// Technical research + architecture
Turn emerging AI into systemsyou can build on.
The Zoop Troop helps fast-moving organizations research, design, and validate agentic systems, workflow automation, and advanced software architectures.
Bespoke engagements · Primarily remote · Available to travel
- 01You see an opportunity for AI agents, but the right system boundaries are not clear.
- 02A promising AI demo needs a dependable path toward real use.
- 03Models, frameworks, or protocols are changing faster than your decision cycle.
- 04A complex workflow needs the right balance of model judgment and deterministic control.
// How we can help
Senior judgment for consequential technical decisions.
01
Design the responsibilities, orchestration, tools, memory, permissions, human oversight, and deterministic boundaries behind useful AI agents.
- Agent and workflow design
- Tool and memory boundaries
- Human intervention
- Reference implementations
Explore agentic systems architecture →02
Investigate models, frameworks, protocols, and architectural approaches before your organization makes a consequential investment.
- Model evaluation
- Technology investigation
- Feasibility research
- Prototype validation
Explore emerging technology research →03
Review an existing direction, surface consequential risks, document decisions, and give internal teams a credible path forward.
- Architecture reviews
- Decision and risk analysis
- Technical roadmaps
- Fractional guidance
Explore architecture validation and guidance →// Our approach
Start with the problem, not the agent.
Sometimes the right answer is an agent. Sometimes it is a constrained workflow, conventional software, or no new system at all. We recommend what the evidence supports.
- 01
Understand the stakes
Start with the decision, workflow, constraints, and cost of being wrong—not a predetermined technology.
- 02
Investigate and test
Research the relevant approaches and use focused experiments to replace assumptions with evidence.
- 03
Design the boundaries
Put autonomy where it creates value and deterministic controls where correctness and accountability matter.
- 04
Create a path forward
Deliver the artifacts your team needs—from decisions and diagrams to prototypes, roadmaps, and guidance.
// Selected experience
Grounded in the work.
Our perspective comes from building and operating products, contributing to collaborative infrastructure, and testing emerging approaches hands-on—not from repeating an industry playbook.
Owned product
BlueClaw Network
A customer-facing AI inference platform that turns distributed compute research into an operating product with routing, metering, account controls, and observability.
Visit BlueClaw Network →Collaborative work
Cloud SPE
Livepeer-funded work in AI and video infrastructure, analytics, data pipelines, model runners, gateways, and open-source systems—contributed through XodeApp alongside Speedybird and other collaborators.
Visit Cloud SPE →Operated initiative
XodeApp + Open Pool
Hands-on experience operating distributed video and AI infrastructure and building systems for worker coordination, performance measurement, and automated settlement.
Visit XodeApp + Open Pool →Research practice
Applied research and experiments
Working explorations of agent orchestration, workflow automation, model behavior, memory isolation, real-time media, local-first analytics, and emerging architectural patterns.
- Years in software architecture
- 24+
- Primary working model
- Remote
- Travel when the work calls for it
- Global
// Start a conversation
Tell us what you are exploring, where uncertainty exists, and what is at stake. We will determine whether our experience and interests align with the problem.