How Audity pivoted from object detection to behavioral AI for law enforcement

May 21, 2026 · Predictable Revenue Podcast
🎧 PodShort 3 min squeezed to 2 AI SprinklerAS AI / ML
Episode artwork
Thomas Aalflin
Co-founder and CCO at Audity
Predictable Revenue Podcast
3 min squeezed to 2
Full episode from Predictable Revenue Podcast
Quotable Moments

I used to read the exact same thing over and over and over again and now I get I get bored and I try and change it up just a little bit every now and then.

Colin StewartHost of The Predictable Revenue Podcast
01:03

But what if you had some founders that have been there before, been there, done that on the revenue side. That's our founder coaching program...

Colin StewartHost of The Predictable Revenue Podcast
01:18

We thought of weird, oddity, and Audity AI. Name wasn't taken, we like the name. Um, it kind of referred to detecting weird or out of the ordinary behaviors.

Thomas AalflinCo-founder and CCO at Audity
02:04
Key Insights
  • The name 'Audity' originated from the idea of detecting 'weird' or 'out-of-the-ordinary' behaviors like violence or aggression captured by cameras, reflecting their core mission.
  • Audity's initial product market fit hypothesis was to apply their AI to public government and agencies (like law enforcement) to detect high-intensity behaviors on surveillance cameras.
  • Early customer validation involved extensive cold-calling to police departments and local surveillance centers, often calling the general police number multiple times to be passed through to the right person.
  • During a visit to a surveillance center, it was observed that law enforcement officers were reactive to incidents (checking footage *after* a call) rather than proactive, highlighting a clear need for AI to detect issues in real-time.
  • The initial AI algorithm, which was a basic object detection system, was refined into a 'very novel behavioral detection AI' based on direct feedback from potential customers.
  • Selling to the police in the Netherlands presented numerous challenges, including navigating laws, regulations, and multiple municipal entities, but they successfully implemented their solution in several cities.
  • The primary objective of their AI was to detect high-intensity behaviors such as aggression and violence on camera footage, which was identified as a key need for law enforcement.
Metrics Mentioned
  • 20 times in a day (Effort involved in cold-calling to reach surveillance center personnel for customer validation.)
  • hundreds of cameras (Ideal customer types like prisons and public transport often have a large number of cameras, which aligns with their license-per-camera pricing model.)

RevBots.ai View:

  • Classic AI Sprinkler move: bolting behavioral detection onto existing surveillance systems.
  • Shows how early-stage founders can force market fit through brute-force outreach.
  • Niche vertical focus (law enforcement) creates defensible wedge for AI startups.
  • ARM playbook would integrate behavioral signals with other public safety data streams.
🎧Full Episode:Predictable Revenue Podcast →