The agents
Delegate your credentialing to purpose-built agents.
Not one all-purpose bot — a team of specialists that takes the whole job, from placement confirmed to submit-ready file.
Assistant Agent
The one you talk to
You talk to it — by text or in the app. It knows every file and rule in your workspace, answers anything, and calls the right agent for any job.
Directs 8 agents · across the workspaceProfile
Operation Lead
One per placement
Every placement gets its own lead. It reads the rules, plans the work, dispatches the specialists — and drives the file all the way to cleared.
Directs 17 agents · inside every placementProfile
Computer Use Agent
SPECIALThe operator
Some verification sources have no API — just a website. So this agent opens a real computer, logs in, clicks through, and does the work by hand, exactly like a person would — then hands the result back to the team.
License lookup
driving the browser · no API needed
Solutions Architect
The builder
Discovers and provisions new integrations and services on demand.
Workspace Management
The librarian
Curates the workspace's knowledge base and catalog.
Workspace Requirements
The rule-keeper
Owns the requirement rules that apply to everyone.
Facility Requirements
The facility expert
Learns each facility's packet and builds its checklist.
Requirement Recipe Planner
The planner
Designs exactly how each requirement gets satisfied.
Staff Context Builder
The people person
Keeps a living profile of every staff user.
Question Answering
The explainer
Answers batches of questions about any file, instantly.
Operation Requirements Lead
The foreman
Turns a placement into its full, materialized requirement list.
Execution Plan Builder
The strategist
Orders every task into a DAG — the plan of attack.
Plan of Action
The scheduler
Lays the whole process out on a timeline, phase by phase.
Step Generator
The decomposer
Breaks each requirement into concrete milestone steps.
Entity Extraction
The reader
Pulls names, dates, and numbers out of any document.
Entity Assignment
The router
Assigns the right party to each step of the work.
Background Check
The screener
Runs the right screening package with your vendor.
Reference Check
The reference-getter
Picks the survey and chases references until they land.
Communication
The diplomat
Texts and emails everyone involved, in a human voice.
Document Acquisition
The fetcher
Finds or creates the exact document a step needs.
Document Authenticity
The detective
Catches tampered, wrong, or expired documents.
Signature Document
The closer
Gets the right forms e-signed, in the right order.
Form Generator
The form-filler
Builds the exact form schema each open question needs.
Vault Intake
The filer
Files every credential into the clinician's vault.
Vendor Assignment
The matchmaker
Matches each requirement to the right vendor.
Wallet Credential Matcher
The matcher
Reuses verified credentials the clinician already holds.
Conditional Requirements Resolver
The adjudicator
Decides which conditional and optional rules actually apply.
When you press go
What actually happens behind the scenes.
One message starts it. A lead agent breaks the file apart, hands each piece to a specialist, the specialists spin up sub-agents — and the whole chain talks back and forth until the work is done.
No one told each agent what to do step by step. The lead decides, delegates, and adapts — the way a great colleague would, at a scale a team never could.
How an agent works
Every agent runs the same loop. Plan, act, verify, report.
Plan
It reads the placement, the facility packet, and your rules — and writes its own task list.
Act
It works real tools: vendor APIs, the ATS, e-sign, and a browser for portals that have none.
Verify
It checks every result against the requirement and attaches the evidence to the file.
Report
It reports back up the chain — and hands anything that needs judgment to a human.
The loop is the safety property: an agent never marks its own work done — verification is a separate step, and the evidence rides with the result.
Enterprise-ready
Autonomy your compliance team can sign off on.
- Audit trailEvery action is logged — who, what, when, and the evidence behind it. Auditors get answers, not archaeology.
- Role-scoped authorityAgents act inside the same permissions your team has — computed per user, per file.
- Evidence-linked answersEvery answer cites the document or record it came from. Nothing is asserted without a source.
- Humans on judgmentFlags, anomalies, and low-confidence calls route to your team before anything moves.
The full oversight model — checkpoints, containment, and evals — is on Safety & human oversight.
Go deeper
Every agent has its own page — what it does, the tools it runs, and its limits.
Questions
What is a credentialing AI agent?
A credentialing AI agent is a purpose-built AI worker that owns one job in the credentialing process — verifying a license, running a background check, chasing a document — and executes it with real tools: vendor APIs, your ATS, e-sign, and a built-in browser for portals with no API. Specialists work under an operation lead that owns the whole placement.
How do the agents work together on a placement?
Every placement gets its own Operation Lead. It reads the facility's rules, breaks the file into its full task list — typically 30–50 requirements — and hands each task to the specialist that owns it. The specialists report back, the lead tracks progress and unblocks stalls, and the file is driven to submit-ready.
Can the agents act without human approval?
Routine execution runs autonomously; judgment stays human. Background check flags route to human adjudication, suspicious documents are escalated with the evidence highlighted, and low-confidence calls wait for a person. Every action is logged either way.
What tools and systems do the agents use?
The ones you already run: Bullhorn, JobDiva, Salesforce, Checkr, Vetty, Yardstik, RefAssured, and Twilio — plus a built-in browser for state boards and vendor portals that have no API. Nothing about your stack has to change.
How is this different from workflow automation?
Workflow automation follows a fixed path and breaks when reality doesn't match it. An agent plans its own path, adapts when a portal is down or a document is wrong, and escalates when it needs judgment — the way a colleague would, not a script.
Which agent does my team actually talk to?
One: the Assistant. It knows every file and rule in your workspace, answers with sources linked, and turns instructions into work for the rest of the team. You never route anything yourself.
Watch the team work a real file.
Bring your hardest placement. Watch the agents delegate, work their tools, and talk it through — start to submit-ready — in thirty minutes.