Executable institutional memory
A company brain that does more than search. It drafts workflows, identifies contradictions, watches decisions age, and suggests the next operating move.
+40%+80%
Ventures
Company ideas for the next strange decade.
ZOAK is exploring companies that compress knowledge, labor, research, compliance, physical operations, and frontier infrastructure into AI-native operating systems.
Thesis map
Out-there, but grounded in systems we know how to build.
These are not polished product pages. They are directions we think are worth building toward: strange enough to matter, practical enough to prototype, and measurable enough to operate.
Outcome-based service companies
AI-native service firms where agents handle research, production, QA, analytics, and reporting while humans own judgment, taste, and relationships.
+60%+140%
Interfaces that assemble themselves
Software that changes shape around a user's goal, role, urgency, and context instead of forcing every workflow through a static dashboard.
+30%+70%
Care plans that behave like living systems
Patient-specific agents that integrate diagnostics, behavior, clinical context, and follow-up loops for high-friction care journeys.
+25%+55%
Precision intervention for crops
AI systems that combine field sensing, robotics, and biological signals to reduce broad chemical use and raise intervention accuracy.
+35%+75%
Cheap defense against cheap autonomy
Software-first detection, coordination, and response systems for environments where low-cost autonomous threats scale faster than traditional defenses.
+50%+120%
Inference infrastructure for long-running work
Tools and orchestration layers optimized for agent loops, memory, context routing, verification, and domain-specific action.
+45%+95%
Compressed manufacturing coordination
Systems that turn supplier discovery, qualification, compliance, risk, and production planning into a faster negotiation and execution loop.
+30%+65%
Operating software for orbital industry
Planning, monitoring, and coordination tools for manufacturing, logistics, inspection, and maintenance in space-constrained environments.
+20%+50%
Tiny teams replacing bloated incumbents
Focused software companies that use AI to deliver the useful 20% of incumbent platforms with dramatically cleaner workflows.
+40%+90%
Public-sector workflow translators
Agents that convert messy applications, evidence, rules, and service requests into compliant workflows humans can actually audit.
+35%+80%
Guidance systems for hands-on jobs
Real-time copilots for field service, labs, manufacturing, and maintenance where workers need context without stopping the work.
+25%+60%
Thesis velocity
How we prioritize what to build.
Market pull
System complexity
Builder filter
We are looking for ideas where intelligence changes the operating model.
- The old workflow is slow because humans are coordinating around bad interfaces.
- The data exists, but no one has turned it into a live operating loop.
- The market is too weird, regulated, physical, or niche for generic AI tooling.
- The winning company will look like software, services, and infrastructure fused together.
Build with us
Interested in building around one of these theses?
We're looking for operators, builders, and domain experts who want to explore frontier company ideas together. Tell us which thesis resonates.