Memory by Contribution
How institutional memory gets built up by contribution rather than by ingestion, and why that distinction governs whether your AI compounds or commoditises.
Author Storefront · Ibex AI
Foundation-AI
Foundation-AI
The Series
Strategic writing on context, governance, and sovereign intelligence.
Whitepapers and essays on how organisations turn context into a moat, and how to keep that moat theirs as frontier models converge. A running series, written from inside the work at Ibex.
1 featured whitepaper
7 companion deep dives
Series in progress
The series
Featured whitepaper
The foundation paper: context, a model of your own, and governance as one architecture. Start here.
Strategic Whitepaper · Foundation
Sovereign Intelligence
Why the enterprise AI advantage is context, and what it takes to trust it. Once frontier models converge, the moat is the institutional context an organisation puts around them, kept sovereign by a model of its own and made trustworthy by governance. The context and governance layers of Foundation-AI as one architecture.
Read paper
Going deeper
Companion deep dives
Close-ups that take one idea from the Scout and agent strategy apart, end to end.
Beyond the Agent: Why Scouts Scale Agentic AI
Why one-off AI agents stall at scale, and why persistent Scouts with memory, shared learning, receipts, and governed autonomy become the real unit of compounding agentic AI.
Read deep dive
Beyond the Lone Agent: The Substrate Behind the Assistant
A fair, named comparison of Hermes, OpenClaw, and Foundation's Triton. Two excellent open-source personal agents, and an enterprise-grade substrate the model lives inside, an AI assistant and digital twin you can give to thousands of employees.
Read deep dive
Beyond the Answer Bot: Why Agentic AI Needs a Living Library for Memory
An answer bot fetches what looks similar; a living library decides what is true, current, permitted, and worth acting on. Why agentic AI needs memory that curates, learns, and proves what it holds.
Read deep dive
Beyond the Big Model: Why Agentic AI Needs Metis
Metis is a private enterprise Vertical Language Model for finance, automotive, supply chain, ESG, telco, media, deeptech, agritech, spacetech, and other verticals where domain signals need to become enterprise judgment capital.
Read deep dive
The Capital Scout: Diligence That Proves Itself
A diligence team of specialist Scouts that each sign and source their work, trace every claim back to its origin, and refuse to conclude on anything they cannot prove.
Read deep dive
The Lighthouse Scout: Innovation Intelligence That Never Sleeps
A two-sided innovation match that helps corporations find the right startups, and helps startups find the corporates that genuinely need them, with the evidence, while the window is still open.
Read deep dive
The Forge Scout: Media Intelligence That Cites Its Sources
A media-intelligence Scout that turns recorded video, audio, documents, images, and 3D models into a sourced, readable brief, ties every line to a timecode in the source, and says so plainly when it cannot pull real content.
Read deep dive
Coming next
In the pipeline
Drafts in motion: titles and framings may shift before publication.
PACT: the Trust Plane
The trust contract between models, contexts, and the humans accountable for what they produce. How a Programmable Accountability Contract Trail makes governance enforceable at runtime.