- Daniel Francis is the founder and CEO of Abel Police, a startup that uses AI to turn body camera footage into police reports, freeing officers from a task that consumes roughly one-third of their time. He got the idea after going on dozens of ride-alongs and seeing how much of policing is consumed by paperwork, and after a personal experience helping a friend escape an abusive relationship where police took 40 minutes to respond. The company now serves 23 police departments and is expanding into related tools.
How Abel Police works and why it matters
-
The core product (Abel Writer): Takes unredacted body camera footage, transcribes it, and automatically fills out both the structured data fields and the narrative section of a police report. Officers review and sign off rather than writing from scratch.
- Police spend enormous time on reports — in some agencies over half their shift — because every detainable interaction requires documentation, even minor ones like someone looking into a store window.
- Before body cams, report accuracy was essentially unknowable. Body cams created a “trustless” system where footage can exonerate officers and the public can verify what happened.
- Officers overwhelmingly hate writing reports. None became police to write. The paperwork burden actively discourages proactive policing because the marginal cost of stopping someone is a 30–40 minute report.
-
Why Axon’s entry helped but wasn’t enough: Axon (a $60B publicly traded company) announced a similar product called Draft One, which legitimized the space. But Axon’s approach was crude — basic transcription fed through ChatGPT — and performed poorly in noisy real-world conditions like street encounters with overlapping voices, road noise, and drunk subjects.
- Abel’s system is far more layered and sophisticated, handling both structured data (the “face sheet” with hundreds of fields per person) and narrative generation.
- Daniel considers Abel’s current output to be 95th-percentile report writing — better than most officers — because officers in the field are managing scene security, investigations, high-risk individuals, and bystanders simultaneously and often miss incriminating statements that Abel surfaces.
Getting into police departments
-
The sales challenge: There are 18,000 police agencies in the US, each with different software, different report requirements, different city attorneys, and different contract demands. Transaction costs are high.
- Daniel’s first approach was driving around the East Bay handing out flyers his girlfriend made. Ten agencies rejected him before Richmond, California said yes.
- The breakthrough came at a police conference (Cal Chiefs), where he met a captain at Richmond who watched a demo — a fake body cam video Daniel made with his phone strapped to his chest warning a friend for littering — and said, “That was bitching. You got to do that here.”
- Richmond was already familiar with the concept because they used to have records clerks transcribe reports over the phone before budget cuts eliminated those positions, so the AI approach felt like a return to something familiar.
-
Why the hardest-to-serve agencies are the last to adopt: Agencies with the worst staffing shortages and most need are often in cities where city councils are hostile to police, making procurement politically difficult. Daniel was told by a CEO of a larger police tech company: “The places that need you most are the last to get you.”
- The playbook is to start with less bureaucratic, adoption-ready departments and build from there — similar to how Flock (automated license plate readers) spread rapidly through low-crime, well-funded areas in the South before reaching major cities.
- The South is less fertile ground for Abel right now because many officers who left difficult political environments in blue states moved there, easing staffing pressure.
-
Criminal Justice Information Security (CJIS): Handling body cam footage requires CJIS compliance — a niche data security standard that varies by state. ChatGPT and general AI tools are not CJIS-compliant. Becoming compliant was a massive undertaking that required Abel to figure it out largely on their own, but it also creates a moat against general-purpose AI tools.
Ride-alongs and understanding the job
-
Daniel has been on 32 ride-alongs and considers them essential to building the product. He spent months during Y Combinator writing code during the day and doing ride-alongs at night, sometimes staying over when officers were held past shift.
- The repeated access built trust. By the fourth or fifth ride-along with the same officer, they were venting about their sergeant and sharing real pain points — something no interview could replicate.
- Policing is far more complicated than outsiders think. Officers simultaneously function as security guards, investigators, and social workers while managing volatile people, bystanders, and their own safety.
- The worst type of cop is a lazy one, because it burdens fellow officers. Unions will even let persistently lazy cops get fired.
- Policing is fundamentally a service job. Daniel’s favorite prototype of a police officer is a former bartender — someone who can be personable but also physically remove someone when needed.
-
Memorable ride-along experiences:
- A 47-minute police chase hitting 130 mph on the highway and 50 mph in the narrow, winding Oakland Hills. Daniel called it the best day of his life.
- A man getting hit by a train, described as “a ball of flesh” where you couldn’t identify body parts.
- Welfare checks on people dead for days, where the house smells “sickly sweet” — a smell the brain briefly registers as pleasant before recognizing what it is.
The broader context of American policing
-
Post-2020 staffing crisis: After George Floyd, policing became less popular as a profession, creating national shortages. Agencies now compete for lateral transfers and are lowering hiring standards — accepting candidates with records they would have rejected a decade ago. Daniel sees this as a nightmare that will generate future problems.
- The real solution is leveraging existing officers to do more through technology — giving them “an extra arm” — rather than lowering the bar for new hires.
- Overtime has shifted from a reward to a curse, and unions are less resistant to productivity tools because their members are exhausted from mandatory overtime.
-
Crime statistics are unreliable: Daniel is deeply skeptical of reported crime numbers. Common manipulation tactics include downcharging felonies to misdemeanors, and revising past quarters’ numbers after the fact to make current numbers look better. This makes crime rate a noisy signal for measuring technology impact.
- A better KPI is measured report-writing time. Abel’s pilot with Belmont showed a 40% reduction, and Daniel believes they can cut it by half.
-
The human cost of policing: Officers regularly face people trying to grab their weapons, sudden violence, and verbal abuse as a daily norm. The contrast with white-collar work — where the worst conflict is a passive-aggressive email — is stark.
- When something goes wrong, cities pay multi-million dollar settlements out of police budgets, forcing the department to operate with fewer resources precisely when scrutiny is highest.
- Daniel’s personal experience helping a friend escape domestic violence — where police took 40 minutes to arrive while an abusive ex banged on the door — gave him a visceral understanding of what the absence of police feels like.
Company strategy and expansion
-
Current scale: 23 police departments. The goal is to expand aggressively, using the saved officer time as the core value proposition.
- Research suggests that for every 100 officers on Abel’s platform, roughly one person’s life is saved per year through increased police presence, more guns off the streets, and more engagement in violent communities.
-
New product (Abel Citizen): A replacement for CopLogic (made by Lexus Nexus), which allows the public to file minor crime reports online for insurance purposes. CopLogic is widely considered terrible — it looks like it was built in 1997, breaks on mobile, and generates unnecessary calls to police.
- Lexus Nexus refused to sign Richmond’s sanctuary city agreement (refusing to promise not to work with ICE), forcing Richmond to cancel their contract. Daniel expects many California agencies to follow.
- Abel Citizen will use a chat interface powered by an LLM to guide citizens through reporting, asking clarifying questions to capture the legally necessary components of a crime (e.g., was the car locked or unlocked? Was the item stolen from your person or left somewhere?) — saving both records departments and officers follow-up time.
-
Why police tech has been neglected: For ideological reasons, talented technologists have avoided the policing industry for 10–15 years, leaving officers using appallingly outdated software. Daniel sees this as both a massive opportunity and a moral imperative — improving police technology is one of the highest-leverage ways to improve American cities and save lives.
Personal background
-
Daniel previously sold a fitness app and experienced what he calls “postpartum depression” after the exit — waking up the next day with no sense of purpose. He traveled to China on a $500 mistake fare (business class) and then moved there for three months working remotely.
- He spent two months backpacking through Japan with a 13L backpack, but found the experience lonely — describing how being surrounded by people who are unfailingly polite but never casual or authentic is a form of profound isolation. “You have never been alone until you are alone in Japan.”
- He has a sheep as a mascot, referencing both the “sheepdog” meme in police culture and the biblical story of Abel sacrificing a sheep.
-
Hardest thing overcome: At 15, Daniel contracted acute disseminated encephalomyelitis, which paralyzed him from the waist down. He couldn’t feel his legs, couldn’t wiggle his toes, and was airlifted to Jacksonville for diagnosis. He had to learn to walk again over months. The experience led him to Buddhism and a deep appreciation for physical capability.