# Evercursion (GaiaLogic, Unipessoal Lda.)

> Open-ended intelligence. Beyond the answer machine. Intelligence as an open-ended process.

Independent AI research in the EU. We build the reasoning and governance architecture around foundation models — turning computation into open-ended intelligence.

- **Canonical URL:** https://evercursion.ai/
- **Markdown Mirror:** https://evercursion.ai/index.md
- **LLM Index:** https://evercursion.ai/llms.txt
- **Complete Single-File Corpus:** https://evercursion.ai/llms-full.txt
- **Legal Entity:** GaiaLogic, Unipessoal Lda. (Evercursion is a brand of GaiaLogic, Unipessoal Lda.)
- **Headquarters:** Travessa dos Pesos, n.º 1, Espiche, 8600-109 Luz, Portugal
- **Contact:** info@evercursion.ai

## What Open-Ended Intelligence Means

Open-ended intelligence is not a larger model. It is the architecture around a model that lets a system question its own assumptions, form new categories, and revise what it knows.
Instead of treating every prompt as a stateless guess, the system reflects on where its reasoning stalls and carries verified context forward — drawing deeper capability from the models you already run.

## Intelligence is a process, not an answer.

Scaling parameters and chaining agents made models fluent, but fluency is not understanding. We build the organizational architecture that turns computation into accountable, open-ended reasoning.

### Beyond the training frame

A single forward pass only compresses what was already present in training. To solve problems that cross the boundary of its training frame, a model needs an external architecture that can build new distinctions at runtime and hold verified state across long horizons.

### Structured reflection, not consensus

Standard multi-agent pipelines average outputs into consensus and smooth over uncertainty. We organize reasoning across distinct perspectives, treat contradictions between them as signals that a category is missing, and store conclusions together with the context in which they held true.

### Strict separation of reasoning and execution

In production systems, a language model should never decide its own permissions. In our architecture, models propose reasoning steps and actions with an inspectable trace, while deterministic rules outside the model govern what is permitted to execute — from compact local models to frontier systems.

## Published Research Notes

- **Note #06 — [Beyond the Scaling Ceiling: Why Open-Ended Intelligence Requires an Organizational Architecture](https://evercursion.ai/research/note-06-open-ended-intelligence)** ([.md](https://evercursion.ai/research/note-06-open-ended-intelligence.md)): Current AI development rests on the bet that scaling parameters, training data, and test-time compute inside statistical models will eventually produce general intelligence. In our foundational paper, Co-Recursive Semantic Self-Extension: A Mathematical Foundation for Open-Ended Intelligence, we prove mathematically why this path hits a structural ceiling. Before any system can learn, it has to decide which signals matter and which are noise relative to an active purpose. Because any fixed purpose — or any recombination of a fixed catalog of values — admits only a bounded set of distinctions, learning inside that frame inevitably plateaus. We show that no isolated machine with a fixed state space and fixed transition rules can generate an open-ended history of new concepts on its own. General intelligence is a property not of an individual model, but of an organizational architecture that turns stalled reasoning traces into new categories, tests competing interpretations across parallel contexts, and preserves the context behind every stored insight. · External Paper: https://arxiv.org/html/2610.31097v1
- **Note #10 — [Co-Recursive Intelligence Under Deterministic Authority: Zero-UI Protocols, Pre-Gate Routing, and Ring-0 Security](https://evercursion.ai/research/note-13-co-recursive-intelligence-under-deterministic-authority)** ([.md](https://evercursion.ai/research/note-13-co-recursive-intelligence-under-deterministic-authority.md)): Contemporary AI systems oscillate between stateless chat silos that forget structured thought and unbounded agent wrappers that let probabilistic models police their own tool use. We describe the first integrated Evercursion prototype that unifies protocol-first, zero-interface correspondence, calibrated pre-gate routing, multi-perspective reflection, and a versioned wisdom graph beneath a deterministic Ring-0 reference monitor — so deep reflection and mathematical trust share the same separation of intelligence and authority.
- **Note #08 — [Small Models Think Better When They Reflect Polycontexturally](https://evercursion.ai/research/note-08-polycontextural-routing)** ([.md](https://evercursion.ai/research/note-08-polycontextural-routing.md)): Scaling test-time compute by sampling longer chains of thought or majority votes inside a single perspective often hits diminishing or negative returns, as extended chains drift or repeat initial assumptions with inflated confidence. Drawing on Gotthard Günther's polycontextural logic and our mathematical theory of co-recursive self-extension, we wrap compact language models in a two-layer reflection architecture: an engine that decomposes problems across objective, subjective, and conceptual standpoints, and a deterministic second-order router that uses disagreements across those standpoints to decide whether to converge, expand, refine, or reset. Across five benchmark suites — GPQA Diamond, ExploreToM, Hi-ToM, BIG-Bench Hard, and MMLU-Pro — structured reflection and routing beat direct answering in twelve of sixteen main conditions, lifting sequential state tracking up to fivefold, outperforming direct calls to flagship frontier models on multi-agent belief tracking at a fraction of the cost, and showing reflective restraint in an open world with no task or reward at all.
- **Note #07 — [Separating Intelligence and Authority in Agentic Systems](https://evercursion.ai/research/note-07-compositional-harness-security)** ([.md](https://evercursion.ai/research/note-07-compositional-harness-security.md)): When an AI coding or enterprise agent is connected to a shell, a filesystem, and a network, the tools it uses to solve problems are the exact same tools an attacker uses to compromise a host. Once untrusted instructions enter the context window through an issue description, a document, or a dependency file, legitimate utilities like bash, curl, git, and tar become dual-use payloads. In this research project, we apply our organizational theory of intelligence to agentic security. We start from a simple architectural principle — the separation of intelligence and authority: a language model may propose actions, but it must never define or judge its own execution permissions. Across 252 benchmark cases spanning single-step attacks, multi-stage kill chains, and everyday software engineering tasks, deterministic permission and dataflow boundaries block 100% of attacks with 0% false positives in microseconds, whereas asking a local language model to judge safety blocks 98.3% of legitimate engineering work.
- **Note #09 — [When Should a Machine Reflect? Criterion Drift, Prosecutor Bias, and Meta-Teleological Stopping in Moral Judgment](https://evercursion.ai/research/note-12-when-should-a-machine-reflect)** ([.md](https://evercursion.ai/research/note-12-when-should-a-machine-reflect.md)): Spending more test-time compute on step-by-step chains of thought or multi-perspective reflection is widely expected to improve language model judgment. Yet across 1,500 balanced commonsense moral scenarios from the ETHICS benchmark — spanning simple one-line actions, adversarially filtered moral traps, and multi-paragraph interpersonal conflicts — aggregate accuracy on complex conflicts barely moves (61.2% to 64.8%) while more than a quarter of individual verdicts flip in opposite directions. Using signal detection theory to separate moral discriminability from the decision threshold, we show that unconditioned deliberation shifts the threshold far more than underlying understanding, trace two concrete mechanisms in the reasoning trees, and demonstrate how a three-contexture pre-gate before reflection eliminates threshold drift at a fraction of the compute cost.
- **Note #05 — [Intelligence First: Knowledge, Learning, and Reflection](https://evercursion.ai/research/note-05-intelligence-first)** ([.md](https://evercursion.ai/research/note-05-intelligence-first.md)): We start from a plain question: what do knowledge, learning, and intelligence actually mean? The usual answers turn out to be borrowed from machines: knowledge as a stock of facts, learning as applying rules, intelligence as solving problems. We turn that order around, so that intelligence — the ability to think beyond the world as it is — comes first, learning becomes the testing of ideas against the world, and knowledge becomes a set of justified ideas about what the world can become. Between mind and world sits a third pole, information, which always involves a process of reflection. Our CIO, Dr. Simon Hegelich, laid the argument out in two talks, in German, linked at the end for anyone who wants it in full.
- **Note #04 — [The Why Question: Why Language Models Do Not Think, and Where a Different Question Starts](https://evercursion.ai/research/note-04-why-question)** ([.md](https://evercursion.ai/research/note-04-why-question.md)): A language model can read a long detective story with the last chapter cut off and name the murderer on the strength of clues scattered through the text, and still be unable to tell you why it chose him. We argue that this gap sits in the architecture: a model cannot answer a why-question about its own answer, and reasoning models, which repeat the model's pass in chains and trees, move the error around through the bias-variance trade-off without closing the gap. Thinking is used here in a strict, testable sense — as the ability to account for why one's own process reached a conclusion. Our CIO, Dr. Simon Hegelich, laid the argument out in three talks, in German, linked at the end for anyone who wants it in full.
- **Note #03 — [RAG in Three Steps: How Retrieval Works, Where It Breaks, and What Comes Next](https://evercursion.ai/research/note-03-rag-three-steps)** ([.md](https://evercursion.ai/research/note-03-rag-three-steps.md)): Retrieval-augmented generation, RAG for short, hands a language model the texts it never saw in training, found through numerical similarity. We work through how that similarity is computed, why it so often returns the wrong passage, and which three routes lead out of the problem. The failures come from semantic nearness that is not relevance, texts of very different length, and the bias-variance trade-off in embeddings with more than a thousand dimensions. Of the three routes we consider the last one — a knowledge graph with logical relations and explicit provenance — the most promising. Our CIO, Dr. Simon Hegelich, laid the argument out in three talks, in German, linked at the end for anyone who wants it in full.
- **Note #02 — [Structured Reflection vs. a Plain Nano Call: Polyreflexion on GPQA Diamond](https://evercursion.ai/research/note-02-polyreflexion-gpqa)** ([.md](https://evercursion.ai/research/note-02-polyreflexion-gpqa.md)): We tested whether wrapping a small model in a structured reflection system beats asking it directly. The model is gpt-5.4-nano in every condition, the exam is GPQA Diamond with 198 graduate-level multiple-choice questions, and the challenger is polyreflexion, our reflection engine. Once both sides see the same material at the final step, every polyreflexion condition beats a plain direct call, with the best configuration reaching 0.460 against a direct baseline of 0.328. An earlier version of this comparison said the opposite because it was quietly unfair, and correcting that is part of the finding.
- **Note #01 — [A Small Model and a Stack of Rules: Our Run at STATE-Bench](https://evercursion.ai/research/note-01-state-bench-gaia-nano)** ([.md](https://evercursion.ai/research/note-01-state-bench-gaia-nano.md)): We entered Microsoft's STATE-Bench Agent Learning Track with a single small model, gpt-5.4-nano, and a frozen rulebook of plain if/then rules, mined deterministically in code from the training conversations the benchmark allows. Two submissions — one whose design was informed by inspecting earlier test transcripts and one strictly blind run that never saw the test set — reach 0.64 and 0.62 pass@1 on v0.8.0, ahead of the verified full-scale entries on that version. What interests us is less the leaderboard rank than the shape of the system: a benchmark built for large agent architectures can be met by something small that you can read with your own eyes.