Drone First Response And Dispatch Triage
Sources: 1 • Confidence: Medium • Updated: 2026-04-11 18:05
Key takeaways
- Garrett Langley says Flock has a team of constitutional attorneys to review product ideas for potential Fourth Amendment issues.
- Garrett Langley says early in Flock's history, a major procurement sales blocker was meeting a requirement for license plate capture performance at about 120 to 150 mph.
- Garrett Langley argues facial recognition should be allowed for severe crimes while restricted for minor offenses rather than being subject to blanket bans.
- Flock OS can ingest 911 calls in real time, correlate them to citywide camera feeds (including non-Flock cameras), run attribute searches such as clothing, and push relevant video to officers.
- Garrett Langley says Flock receives the vast majority of its payments by check (roughly 80% to 90% or more) and says none of their customers pay via ACH.
Sections
Drone First Response And Dispatch Triage
- Garrett Langley says Flock has a team of constitutional attorneys to review product ideas for potential Fourth Amendment issues.
- Garrett Langley says Flock's policy preference is to dispatch drones in response to triggers like 911 calls, gunshot detection, or a stolen car rather than flying around looking for suspicious activity.
- Garrett Langley says that during flight from launch to destination, the drone camera defaults to pointing at the horizon to avoid looking into private areas, while allowing an operator to tilt down if desired.
- Garrett Langley says Flock dispatches drones from charging docks when needed rather than keeping them continuously in the air.
- Garrett Langley says the drone dock is engineered to keep lithium-ion batteries within temperature limits while charging and operating across adverse weather conditions.
- Garrett Langley says Flock's drone development approach is to build the best camera payload first and then design the airframe and flight system around that payload.
Hardware Plus Field Ops As A Scale Moat And Bottleneck
- Garrett Langley says early in Flock's history, a major procurement sales blocker was meeting a requirement for license plate capture performance at about 120 to 150 mph.
- Garrett Langley says Flock's camera system keeps the camera effectively offline until a vehicle is detected in order to reduce power usage and operating cost.
- Garrett Langley says Flock operates at large physical-installation scale and pulled about 77 permits per day last year.
- Garrett Langley says hardware demand forecasting at Flock requires 12 to 18 months of lead time and must be done at the geographic level due to first-party install and maintenance.
- Garrett Langley says Flock uses a continuously running radar attachment tilted backward to prime the camera before a vehicle enters view on busy roads with tight angles of incidence.
- Garrett Langley says Flock views taking a picture as its most expensive operation, followed by sending data to the cloud, followed by keeping the computer powered on.
Privacy Legal Constraints As Product Requirements
- Garrett Langley says Flock has a team of constitutional attorneys to review product ideas for potential Fourth Amendment issues.
- Garrett Langley argues facial recognition should be allowed for severe crimes while restricted for minor offenses rather than being subject to blanket bans.
- Garrett Langley says Flock's policy preference is to dispatch drones in response to triggers like 911 calls, gunshot detection, or a stolen car rather than flying around looking for suspicious activity.
- Garrett Langley says that during flight from launch to destination, the drone camera defaults to pointing at the horizon to avoid looking into private areas, while allowing an operator to tilt down if desired.
- Garrett Langley says Flock does not use facial recognition because it is too controversial.
- Garrett Langley says Virginia legislation set a 21-day retention period, mandated formal auditing, limited use to criminal investigations, and restricted participation with the federal government.
Real Time 911 To Video Intelligence Workflows
- Flock OS can ingest 911 calls in real time, correlate them to citywide camera feeds (including non-Flock cameras), run attribute searches such as clothing, and push relevant video to officers.
- Garrett Langley describes a case where the system helped go from a 911 call to an arrest in about 17 minutes by searching for a suspect wearing white Converse sneakers.
- Garrett Langley says AI can take 911 calls during surge events and can pre-build investigation context from the call for detectives.
- Garrett Langley says Flock builds investigator-facing agents that pre-complete time-consuming case busywork so investigators can focus on judgment-heavy tasks.
- Flock is working on real-time analysis of 911 calls.
Procurement And Payments Friction In Govtech
- Garrett Langley says Flock receives the vast majority of its payments by check (roughly 80% to 90% or more) and says none of their customers pay via ACH.
- Garrett Langley says government procurement is exceptionally slow and laborious, with most purchases going through an RFP process.
- Garrett Langley asserts that many government RFPs are written to fit a single vendor, making the process a delay mechanism rather than a true competitive bid.
- John Collison says Stripe can provide a mailing address to accept customer checks and convert them into digital funds, and Garrett Langley indicates Flock uses Stripe for this check processing.
- Garrett Langley says some government spend thresholds allow a police chief to purchase roughly $25,000 to $50,000 without going through the full procurement process.
Watchlist
- Garrett Langley says Flock has a team of constitutional attorneys to review product ideas for potential Fourth Amendment issues.
Unknowns
- Are the drone response-time claims (e.g., ~68 seconds vs ~7.5 minutes) representative across jurisdictions, incident types, weather, and regulatory constraints, or are they best-case examples?
- What is the full distribution of outcomes for the 911-to-video workflow (time from 911 intake to identification, to dispatch decision, to arrest/clearance) and how does it compare to pre-deployment baselines?
- How is the 'helped clear just over one million crimes' metric computed at case level, and what fraction is attributable to decisive evidence vs incidental involvement?
- Is the claimed public availability of audit trails real in practice (who can access what, under what process), and how often are audits actually performed and enforced?
- How do different states' cloud legality and retention/audit requirements affect product deployment architecture (cloud vs hybrid), feature availability, and sales cycle duration?