MANIFESTO
Organizations have more data than ever. Yet they still lack systems to think better.
By Miguel Ángel Sánchez Ciria · May 2026, revised September 2026 · ~5 min read
1. The paradox of abundance
There is more data available than at any moment in the history of humanity. More tools to process it, more infrastructure to store it, more algorithms to find patterns in it. And yet, inside most organizations, the same questions keep coming up: why did we decide this? what were we considering when we decided? who signed off? what alternatives were discarded and why?
The answers exist. They are in last quarter's Slack threads, in PDF decks no one ever reopened, in wikis that were maintained for three months, in the memory of people who rotated to another company. The accumulated cognition of the organization lives scattered, with no common substrate to preserve it.
That is the paradox of abundance: having more data than ever and, at the same time, having no system to think better.
2. Why current approaches don't solve this
The industry has answered with several categories of tool. Each one solves something. None of them solves this.
General-purpose chatbots and assistants (ChatGPT, Claude, Gemini) offer powerful conversational intelligence, but they are not built to sustain long-running work. In long conversations the context degrades, the model fills gaps with invented details, and it proposes again options that had already been ruled out. Memory, when it exists, is optional, individual, and lives at the model provider.
Code copilots (Copilot, Cursor) autocomplete in the editor. They are useful for generating code but do not govern the decision cycle around that code: what was decided, who signed, why. A three-year audit is impossible.
RAG systems solve "search your documents and generate an answer based on them". Useful, but passive: it depends on someone having written the right documents in the first place. Cognition that doesn't get written down doesn't get retrieved.
Agent frameworks (LangGraph, CrewAI, AutoGen) optimize multi-agent autonomy. The question they don't answer is responsibility: who signs off on each action, how the decision can be reconstructed, how a wrong state can be reverted.
Wikis and KMS tools (Confluence, Notion, SharePoint) document passively. The organization has to dedicate hours to maintaining something that, normally, few people ever consult again.
Each of these categories is valuable in its domain. None of them preserves the accumulated cognition of an organization with traceability and governance.
3. The third path
The two options the market assumes are: ungoverned autonomous AI, or unamplified humans. The third path is governed, persistent, accountable intelligence.
This means four concrete properties:
— Capture context without asking anyone to write extra documentation. Cognition is preserved inside the natural cycle of AI-assisted work.
— Structure thinking instead of leaving it in threads, emails, and people's memory. The way something was decided becomes explicit.
— Govern decisions assisted by AI through a formal cycle: the model's proposal, the human review, the signed accountability. Natively auditable.
— Accumulate intelligence over time, surviving human turnover and the replacement of the underlying language model.
4. INTENTIA+ — what it is, in honest terms
INTENTIA+ is the infrastructure that implements that third path: the governed record of what an organization decides with AI. It is made of five pieces:
— The cognitive repository (CPR, Cognitive Persistent Repository) is the persistent, governed memory of each workspace: facts, decisions, documents and their versions, with who added what and when. The useful intelligence of INTENTIA+ is not in the language model, which gets replaced. It is in what this repository has accumulated about your organization.
— AICA (AI Cognitive Agent) is the cognitive agent. It greets you on your home page and tells you what is waiting for you, what is about to expire and where you left off. It proposes and prepares, but it never decides or acts on its own.
— Multi-source: you connect the folders where the work already lives, and their content reaches the assistant, which always says which documents it read and why.
— Multi-purpose: the same product serves a law firm, a consultancy or a software team. Each workspace has its own purpose, documents and rules.
— Context rules turn the organization's rules into product objects, with a scope, a version and a declared consequence. They reach the assistant with every request, and the record shows which ones applied.
All five are governed by AI-REPO+, the governance framework, and its HHAC cycle (Human-Human-AI Cycle): the AI proposes, a person approves, rejects or amends, and what gets approved is recorded with its reasoning. Nothing enters the organization's memory without that human step.
Built this way, the organization works with its own OpenAI, Anthropic or Google account, so the AI, the bill and the data remain its own. Every answer says what it read, which rules governed it and what it cost. Decisions nobody resolves in time expire instead of waiting forever. And cognition persists when people rotate or the model changes.
5. What INTENTIA+ does NOT promise
Honesty is part of the architecture. INTENTIA+ does not replace the human team: it amplifies. It does not promise "total control" or "safe AI": it reduces the risk of bringing AI in with explicit governance, complies with EU AI regulation (the AI Act) and, as the work happens, produces the evidence that whoever has to show how AI was used will need.
It is not an academic promise: INTENTIA+ works today, is used by invited organizations and runs on servers in Europe. What it does, and what comes next, is on the product page.
And it does not solve the cultural problem on its own. If the organization does not want to preserve cognition, no system will preserve it for it.
6. An invitation
INTENTIA+ works today with invited organizations. If the problem described in this text is recognized inside your organization — and you believe the third path is worth exploring — write to us.
— Miguel Ángel Sánchez Ciria
May 2026