--- THE FUTURE PROTOCOL ---
We formalize the interface between human intention and computational systems. The protocol is the product.
THESIS
The bottleneck was never the tools. It was how we talked to them.
Every team that builds complex systems — games, simulations, creative engines — hits the same wall. The AI can generate. The pipeline can execute. But the gap between what you mean and what the system does remains unformalized. Decisions are scattered across documents, chat logs, and memory. Context is lost between sessions. Intent is ambiguous.
VisProto exists to close that gap. We build protocols — structured, machine-readable, human-auditable — that make intention explicit. When the protocol is right, everything downstream becomes tractable.
APPROACH
Four principles that govern everything we build.
Define what exists before building systems to manipulate it. A game has actors, scenes, dialogues, items — each with structural relationships. We model these as typed entities with explicit contracts, not as free-form text.
AI should know deeply about its domain and nothing about what lies outside. A dialogue agent understands character voice, narrative context, and linguistic style — but never touches the balance spreadsheet. Boundaries create reliability.
When individual agents each hold a piece of the structure, their collaboration produces results that no single generalist model can match. Understanding emerges from specialization, not from scale alone.
The protocol is not documentation written after the fact. It is the product itself — the formalized pattern of how humans and machines should communicate in a given domain. Ship the protocol, and the tools follow.
PRODUCTS
A specialized AI agent runtime for game development. 16 domain-expert agents — dialogue, narrative, balance, systems, QA, localization, and more — each structurally aware of your project data. Not another chat wrapper. A team that understands your game.
A substrate for coherent world ontologies. Every fictional world — a game, a novel, a film universe — accumulates lore, rules, and relationships that become impossible to hold in one mind. Worldseed provides the structural foundation to keep world knowledge consistent, queryable, and alive.
ECOSYSTEM
Tested in production. Every day.
VisProto is not a research lab. Our first and deepest client is SenseGames — a game studio building complex action RPGs. SenseClaw runs against real production data: 200+ characters, 5,000+ dialogue lines, complete item and skill databases. Every protocol we write has been tested against the chaos of actual game development.
VisProto defines how intention maps to execution. Ontology schemas, agent boundaries, tool contracts.
SenseClaw instantiates protocols into specialized agents. Each agent is a domain expert with real project context.
Production data from game studios feeds back into protocol refinement. The system gets sharper with every project.
ABOUT
We study the boundary between intention and execution. What mattered was never the pipeline — it was the protocol.
Founded in 2024. Based in Shenzhen, with infrastructure in Singapore. VisProto is a one-person methodology company — deliberately small, deliberately deep. Theory follows observation. Observation follows practice.
Define what exists before building systems to manipulate it. Structure precedes intelligence.
Intelligence respects the boundaries you define. Agents that know everything are reliable at nothing.
Systems understood structurally, not merely used. Depth of understanding is the only moat that compounds.
If this resonates.
vi@viskem.com