed.
Close curriculum gaps	Find unmapped competencies in a framework you pick, then propose links to existing skills, evaluation questions, and checkoff steps.

Automated placements options

Before sending, you can toggle matching rules:

Require student compliance — skip students who are not marked compliant
Require compliant preceptor on track — only use tracks that already have a compliant preceptor, or assign one first when creating a new track

Pin a classroom from Context to limit planning to one class. Otherwise the skill plans across active classes with unassigned members. Distance uses home ZIP codes (see Users) and is estimated ZIP-area miles—not driving time.

Close curriculum gaps options

Pick the competency framework in the composer before sending. The skill maps only onto existing performance items—it does not create new skills, evaluation questions, or checkoff steps in the same run.

Agent actions and approval cards

Some requests go beyond read-only analysis. When the agent is ready to change data, it saves a proposal and shows an approval card in the chat. Nothing is written to your institution until you approve.

What the agent can propose

Area	Examples of proposed changes	Not included
Placements	Add or edit clinical sites; parent location groups; site units and capacity; schedule tracks and rotation segments; assign students or preceptors to tracks; nearby placement drafts	Contract file uploads (metadata only); replacing Suggest Assignments on the Placements page for manual one-click drafts
Compliance	Create, edit, or delete compliance templates; update the standard template’s requirement list; assign templates to users	Uploading documents; approving or rejecting individual compliance items
Curriculum mapping	Link unmapped competencies to existing skills-template items, evaluation questions, or checkoff steps	Creating new skills, evaluations, or checkoff steps

You can also ask in plain language—for example “create a track at Memorial for the Spring class” or “assign the standard compliance template to these students”—and the agent will query first, then propose the matching operations.

Student creation, billing changes, and other admin tasks outside these surfaces are not available through Agents today.

How the approval workflow works

Propose — The agent gathers ids and context, captures a rollback snapshot of affected records, and queues a card labeled Review.
Review — The card shows a short explanation and a bulleted list of each change. For multi-step proposals, uncheck rows you do not want before approving.
Approve — Click Approve (or Approve N when only some rows are selected). HealthTasks applies the selected operations. The card updates to Applied.
Dismiss — Remove a proposal without applying it. Dismissed cards disappear from the thread.
Roll back (optional) — After approval, Roll back reverses the applied changes using the snapshot taken before approval. The card returns to a state where you can Apply again if you change your mind.

Important rules:

You must approve — The agent will not claim a write is complete until you approve. If you see a card, review it before assuming Placements, Users, or Templates already changed.
Thread owner only — Only the admin who started the chat can approve, dismiss, or roll back. Colleagues with a shared thread can read the conversation and cards but cannot act on them.
One pending proposal per type — A new placement, compliance, or curriculum proposal in the same chat replaces the earlier pending proposal of that type.
Stale proposals — If someone edits the same site, template, or mapping in the product while a card is pending, approval may fail with a message to dismiss and ask the agent to re-propose.

Approval card statuses

Status	Meaning
Review	Waiting for your decision
Applied	Changes are live; Roll back is available
Rolled back	A previous apply was undone; you can approve again
Dismissed	Closed without applying

Partial approval

When a card lists multiple changes, use the checkboxes to include only the rows you want. Dependent operations stay grouped—unchecking a parent row may uncheck related rows. Approve at least one selected row to apply.

Insight Reports

The sidebar lists five pre-built reports. Each generates a structured analysis without writing the prompt yourself:

Report	Focus
Overall Performance	Program health, metrics, and priorities
Curriculum Mapping	Structure and gaps across competencies, COs, and POs
Competency Analysis	Competency coverage and accreditation readiness
Course Objectives	Course-level objective achievement
Program Outcomes	Highest-level outcome effectiveness

Select filters when prompted (such as cohorts or templates), then generate the report. You can export finished reports for meetings or documentation.

Each report run consumes Agent usage from your institution’s monthly allowance.

Usage and billing

Agent usage is a monthly allowance for your institution:

Limits reset on the 1st of each month (UTC)
A usage indicator in Agents shows how much capacity remains
When you hit the limit, wait for the next reset or upgrade on Billing → Agents

Standard usage is included for eligible institutions; higher monthly tiers are available as an add-on if your team needs more capacity. Exact allowances appear on Billing. Bulk Personalized Learning Plan generation in classrooms also draws from this pool (see Classrooms).

Tips

Use Skills for placement matching and curriculum-gap mapping—they are faster and more consistent than free-form prompts.
Read every approval card before approving; use partial selection when you only want part of a draft.
After rolling back, fix the underlying issue in chat or in the product, then approve again or ask for a fresh proposal.
Use Insight Reports for recurring leadership or accreditation reviews; use chat for ad hoc questions and operational drafts.
Scope reports and placement runs to the cohorts and classrooms that matter for the decision at hand.
Prefer Agents in training materials—not the legacy “AI Insights” name.

Troubleshooting

Agents missing from the sidebar
Confirm you are an institution administrator on an Intelligence plan. Faculty and students do not see this item.

Usage limit reached
Usage resets on the 1st (UTC), or an administrator can upgrade the Agents allowance on Billing.

Report will not generate
Confirm required filters are selected and that your templates and program data exist for that scope. Retry after a moment if generation was interrupted.

Approval card will not apply
The underlying record may have changed after the proposal was created. Dismiss the card and ask the agent to propose again. Confirm you are the admin who started the thread.

Cannot approve on a shared thread
Only the thread owner can approve or roll back. Open your own chat or have the owner approve.

Agent says changes are done but Placements looks unchanged
Writes require approval. Look for a Review card in the thread and approve it.

Rollback failed
Some later edits may block undo. Note what changed, fix manually in Placements, Users, or Templates, or ask the agent for a corrective proposal.

Old AI Insights link
Old AI Insights bookmarks redirect to Agents. Update bookmarks and help links to Agents.

Placements: Manual schedules, Suggest Assignments, and the compliance matrix
Users: Cohort codes and home ZIP used in placement matching
Compliance: Document uploads and review (separate from template proposals in Agents)
Templates: Curriculum frameworks and performance items linked by gap-mapping
Classrooms: Bulk PLP generation and classroom-scoped context
Billing: Agents usage tiers and upgrades
Reports: Non-AI operational reporting
AI-Insights: Short redirect note if you still land on the old name

Support

For Agents access or usage questions: support@healthtasks.ai