Companies can get an AI agent running in an afternoon. The hard part is putting that agent in a room with counsel, compliance, and capital without the conversation collapsing into theory.
That is why C-LAW-D matters as a case study. On Monday, May 4, 2026, Purple Horizons was the tech partner for an invite-only convening at Holland & Knight in Miami, co-hosted by Florida State University College of Law and Lab22c. The room mixed operators, counsel, investors, and scholars. The brief was simple: understand Claude, sovereign AI, and the law in the same afternoon, with a working agent actually in the room.
The discussion itself stayed off the record. That was the point. What we can publish is the design: how a legal room was built so people could see a live agent before they argued about who is liable for one.
Why a legal room needs a live agent
Most AI-and-the-law events are panels about hypotheticals. Someone describes a chatbot. Someone describes a lawsuit. Nobody in the room has touched the system they are regulating in their heads. Counsel leaves with notes. Operators leave unconvinced. The gap between those two groups is where companies get stuck.
Purple Horizons had already seen the demand on the builder side. In late February, OpenClaw Miami packed The LAB in Wynwood. Roughly 200 people made it in. Another 500 sat on the waitlist. Inc. Magazine covered the night as a signal that business leaders wanted to run agents on their own hardware, not only in someone else's cloud.
That event was for people ready to install. C-LAW-D was the next conversation: what happens when those agents leave the meetup and show up inside a firm, a fund, or a government-adjacent workflow. The legal questions are not abstract once the agent can send a message, file a document, or act while nobody is watching.
10 AI signals that matter for your business. Every week. No hype.
C-LAW-D stood for a specific intersection: Claude, sovereign AI, and the law. Lab22c and FSU College of Law co-produced it. Holland & Knight hosted. Purple Horizons ran the technology. The guest target was a strategic 50, invited from a list of about 100. Runtime was two and a half hours, 1:00 to 3:30 p.m., business casual, Miami smart.
The audience mix was the product. Legal (Holland & Knight attorneys and clients). Academic (FSU Law faculty and scholars). Tech (founders and builders). Capital (investors and operators). Policy (people who actually have to live with sovereign-AI rules). The room was curated so no single tribe could talk past the others.
The program had two panels with a live demo in between. Panel one put operator and capital reality on the table first: how founders and investors bet on sovereign AI before the rules exist. Panel two was the legal anchor: who controls the machine, where jurisdiction sits, and how liability works when an agent acts. Opening remarks came from Lab22c, the FSU College of Law dean, and Holland & Knight. Purple Horizons did not open with a slide. We opened with hardware.
What Purple Horizons ran
Arrival: OpenClaw show and tell
Guests arrived at 1:00 for 45 minutes of cocktail-style networking. Ralph Quintero and Gianni D'Alerta ran a dedicated OpenClaw station. One-on-one, not a pitch. Counsel could ask what the agent actually does. Operators could show what they had already built. The greeter problem from OpenClaw Miami did not apply here. This was a small room. The job was signal, not throughput.
The stack was the same local-first story we use with companies: the agent runs on hardware the operator controls, channels like Telegram or Slack are optional, and data does not leave the machine unless someone chooses that. For a legal audience, that is not a feature list. It is the beginning of a custody conversation. Who holds the keys. Who can see the prompts. What happens if the laptop walks out of the office.
The 10-minute bridge
After the first panel, Purple Horizons took the full room for a 10-minute live OpenClaw demo. The run of show called it a proof point, not a pitch, and that was the right order. A demo before the operator conversation is theater. A demo after people have named the risk is a different object. The room had just talked about compliance, cybersecurity, and unpriced liability. Then they watched an agent do something.
That sequence is the same logic we use in an AI workshop for companies and in a buildathon. Foundation first. Then the build. Nobody leaves having only watched.
Moderating the governance panel
Gianni and Ralph moderated the second panel, on constitutional law and institutional governance. The framing question was who controls the machine. After a live agent, that question is no longer academic. Jurisdiction, speech, liability, and how you train the next generation of lawyers all land differently when the room has just watched software act.
We will not recap who said what. The case study is the role: operators in the moderator chair, not a vendor in the back of the room. That is how Purple Horizons shows up. Founders in the room. The stack on the table. Counsel asking questions of a thing they can see.
What the day was built to prove
Sovereign AI is not only a nation-state slogan. For a company it means something plainer: who hosts the model, who sees the data, and who is on the hook when the agent acts. A cloud chatbot and a local agent are different legal objects. Treating them as the same product is how teams walk into a compliance problem they did not budget for.
The second proof is about rooms. You cannot sequence this work as "tech first, legal later." If operators and counsel never sit together, the company gets two strategies. One ships. One blocks. C-LAW-D forced the overlap on purpose: capital and operators first, a live demo as the hinge, then the legal frame while the demo was still warm.
The third proof is the same one in the Palmer Trinity buildathon and the TradeStation training work. People change what they believe when they build or watch the thing, not when they hear a keynote about it. Schools, fintech teams, and law firms are different institutions. The method is not.
What companies can take from C-LAW-D
Put the agent in the room before the policy conversation. A governance memo written against a slide deck will not survive first contact with a working agent. Let counsel see the system. Then write the rules.
Mix operators with counsel in one sitting. Sequential meetings produce sequential truths. A two-and-a-half-hour room with both groups in it produces a shared picture of risk, which is the only picture a company can actually operate on.
Keep the first version small and off the record if you want honest talk. C-LAW-D was invite-only on purpose. Fifty people who will say the quiet part is more useful than 500 people performing expertise. Public recaps can wait until the company has a position it is willing to sign.
Leave with a next step, not a deck. For Purple Horizons that next step is the same one we run with operating companies: a workshop on prompting and safe use, then a short build on real work. The legal room is not the end of the work. It is how you start it without walking into a wall.
FAQ: C-LAW-D and AI agents in the legal room
What was C-LAW-D?
C-LAW-D was an invite-only convening on May 4, 2026 at Holland & Knight in Miami. FSU College of Law and Lab22c hosted. Purple Horizons was the tech partner. The day sat at the intersection of Claude-powered agents, sovereign AI, and legal architecture.
Why was the event off the record?
Because the useful version of this conversation includes risk, liability, and what people will not say on a stage. Off the record is how you get operators and counsel talking to each other instead of performing for an audience. This case study covers the format and Purple Horizons' role, not quotes from the room.
How is this different from OpenClaw Miami?
OpenClaw Miami was a packed public workshop: bring a machine, leave with a working agent. C-LAW-D was a small legal and capital room: see the agent, then talk about jurisdiction, liability, and who is allowed to run it. Same stack. Different job.
What does sovereign AI mean for a company?
In this context it means control. Where the model runs, where the data lives, who can access the agent, and which jurisdiction you are in when something goes wrong. It is not a slogan. It is an operating choice with legal consequences.
Can a company run a smaller version of this internally?
Yes. Put operators, counsel, and whoever owns risk in one room. Show a live agent on real work. Then write the rules. Purple Horizons runs that as a company workshop, not as a conference.
Who led this from Purple Horizons?
Co-founders Gianni D'Alerta and Ralph Quintero. They ran the OpenClaw station, the live demo, and moderated the governance panel. Same operator-in-the-room model as the rest of the firm's training work.
Gianni D'Alerta
Co-founder of Purple Horizons and AI Innovation Fellow at FIU's Center for Leadership. Transforms complex tech into business breakthroughs.
