Arash Jafari

An AI that looks over your shoulder - in the good way

Somewhere in a skilled nursing facility tonight, a director of nursing is staring at a CMS Form 2567. It is a list of citations, written in the dense regulatory dialect of F-Tags and severity scores, and it demands a response called a Plan of Correction. Get it wrong and the facility risks fines, oversight, or worse. Get it right and you have spent the small hours of the morning doing it instead of sleeping. This is the unglamorous machinery of American long-term care, and it is exactly where Arash Jafari decided to spend his twenties.

Jafari is the CEO and co-founder of Clearpol, a San Francisco company that describes itself, with refreshing plainness, as "your AI ally in post-acute and long-term care." The pitch is not science fiction. Clearpol's software reads an operator's policies, procedures and clinical data, compares them against the live tangle of state and federal regulation, and flags where reality is drifting from the rules. When a deficiency does land, the AI drafts the Plan of Correction. The company says it has written more than 3,000 of them, and that operators report a 100% acceptance rate on what gets submitted.

What makes the story worth telling is not the product alone. It is the route Jafari took to get here. He did not come up through hospital administration or health-insurance back offices. He came from rocket science.

From satellites and spacecraft to survey readiness

Before Clearpol, Jafari's resume read like a recruiting brochure for the space program. At NASA's Langley Research Center he worked as a graduate intern in deep learning, building models to predict severe weather events from satellite imagery. At NASA's Jet Propulsion Laboratory - the lab that lands rovers on Mars - he was a graduate software engineer working on interplanetary physics and robotics simulation. Before that, at UC Irvine Health, he was a junior specialist coordinating a brain-MRI study of high school football players. Weather, spacecraft, neurology. Three domains, one constant: take messy real-world signals and teach a machine to find the pattern that matters before it becomes a problem.

That through-line is the whole company. Severe weather, a failing thruster simulation, an unflagged infection trend in a nursing home - to an engineer, they rhyme. Each is a system generating data faster than a tired human can monitor it, and each rewards the person who can see the risk early. Jafari simply moved the same instinct from the upper atmosphere to the nurses' station.

The co-pilot, not the autopilot

Jafari is careful about the metaphor he uses for the technology. He calls Clearpol's AI a "co-pilot" - something that looks over the shoulder of staff and administrators, surfacing the regulatory insight they need rather than replacing their judgment. In an industry understandably wary of black boxes making care decisions, the framing matters. The human stays in command. The machine handles the reading, the cross-referencing, the 2 a.m. document drudgery that no one became a nurse to do.

The market he chose is enormous and chronically underserved by good software. Skilled nursing facilities operate under some of the heaviest regulatory loads in any industry, while running on thin margins and chronic staffing shortages. Compliance is not a side quest; it is existential. Yet much of the work is still done by hand, in spreadsheets and binders, by people who would rather be at the bedside. Clearpol's bet is that the most humane thing AI can do here is not diagnose patients - it is to give exhausted humans their nights back.

By 2025 the bet was being validated in public. Clearpol was welcomed as a Preferred Product and Service Provider by the California Association of Health Facilities, an industry stamp of trust. Then, in November, came the headline that put the company on the map nationally.

A first-in-the-nation launch

At the AHCA/NCAL 2025 Convention, Clearpol and the Health Care Association of New Jersey unveiled the HCANJ Survey Intelligence Hub - billed as the first state-level, AI-powered survey intelligence platform in the country. The idea is to give an entire state's providers a shared dashboard: track facility and unit-level trends across F-Tags, severity and repeat citations; benchmark against state and national norms; prioritize fixes in care planning, supervision, medication management and infection control. Where a single facility once saw only its own deficiencies, an association can now see the weather system forming across all of them. The engineer who once predicted storms from orbit built, in effect, a radar for regulatory risk.

"This collaboration brings cutting-edge analytics to the frontlines of care," Jafari said of the partnership. It is the kind of sentence that could read as boilerplate, until you remember the person saying it spent his early career making analytics that had to survive contact with actual hurricanes and actual spacecraft.

The quiet credentials

Jafari keeps a low public profile, which is itself a little unusual for a venture-backed founder. He graduated Magna Cum Laude and was inducted into Phi Beta Kappa. He pursued graduate work in data engineering at the University of California, Riverside, where his earliest surviving piece of online writing is a 2019 Medium explainer with the wonderfully unglamorous title "How Search Engines Work." He is fluent in Farsi and has working knowledge of Spanish. He co-founded Clearpol in 2020 and has been its CEO since the beginning.

There is no flashy founder mythology here, no manufactured origin tale. What there is, instead, is a coherent obsession: take a high-stakes domain drowning in its own data, and build the system that lets a human catch the important thing in time. He has now applied it to the weather above us, the machines we send to other planets, and the people we care for at the end of their lives. The last of those is the least photogenic and arguably the most consequential.

Clearpol today reports more than 900 facilities served, millions of patient-days analyzed, and a security posture - HIPAA compliance, SOC 2 Type II, SOC 3 - that signals it intends to be infrastructure, not a novelty. The company has raised seed funding and built out a full suite: Clinical AI, the POC Writer, Insights Pro, Regulatory Intelligence, and the Survey Intelligence tools now reaching the state-association level.

Ask what Arash Jafari is really building and the honest answer is not "compliance software." It is time. Specifically, the time a director of nursing gets back when the form fills itself, the time a resident gets when a risk is caught early, the time an entire industry saves when the fine print finally has something fast enough to read it. He spent his early career watching things from very far away - storms from orbit, robots on other worlds. The work he does now is about watching the thing right in front of us, the one we keep being too tired to see.