Hanzo runs multiple AI agents across its ventures right now — not as a roadmap item, not as a pitch deck slide, but as operational reality. A single founder runs six ventures with 14 employees because AI handles the operational grind that would otherwise require a team five times that size. This is not a story about what AI could do someday. This is what it does for us today, across every business in the portfolio.
We did not set out to be an AI company. We set out to build businesses that could not scale without it. Every AI system we deploy exists because a specific bottleneck demanded automation — not because we wanted to slap "AI-powered" on a landing page. The result is an infrastructure layer that runs subscriber engagement, billing, content generation, employee management, and media distribution across all of our ventures simultaneously.
VelvetChat: AI as the chatter's weapon
VelvetChat is where our AI stack is deepest. The system suggests real-time responses to subscriber messages by analyzing conversation history, spending behavior, and subscriber preferences. It scores subscribers into spending tiers and flags upsell opportunities before a human chatter even reads the message. It generates content using LoRA-trained image models fine-tuned on each creator's likeness. It clones model voices through ElevenLabs for AI-generated voice notes that sound indistinguishable from the real person. The human chatters are still essential — they make the judgment calls, build the relationships, handle the edge cases — but AI gives them superhuman context on every interaction. Real chatters, supercharged by AI.
The operational backbone
The customer-facing AI gets the attention, but the bulk of our agent infrastructure is operational. Stripe ACH billing runs on full automation — tier calculation based on creator earnings, invoice generation, 7-day and 3-day and 1-day payment reminders, overdue recovery sequences, and payment celebration notifications. All AI-orchestrated, no human touches it unless something breaks. Our Discord bots run 15+ tools for employee management: clock-in and clock-out tracking, shift scheduling, model credential lookups, channel messaging, training mode management, hours adjustment, and billing lookups. An employee can ask the bot who is on shift, check their hours for the pay period, or look up a model's custom pricing — all through natural language in Discord.
On the content side, we train LoRA models on Replicate for AI image generation and use ElevenLabs Eleven v3 for voice cloning with tuned parameters for each creator. Medtrics uses AI to help physicians surface relevant clinical discussions and research. The lead qualification system scores incoming prospects across multiple signals and routes only pre-qualified opportunities to the team. Every one of these systems was built because a human was spending hours on a task that a well-scoped agent could handle in seconds.
AI is the operating system, not a feature
The key insight we have learned building all of this: AI is not a product feature. It is the operating system of the holding company itself. Without it, running six ventures with 14 people would be physically impossible. With it, the math works — because AI handles billing, content generation, subscriber intelligence, employee scheduling, lead scoring, and media distribution while the humans handle strategy, creative direction, and relationship building. The companies that win with AI will not be the ones with the fanciest models. They will be the ones that deployed it into every operational seam of their business until the humans only touch what humans are actually good at. That is what we are building at Hanzo — one agent, one workflow, one venture at a time.
