Healthcare AI from the ITECS Dallas team
AI Solutions for Healthcare Practices
ITECS builds governed AI for specialty physician groups, dental groups, surgery centers, and home health and behavioral health providers of 50 to 300 people. Agents verify coverage, determine prior-authorization requirements, assemble cited packets, and work denials under a business associate agreement, while staff and clinicians approve each submission and clinical decisions stay with clinicians.
Practices do not need AI that guesses at medicine. They need AI that clears administrative work before the patient arrives, keeps protected health information to the minimum each task needs, and leaves every clinical and payer decision with a named person.
Healthcare signal map
Access + revenue + privacy
Access
Eligibility, scheduling, and intake
Revenue
Authorizations, denials, and appeals
Privacy
BAAs, minimum necessary, and audit trails
HIPAA discipline behind healthcare AI
Healthcare AI from a team that already runs HIPAA-aligned IT in Dallas.
ITECS has supported Dallas healthcare organizations since 2002 with managed IT, cybersecurity, and HIPAA compliance support, and its business associate agreement is included at no additional cost for healthcare clients. ITECS AI builds on that base: risk analysis, access control, and audit logging come before any agent touches protected health information. Read about ITECS HIPAA compliance support.
13 hours
Weekly physician and staff time spent on prior authorizations, across about 40 requests per physician.
AMA prior authorization survey, 202524 minutes
Provider time for one manual prior authorization, against 10 minutes when it runs fully electronic.
2024 CAQH IndexJan. 1, 2026
Since then, Texas has required providers to disclose AI used in relation to health care service or treatment by the date of first service.
Texas HB 149 (TRAIGA)
Operating pressure
Healthcare AI Starts With Access, Revenue, and Privacy
The first step is not picking a model. It is finding where front-desk and billing staff rework the same coverage and authorization problems, and where protected health information needs firmer boundaries before AI touches it.
Authorization pressure
Requirements vary by payer, plan, and service, and they change. Staff discover gaps on the phone or after a denial.
- Coverage changes found at check-in instead of before the visit
- Prior-authorization rules that differ by payer, plan type, and site
- Packets assembled by hunting through the chart
- Payer status checks by phone, portal, and fax
Denial pressure
Denied dollars pile up faster than staff can work them, and few appeals get filed even when the documentation exists.
- Denials worked in arrival order instead of by value and deadline
- Appeal deadlines tracked in spreadsheets
- Root causes that never reach the front desk
- Write-offs decided without a consistent policy
Privacy pressure
AI is only usable when PHI stays inside business associate agreements, minimum-necessary limits, and Texas disclosure rules.
- Consumer AI accounts with no business associate agreement
- Chart access broader than each task needs
- Texas AI disclosure duties for services tied to patient care
- Substance use disorder records that need separate consent handling
Example revenue-cycle signal board
The Metrics a Healthcare AI System Should Make Visible
An authorization and denial-prevention program should show the practice administrator denied dollars, what is recoverable, which scheduled services still lack an authorization, and where front-end checks slip. These figures are illustrative for a fictional 120-person specialty group; discovery replaces them with the practice's own data.
Denial bridge
Denied dollars by root cause
Initial denials this month
$610K
11.3% of $5.4M in submitted charges
Recoverable by appeal or corrected claim
$386K
Denials with documentation on file or a correctable error
Scheduled services missing an auth
41
Next 10 days, with $236K in expected reimbursement at risk
Visits re-verified 72 hours out
94%
The rest routed to the front-desk exception worklist
What leadership sees
- Ties each denied dollar to payer, denial code, service line, and site
- Separates preventable front-end failures from payer medical-necessity disputes
- Routes packets, appeals, corrected claims, and write-offs for staff approval
Healthcare AI use cases
Where AI Creates Measurable Value for Practices
Start with administrative work where the data already arrives and staff already approve the result: coverage, authorizations, denials, and intake.
Eligibility & Benefits Re-Verification
Run eligibility checks at scheduling and again 72 hours before the visit, flagging terminated coverage, plan changes, coordination-of-benefits conflicts, and network status.
Coverage problems surface before the patient arrives
Prior-Authorization Detection & Packets
Check each scheduled order against a maintained payer-rules library, then map minimum-necessary chart evidence to each payer criterion with citations.
Each scheduled service has an auth status and an owner
Denial Triage & Appeal Drafting
Classify remittance denial codes and payer letters by root cause and deadline, then draft corrected claims or appeals that cite payer policy and the chart.
Denials worked by value and deadline
Referral & Fax Intake Classification
Classify inbound faxes and portal documents, match them to the patient and the right worklist, and flag urgent items for staff.
Referrals and payer letters land with an owner
Coding & Charge Review Assistance
Check documentation-to-charge consistency, missing modifiers, and bundling conflicts before claims go out, routed to a certified coder who keeps the code decision.
Fewer avoidable coding edits
Credentialing & Payer Enrollment
Track CAQH ProView attestations, license and DEA expirations, and payer enrollment applications, and assemble packets for staff to submit.
New providers become billable without avoidable gaps
Patient-Message Drafting With Clinician Review
Draft replies to portal messages and refill requests for a clinician or nurse to edit and approve before anything is sent.
Faster inbox turnaround, accountability intact
Policy & Payer-Manual Knowledge Base
Answer front-desk and billing questions from the practice's own policies, payer manuals, and HIPAA procedures, with citations.
Consistent answers and faster onboarding
Flagship enterprise agent
Authorization & Denial Prevention Agent: Cleared Before the Patient Arrives
An illustrative solution design, not a reported client result. The agent is administrative only: it compares documentation already in the chart with published payer criteria and leaves clinical and payer decisions with a named person.
Verify coverage before the visit
The agent runs eligibility checks when a visit is scheduled and again 72 hours before it, normalizes the plan type, and routes coverage, network, and coordination-of-benefits problems to the front desk.
Decide whether authorization is required
A maintained payer-rules library returns required, not required, exempt, or unknown for each scheduled service, including Texas gold-card exemptions and the Medicare prior-authorization programs that began in Texas in 2026.
Assemble a cited packet
It retrieves only the chart elements each payer criterion allows, links each criterion to its evidence or marks it as a gap, and drafts the request and any letter of medical necessity for clinician signature.
Staff and clinicians approve
The authorization coordinator approves or edits the packet and the clinician signs any letter. Staff submit through the payer API, portal, or fax, and the agent tracks the payer's response clock.
Human approval point
Work denials and learn
It ranks returned denials by value and appeal deadline and drafts appeals that cite the payer's stated reason. Write-offs need approval, and root causes update the front-end rules.
Human approval point
What it reads
- Schedules, orders, referrals, demographics, and insurance from the practice management system
- Eligibility, claim status, and remittance transactions through the clearinghouse
- Only the chart elements each payer criterion allows, from the EHR
- Payer policies, gold-card exemption notices, and payer letters from fax and portals
What it never does on its own
- Makes a clinical or medical-necessity determination
- Writes or changes clinical documentation
- Submits, appeals, adjusts, or writes off without a named person's approval
- Sends protected health information to a service outside the business associate agreement chain
Two kinds of healthcare agents
Practice-Wide Agents and Desktop Agents
ITECS builds two kinds of agents for healthcare organizations. Each has its own owner, controls, and rollout path.
Engineered by ITECS
Enterprise hosted agents
Engineered by ITECS, connected to the practice management system, EHR, and clearinghouse, and run in a governed hosted environment with a review portal and audit trail. Practice leadership owns them: the administrator, CFO, or revenue cycle director.
- Authorization and denial prevention across payers
- Referral and fax intake routing
- Credentialing and payer enrollment tracking
Built with your team
Personal desktop agents
Built with department staff in tools a business associate agreement covers when configured correctly, such as Microsoft 365 Copilot, ChatGPT Enterprise with its regulated workspace, or Claude Enterprise with the HIPAA configuration, then rolled out through training.
- Payer-manual and policy lookups for billing teams
- Patient-friendly balance explanations from approved templates
- Summaries of payer policy updates for the front desk
Systems healthcare agents read
- athenahealth
- eClinicalWorks
- NextGen
- ModMed
- Greenway
- Dentrix
- Open Dental
- WellSky
- Homecare Homebase
- Waystar
- Availity
- Microsoft 365
Technologies ITECS evaluates and integrates; listing a product does not imply a formal partnership or a certified integration. Discovery confirms the systems and interfaces each practice runs.
Start here
Healthcare AI Readiness Assessment
ITECS starts by mapping the business case, PHI flows, vendor agreements, and governance model before recommending an agent build.
Map revenue-cycle priorities
Identify where coverage, authorization, denial, and intake work leaks staff time or reimbursement.
Trace PHI and vendor agreements
Document which systems hold PHI, which tasks need it, and whether each AI service sits inside a business associate agreement.
Rank use cases by economics
Prioritize by denied dollars prevented, staff hours returned, risk reduced, and feasibility, starting where data already arrives.
Write the governed path
Document the AI policy, minimum-necessary retrieval, approval points, Texas disclosures, and audit logging before any production build.
Governance
Built for Protected Health Information
ITECS designs healthcare AI around HIPAA, Texas privacy and AI laws, and the practice's own policies, with clinicians and staff keeping the decisions.
Administrative by design
The agents compare existing documentation with published payer criteria. Clinical judgment stays with clinicians, the same principle Texas SB 815 now applies to insurers' utilization review.
- Matches and gaps shown with citations, not medical-necessity judgments
- Gaps go to clinicians as non-leading queries
- Clinicians sign letters of medical necessity and medical-necessity appeals
- Section 1557 review for any tool near clinical decisions
PHI boundaries
Each service that touches PHI sits inside a business associate agreement chain, and retrieval is limited to the minimum each task needs.
- BAA chain documented: practice, ITECS, and each model or cloud subcontractor
- Minimum-necessary retrieval templates for each payer criterion
- Texas AI disclosures and US-only storage for EHR-derived content under SB 1188
- SOC 2 Type II applies only if Promus managed cloud hosting is selected
Security
Security for Healthcare AI Workflows
Healthcare AI touches protected health information, payer correspondence, and chart evidence. ITECS designs these systems with criterion-scoped access, immutable audit logs, and staff approval before anything leaves the practice.
FAQ
Healthcare AI FAQ
Will the AI decide medical necessity or change clinical notes?
No. The agent compares documentation already in the chart with the payer's published criteria and shows each match or gap with a citation; clinicians sign any letter of medical necessity, and staff approve each submission.
Do we need a BAA with ITECS and with the AI vendor?
Yes. Any service that creates, receives, maintains, or transmits PHI for you is a business associate, including the model provider ITECS uses as a subcontractor; ITECS includes a BAA for healthcare clients and runs agents only on services the vendor lists as BAA-eligible.
Will it work with our EHR, practice management system, and clearinghouse?
Discovery confirms your specific systems, such as athenahealth, eClinicalWorks, NextGen, ModMed, Dentrix, WellSky, Waystar, or Availity. Where a payer has no API, the agent prepares the portal or fax submission for staff to complete.
Don't the 2026 and 2027 CMS prior-authorization rules make this unnecessary?
They help, but they do not replace it. CMS-0057-F requires Medicare Advantage, Medicaid, CHIP, and federal marketplace plans to offer prior-authorization APIs by January 2027, but it does not cover state-regulated commercial or self-funded plans, assemble chart evidence, or work denials.
Do we have to tell patients we use AI?
In Texas, likely yes for many uses. TRAIGA, effective January 1, 2026, requires providers to disclose AI used in relation to health care service or treatment by the date of first service, and a plain-language notice with a link is allowed; have counsel confirm your wording.
