There’s a process in healthcare that costs billions of dollars a year, takes three to four months to complete, still relies on fax machines in 2025, and affects every doctor, nurse, and therapist who wants to see patients and get paid by insurance.
Most people have never heard of it.
It’s called credentialing. And for decades, it barely changed.
What credentialing actually is
Before a clinician can bill insurance, someone must verify that they are who they claim to be and meet a long list of requirements. Medical school and residency. State licenses. Board certifications. DEA registration. Malpractice insurance. Federal exclusion lists.
This scrutiny is necessary. Healthcare depends on trust.
The problem isn’t what gets verified. It’s how.
Each insurance payer requires its own application. Each state has its own licensing rules. Each payer wants the same information formatted differently. A provider joining a practice that accepts 10 insurance plans across three states may need to submit dozens of separate applications.
The underlying data already exists. State medical boards maintain licensure records. The DEA tracks registrations. Federal agencies maintain exclusion lists. Credentialing organizations like CAQH collect standardized provider profiles.
Yet for years, verifying this information meant manual work: phone calls, paper forms, scanned documents, fax transmissions, and endless follow-ups. Information was re-entered by hand into payer portals that assumed the presence of a human operator on the other side.
The result is a familiar timeline across healthcare organizations: 90 to 120 days from hire to insurance billing.
During that window, providers often can’t generate reimbursable revenue. Practices either delay care or absorb the cost. Across the system, the financial impact totals billions of dollars in lost revenue and administrative spending each year.
Why didn’t the software fix it?
Credentialing software has existed for a long time. Dashboards improved visibility. Digital forms replaced paper. Status tracking became cleaner.
But the underlying work didn’t change.
Someone still had to gather data. Someone still had to fill out payer applications. Someone still had to log into portals and click through forms. Someone still had to notice when an application was rejected, often weeks later, and start over.
Error rates remained high. Missing documents, inconsistent data, expired licenses, or mismatched identifiers frequently triggered rejections. Each rejection added weeks to the process.
Most legacy platforms digitized a manual workflow without eliminating it. They made credentialing easier to manage, but not faster or more reliable.
What changed
The inflection point came when a new generation of companies treated credentialing as what it actually is: a data and verification problem, not a form-tracking problem.
The logic of credentialing is deterministic mainly. A license is valid, or it isn’t. A provider appears on an exclusion list or doesn’t. A document is current or expired. The sources of truth are known.
This is exactly the kind of work modern AI systems handle well.
AI-native credentialing platforms can query thousands of primary sources in parallel: state licensing boards, federal databases, board certification organizations, and more. What once required weeks of phone calls and follow-ups can happen in minutes.
Provider data can be mapped automatically to each payer’s specific requirements, generating dozens of distinct applications without manual re-entry. Browser automation can submit applications through payer portals that don’t offer APIs, replicating the actions of a human user actions at machine speed.
Critically, validation happens before submission. Missing fields, inconsistent data, or expired documents are flagged early, reducing rejections and rework.
The result is a step-change in outcomes: credentialing measured in days instead of months, and first-pass approval rates significantly higher than legacy approaches.
One example
Assured is one of the companies applying this approach.
It was founded in 2024 by Rahul Shivkumar and Varun Krishnamurthy, who previously built and scaled a multi-state virtual care business. As they expanded across states and payer networks, credentialing became a recurring bottleneck: slow, fragmented, and operationally fragile.
Assured’s platform verifies credentials across 2,000+ primary sources simultaneously and automates payer submissions end-to-end. Most submissions are completed within 48 to 72 hours of receiving provider information, with first-pass approval rates reaching 80 to 95 percent depending on payer mix.
In early 2025, the company received NCQA Credentials Verification Organization (CVO) certification, a widely recognized industry standard for quality and compliance. It has since raised $6 million in venture funding from First Round Capital to expand its platform.
The pattern is consistent: when the manual work disappears, timelines collapse.
Why this matters beyond credentialing
Credentialing is a narrow problem, but a revealing one.
Many industries rely on operational processes that feel immutable. They involve multiple counterparties, fragmented data sources, regulatory oversight, and legacy systems. Software helps organize the work without actually doing it.
AI changes that dynamic. When systems can execute tasks end-to-end, retrieving data, validating it, and interacting with legacy interfaces, the bottleneck shifts from human capacity to system design.
Healthcare credentialing is an early example. Similar transformations are already underway in insurance claims processing, compliance verification, legal review, and financial underwriting.
The shift isn’t about AI assisting humans. It’s about AI operating systems that were previously assumed to require constant human intervention.
Where it goes from here
Credentialing teams across healthcare report high turnover and burnout. The work is repetitive, manual, and unforgiving, precisely the kind of labor software promised to eliminate but historically did not.
Organizations that move to AI-native approaches rarely revert. The gains in speed, reliability, and recovered revenue are too large to ignore.
Fax machines haven’t disappeared from healthcare yet. But they’re handling fewer applications every month.
The multi-billion-dollar paperwork problem isn’t solved everywhere. But it is solvable. And for the organizations that have already made the shift, the 90-day wait is no longer part of the job.
