AI is gaining authority faster than the security boundary around it.
GCF FrameWorks is building the governed control layer between AI intelligence and the systems, data and tools it can affect. ISAC is the first persistent AI system built to prove the architecture.
Why now
AI systems are increasingly connected to memory, files, browsers, APIs, code execution, credentials and external services. That turns model failures from answer-quality problems into potential system-level incidents.
What exists today
GCF's core architecture is built as the current baseline. ISAC runs internally on the framework with working governance controls, memory and model infrastructure, and substantial automated test coverage. Internal evidence includes a governance-gate test in which a synthetic destructive request is blocked before execution.
Round at a glance
£250k pre-seed
Working assumption ~£1.42m pre-money. To confirm at term sheet.
35 / 55
Foundation Stability Audit, Phase 3 in progress.
What the round unlocks
Move beyond founder-constrained delivery
Add focused engineering capacity and product hardening.
Test the claims properly
Independent review, adversarial testing and isolated security environments.
External evidence
Controlled evaluators, design-partner feedback, workflow results and willingness-to-pay evidence.
Deeper technical R&D
Support larger test workloads and future capability-per-compute research.
Documents
Three documents. All public, no email gate.
Architecture spec
Full design specification covering the governance boundary, memory, scoring and routing.
Read on Notion ↗Technical evidence
Runtime logs, governance block configuration and prior test outcomes.
Open technical evidence ↗// all three are public; no email gate, no Stripe checkout, no NDAs
Long-term opportunity
The initial commercial direction is private-sector persistent AI. Consumer systems come later. Robotics is a long-term extension where governed authority and human override become even more important because AI actions can become physical. Separate future R&D will explore efficient models, but the core principle remains unchanged: capability can increase without authority increasing automatically.