Making Cities Awesome: Peregrine’s Nick Noone & Ben Rudolph
In a Nutshell
Peregrine builds AI infrastructure for public safety agencies that joins existing data sources instead of collecting new data, emphasizing customer data ownership, strict governance controls, and privacy preservation over surveillance models. The company operates through forward-deployed engineers who embed with police, fire, and emergency departments to solve their specific operational problems, rejecting Silicon Valley's "move fast" approach in favor of deep institutional understanding and patient relationship-building. Their core strategy focuses on creating an institutional memory layer for cities by turning massive unstructured datasets (like cold case files) into actionable intelligence through specialized agents, while deliberately avoiding network effects that would centralize sensitive community data.
These notes were generated by AI and may contain inaccuracies.
Peregrine is built around leveraging technology to work with cities, counties, and states to impact the places people live. At the bottom of the pyramid for making cities awesome is the idea of safety - people need objective safety and need to feel safe. When that stability is there, amazing things become possible. The company aims to create backbone infrastructure at the modern city level to help preserve individuality and privacy while also bringing people together.
Peregrine is building AI that protects cities and communities while rejecting the surveillance state. The company focuses on forward deployed engineering, civil liberties, and public safety applications.
Nick Noone ran Palantir's SOCOM unit and deployed into dangerous high stakes intelligence operations in the Middle East. Forward deployed engineering means psychologically going in and owning or co-owning the customer's problem. The goal is getting all the way to the outcome with and on behalf of customers. Technology and skills are part of the answer, but the real answer is getting to the outcome at all costs, ideally three to five times faster than any other person or team could do. At the end of the day, it's the customer's win, not the company's win.
When building Peregrine, the forward deployed engineering concept had been around for a while. Feedback indicated that when really smart people from top five or top ten schools would go into complicated places like the LAPD and quickly try to understand 30 years of institutional context - not just at the data level but at the human level - the idea that moving fast is even possible in that environment is very counterintuitive to the customer. This often shows up as ego to organizations where people have dedicated their entire lives to particular roles. Silicon Valley misses that letting go of intelligence, skills, and abilities, suspending ego, and really getting into the customer's context requires high empathy and patient ways of working.
While Nick was deploying into the Middle East, Ben Rudolph was doing refugee work in Africa and India. Ben graduated school with two offers: one to work at a small startup called Airbnb, and the other with the UN Refugee Agency. He chose the UN Refugee Agency and deployed to the Sudanese border and Colombian border. He was deeply humbled and understood these types of situations and where technology can help and where it can't help.
UNHCR is a deeply disconnected organization that does phenomenal work but has data often in spreadsheets that was very difficult to make sense of. This formed Ben's thesis that a lot of problems are downstream of data problems where significant impact can be made.
After UNHCR, Ben joined Demagi, a small company building last mile healthcare solutions in underresourced places around the world. One project involved working with the Indian government on applications to help folks in rural India adhere to their tuberculosis drugs. The technology was deployed and rapidly spread throughout the country.
In 2016, Nick and Ben got together on a vacation with gymnastics teammates and had a conversation 10 years after graduating about whether they should build something together. They met in San Francisco combining Nick's national security and US government experience with Ben's humanitarian work experience. Nick had been flying between Baghdad and Middle East locations while Ben was commuting to Africa. They recognized that cities are where people live and thrive, where the majority of lives are spent.
They determined that at the bottom of the pyramid for making cities awesome is safety - people need objective safety and need to feel safe. When that stability is there, amazing things become possible. They wanted to deploy inside institutions while eliminating mishandling of information in complex organizations with access to super sensitive data. The goal was creating backbone infrastructure at the modern city level to preserve individuality and privacy while bringing people together.
Over two dozen people said no before San Pablo said yes. The company started on February 26th of 2018, though they had incorporated a month earlier. February 26 is when the San Pablo Northern California Police Department permitted them to step into the building with badges, desk space, and access to information.
They researched subject matter experts in public safety at the city scale and found an article about Brian Bouar, a young up-and-coming commander catapulting up the ranks in a paramilitary organization. An article about Operation Red Reach, one of the most iconic cross-jurisdictional gang affiliated narcotics investigative work in Northern California, featured Brian as a covert operator trying to move a stuck investigation forward. They called Brian and said they didn't know much but might have some utility, and asked if they could come in, ask questions, learn, and maybe build something over time.
The first ride along was with Oakland PD during their deep research phase. It was an overnight experience that was eye opening. Most people from Silicon Valley going on ride alongs are reporters or community interest organizations, not engineers. They received white glove service and learned it is difficult to penetrate and build trust with these organizations. The lessons signaled they had to take their time.
The company started in 2017-2018 as defund the police was gaining momentum. Nick describes himself as an experience-based learner who is tactile and wants to build bottoms-up understanding through action. This mentality leads to questioning assumptions, not in a low trust skeptical way, but to understand truth through curious adventure.
Nick has a personal affinity for working with people who feel misunderstood and helping them feel more understood. During COVID and protests, they were holding complexity - they had friends protesting police outside the women's center in San Francisco while driving across the Bay Bridge every day to work with gang homicide investigators. They were holding both truths simultaneously.
They established empathy by jumping into difficult situations rather than looking from the outside. They realized things thought to be wrong are actually complicated with many gray zones, and the deeper they go the more nuanced and textured the problem becomes.
At that time they were only working with police departments, though now they work with police, fire, emergency management, health services and more. Engineers wouldn't even respond to recruiting calls during that polarized period.
Peregrine is structurally different from data collection companies like Flock or Axon. Most companies operating in public safety since before 2018 built their business on data collection - hardware or software solutions installed in customer bases that are fundamentally about input, whether passive collection through sensors or humans inputting information directly.
Peregrine inverts that model. The company is in the business of joining disparate information to provide more secure, properly governed solutions that sit on top of pre-existing systems, helping people do better work with more precision and accuracy. Being not in the business of bringing more data to customers but building solutions that drive greater precision in how humans interact with their data flips the historical model.
The business incentive is opposite - not about maximizing data collection and creating network effects. The world is concerned about amalgamation of data and central authoritarian bodies having access to privileged information. Peregrine wants to decentralize to get to a world where they are not bringing more data to customers that they don't already have. The fundamental problem is customers can't utilize the data they have or utilize it in a way that's secure and high trust for communities served.
From day one, the company focused on permission controls, data governance, and sovereignty as necessary ingredients for deploying to high stakes institutions. They think about trust with both the community and public servants serving community members.
Each customer owns their own data - the organization serving the community means that is the community's data and the organization's data, not Peregrine's data. They provide controls and capabilities to allow secure rollout within departments and selective sharing when needed. Without these solutions, people put data in the back of a car and drive it across a city, which actually means less control.
When initially deploying AI to organizations, users often get a nice search - looking for an address and seeing everything that happened at that address in a nicely formatted way without looking through rows. As users get more used to the product, the floor gets raised for everybody across departments, enabling interesting deep types of analysis that were previously impossible.
One example involves a county in Florida that had to do over 100 water rescues. They had never done this many before and started interrogating the question with Peregrine. The AI uncovered that weather patterns had happened before, but after iterations and the agent doing deep research, it revealed these weather patterns had never occurred for three consecutive days. Those patterns create sand channels that are perfect conditions for rip currents causing issues for people in the water. This is actionable for the agency to overlay with terrain maps.
Another example involves a detective investigating a threat against a synagogue trying to find other anti-Semitic threats against two synagogues in the area. Keywords are difficult for finding this information, but AI enables semantic understanding of all accessible information, pulling out patterns of threats occurring against synagogues.
95% of the work is what happens before the user types in the question - all the preparation to get the AI ready to answer accurately and cite questions accurately. The company spends significant engineering points on data preparation getting the AI ready. They have many agentic use cases.
AI allows writing software rapidly when the cost of generating software is virtually zero. The platform provides deployment teams with security and governance controls and APIs needed, allowing them to write the world with software. Deployment strategists can create solutions like hurricane simulators using AI, enabling rapid innovation cycles.
The transcript cuts off mid-sentence at: "Can they go from idea"
Peregrine enables customers to take ideas from concept to execution through partnership with their deployment team. While some customers leverage the platform independently, most development work comes through the deployment team providing technical expertise. This dynamic is already changing as customers gain confidence to embrace innovation and let the best method win.
Peregrine provides domain expertise and deeply understands law enforcement, fire departments, and EMTs, then provides technology to enable their most important missions. Technology can change over time, but the core strategy remains unchanged. Most capabilities are downstream of high-quality data - charts, agents, and future technologies are only as good as the underlying information accuracy and access.
Peregrine has an agent that enables deployment strategy teams to integrate customer-owned data agentically. They've integrated tens of thousands of data sets across customers and built an evaluation set to deterministically evaluate agents for completeness and correctness on integrations. About 90% of Python notebook integrations are now written by agents with deployment team oversight. These agents run for hours, analyzing databases, understanding ontology, and splitting work into sub-agents that communicate back to an orchestrator agent.
The first operational agent built was a cold case agent. These investigations involve uploading 200-300 gigabytes of data including videos, audio, images, and PDF files. Some police departments still have paper records in large boxes with attached CDs. The cold case agent was developed in partnership with a customer who had exonerated a wrongly convicted individual. They provided all case evidence and data, and the agent was trained to run for 30-60 minutes to glean insights, eventually reproducing the detectives' original results.
The cold case agent is now used in several US departments. In a Wisconsin county case involving 300 gigabytes of data, the agent identified and placed a suspect not just at the crime scene but at the body discovery location. This was accomplished using a few cell phone call detail records scattered among extensive data, demonstrating the agent's ability to surface critical evidence from massive datasets.
Trust is maintained by not taking credit or publicly trumpeting achievements with these organizations. Being quiet professionals and empowering customers is why they gain access to subsequent problems. This approach is particularly important given high public distrust in AI, and these success stories represent powerful examples of AI improving lives and communities.
Network effects in public safety organizations are strong. Peregrine protects against organizations potentially accessing data outside legal boundaries. The approach preserves data sanctity and ownership for each agency while building connective tissue and interoperability. Data governance and permissioning logic in AI contexts become critically important.
Peregrine operates as an "anti-network effects business" by building high-integrity, transparent solutions while letting go and avoiding overly rapid growth. The business practices letting go by making solutions transparent to customers and their constituents.
Peregrine does not impose technology decisions on customers. The approach is to help customers understand their operational context, bring to light all considerations, and help usher in the right way of thinking about deploying technologies. Most public safety agencies and communities choose not to implement facial recognition, with decisions varying by subjective preference. The company follows customers and the law while creating clarity around institutional norms.
Cities call each other for advice on technology decisions. The value comes from helping organizations streamline access to the best possible information so they can make optimal decisions about technology adoption.
Peregrine delivers technology to underdog markets that Palantir doesn't serve with their eight-figure minimum contracts. Building a vertically integrated tech stack from network access through permissioning, governance, ETL, pipelines, ontology, UX, and APIs requires audacity. The company takes time to deliver solutions all the way to outcomes and owns responsibility with customers for achieving results, which initially seemed unscalable but resulted in a platform that scales effectively.
The deployment strategists are selected and trained to be smart, humble, and deeply embody the mission. The company aligns on character qualities including pursuit of excellence. Environmental factors include trusting individual contributors and using the analogy of sending engineers into a "dark cave" with a tool belt, maintaining integrity and principles while achieving objectives. Innovation happens at the farthest fringes of the organization, with engineering, product, and design fed through the forward deploy team's innovation process.
Forward deployed engineers function as R&D and growth rather than cost centers. After landing in a customer base, they conduct within-customer pilots, sprinting on additional use cases. Landing and expanding in customer bases happens through leading from the front via forward deployed engineers.
New products emerge directly from field work rather than theoretical planning. One example involved typing code across the Bay Bridge to meet an urgent detective request from Aaron Blazale. The forward deploy motion impacts products in two ways: identifying basic primitives the platform needs, and creating an innovation lab where deployment strategists build unique capabilities for specific customers.
For the longest time, Peregrine had no ability to edit specific fields and properties. A forward deployed engineer created an integration based on comments written on objects, which would be read by Peregrine to update properties. This hacky solution signaled the need to implement proper editing functionality. The approach prioritizes achieving objectives over perfect technology implementation.
A deployment strategist built a hurricane simulator in Peregrine. Another built a system integrating 911 call times and budgets that could estimate how many people would be impacted by placing a fire department in a city. This was built by someone who had been at the company for only a month.
The vision is for Peregrine to become the institutional memory layer for 10,000 cities. This requires an operating model focused on delivering technology infrastructure that empowers organizations to work with their data as they would like and are required to do. Every jurisdiction is unique, and preserving that uniqueness is considered beautiful. The path requires serious integrity, moral compass, and core values in actions and ways of working.
Supporting state, county, and city-level public safety was historically attempted by major organizations that failed because delivering tailored solutions at affordable price points was never possible. Creating efficiency to drop prices by orders of magnitude has earned the right to attempt the institutional memory vision. The institutions hold the power, not the technology provider.
Keep Sequoia Capital in your library
Save the videos and channels worth coming back to, and find them again in one place.
