Ideas
Find the consequential problem beneath the visible request and state the hypothesis clearly.
Independent AI Product & Architecture Lab
Cobalt Giraffe is an independent AI product and architecture lab. The lab explores where AI and agentic systems can solve difficult real-world problems, then architects, prototypes and tests the ideas worth pursuing.
Some ideas work. Some change direction. Some fail.
Failure with evidence is a useful result.
Cobalt Giraffe is an environment for examining difficult technology problems, shaping defensible architectures and testing what deserves to exist.
Find the consequential problem beneath the visible request and state the hypothesis clearly.
Make responsibilities, boundaries, data, controls and human authority explicit before autonomy grows.
Replace confident assumptions with prototypes, evaluation and observable evidence.
Record what the evidence supports, what failed and what remains unresolved. Build further when the findings justify it.
Research, architecture and product engineering, with assumptions, decisions, evidence and limitations made explicit.

Identity, authority and autonomy - with controls where consequences occur.
Research in progress · Working architecture hypothesis; not validated
Explore the modelA disciplined route from an ambiguous problem to a decision: continue, iterate or kill.
Northstar tests whether authority checked earlier still holds when an action takes effect. Its synthetic schedules retain the counterexamples, including the failures that limit the claim.
Northstar · LEARNING / RESEARCH POC. NOT PRODUCTION APPROVED.
Inspect the Northstar investigationPublished research and evolving architecture models on enterprise architecture, agentic systems and AI governance. Each piece makes its argument, limits and current status clear.
Research, architecture and experimentation
Cobalt Giraffe examines difficult technology problems without forcing every idea toward a predetermined answer or product.
The lab investigates systems that must work inside real organisations, where governance, security, traceability and human accountability matter as much as model capability.
Founded by Philip Gregory, Enterprise & AI Architect.
About the lab