<aside> 📘

Document status: Master product vision and top-level architecture for ISAC. GCF 2.0 is the frozen cognitive-framework baseline. This page defines the product intent, system boundaries, architecture principles and links to the detailed subsystem specifications.

</aside>

<aside> 🟢

Implementation status: 25 September 2026: Stage 35 is complete and Foundation Stability Audit Phase 3 is in progress before Stage 36. ISAC remains in active prototype development. This documentation is an engineering source of truth and does not imply commercial release readiness.

</aside>

Executive Summary

ISAC is a persistent AI system built on GCF 2.0. It is designed so that models remain replaceable cognitive engines rather than becoming the identity of the product. Identity, memory, tools, permissions, user relationships, resource management and system behaviour remain part of the wider architecture.

GCF provides the governed cognitive foundation for planning, prediction, resource-aware cognition, recursion, background cognition and learning. ISAC provides the persistent user-facing system that coordinates those capabilities across local and optional external models.

The engineering objective is to maximise useful capability per unit of compute while maintaining explicit security, permission and resource boundaries. Local-first operation is the baseline; cloud and BYOK providers are optional extensions.

<aside> 🛡️

Security and governance position: GCF is designed to address the growing security concerns that arise as AI systems gain access to persistent memory, tools, providers and external actions. Authority, permissions, credentials and auditability are kept outside any single model. These are architectural controls that remain subject to testing and independent validation, not a claim of absolute security.

</aside>

Subsystem Architecture Index

• Identity, Personality & Relationship

• Memory & Security

• Models & Provider Runtime

• Environment & Smart Home

• Skills & Learning

• Goals & Cognitive Initiative

• Multi-Device & Guardian Node

• Offline Knowledge Vault

• Voice & Accessibility

• First-Run Experience & UX

• Benchmarks & Evaluation

• Roadmap & Development Stages

<aside> 🌐

Mission: Build a persistent personal intelligence that becomes more useful through understanding, memory and experience - not simply by consuming more compute.

</aside>

<aside> 🧠

Persistent-ISAC principle: Hardware and model providers may change without creating a different ISAC. Cognitive capacity can scale up or down while ISAC's identity, relationship history, memory and experience remain continuous.

</aside>

Product Rationale

ISAC is not intended to be another chat interface or a conventional model wrapper. The product is being developed as a persistent personal AI system that can maintain continuity, practical capability and a stable relationship with the user over time.

Early development showed that improving the underlying model alone does not solve the wider system problems of planning, memory, permissions, tool access, resource management, continuity and governed learning. That led to GCF: a cognitive framework that provides those system-level capabilities around replaceable models.

ISAC is the first AI system being built on that framework.

User Experience Goals

• Be something a person can turn to for help, support and practical assistance.
• Maintain continuity instead of resetting every session.
• Learn how the person communicates, works and prefers to be helped.
• Offer useful proactive input when appropriate rather than remaining entirely passive.
• Preserve one recognisable ISAC identity across hardware and model changes.

Engineering Goals

• Prove that useful AI capability can remain small and resource-conscious.
• Use architecture, memory, skills, tools and local knowledge to reduce dependence on model scale.
• Remain genuinely useful offline.
• Scale with GCF instead of assuming unlimited RAM, VRAM or cloud compute.
• Measure intelligence against resource cost rather than parameter count alone.

<aside> ⚡

Core thesis: System capability should come from the combination of GCF, memory, skills, local knowledge, tools and experience - not from permanently increasing model size.

</aside>

<aside> 🎯

ISAC is an adaptive AI system built on GCF 2.0 - not a single model. Its purpose is to stay small, efficient and personal while using architecture, tools, memory and governed cognition to perform above the expectations of its resident model size.

</aside>