cosenlab
Research

Cosen Lab Research

We study what does not yet have an answer.

Systems, architectures, and principles that can change how people and organizations work with intelligent technology. From the question to a working system.

AI governanceCognitive systemsKnowledgePeople and AIEnterprise
  1. 01

    Question

    What remains unresolved?

  2. 02

    Experiment

    What can we test?

  3. 03

    Architecture

    What survives implementation?

01Research agenda

What are we
trying to understand?

Our research starts from the questions that sit beneath products and interfaces: the architectural problems that emerge when intelligence, memory, evidence, and human authority are part of the same system.

  1. 01

    Cognitive infrastructure

    How can digital systems preserve context, memory, and intent over time?

  2. 02

    Governed intelligence

    How can increasingly capable AI remain inspectable, controllable, and accountable?

  3. 03

    Evidence-based knowledge

    How do we distinguish information from knowledge that is supported, current, and traceable?

  4. 04

    Human–AI decision-making

    Where should machine intelligence stop and human authority begin?

  5. 05

    Enterprise intelligence

    How can intelligence be brought into complex organizations without replacing their systems of record?

02Active research programs

Research does not stay on paper.
It becomes systems.

Six interconnected programs, not separate departments. Every Cosen Lab product is where a hypothesis meets constraints, users, and operational reality. Select a program.

Program · systemClabbitQuuipuIRIS Core AIIRIS DiscoverySempliGoIRIS applications

Program 03 · Knowledge · active research

Epistemic integrity

How should a system behave when knowledge is incomplete, uncertain, or contradictory?

We study how to preserve uncertainty and bound answers, without turning an absence of evidence into false confidence.

Tested in IRIS Discovery

Research questions

  1. Q1How does what is unknown stay unknown?
  2. Q2When is an inference admissible?
  3. Q3How should conflicting evidence be represented?
  4. Q4How does confidence remain explainable?
03Open questions

Questions we
keep coming back to.

Research means resisting the temptation to answer too quickly. Some questions remain useful precisely because they expose the limits of what current systems can do safely.

  1. 01Can an intelligent system know when it does not know?
  2. 02What should an AI system be allowed to remember?
  3. 03Can governance become part of computation itself?
  4. 04What happens when intent replaces navigation?
  5. 05How do we preserve human authority as machines become more capable?
  6. 06Can regulation become operational knowledge?
  7. 07Can enterprise intelligence exist without replacing enterprise systems?
04Method

How we investigate.
A cycle, not a line.

We move between conceptual models and working systems: we make assumptions explicit, then build, test, and revise without hiding what remains uncertain.

  1. 01

    Frame

    Define the question and its boundaries.

  2. 02

    Model

    Build conceptual models, ontologies, and architectures.

  3. 03

    Prototype

    Turn hypotheses into working systems.

  4. 04

    Stress-test

    Scenarios, constraints, and failure modes.

  5. 05

    Observe

    What works, what fails, what remains unknown.

  6. 06

    Revise

    Update the model without hiding uncertainty.

Every revision leads back to a sharper question.

05Research materials

Not everything we learn
becomes an article.

Research produces different materials depending on how mature the question is. Notes, papers, and frameworks will be published as they become ready to be read, discussed, and challenged.

  • Research notes

    Short explorations of emerging questions

  • Frameworks

    Structured conceptual models

  • Architecture papers

    Technical models derived from research

  • Experiments

    Working tests of specific hypotheses

  • Data and evaluations

    Where publication permits

  • Specifications

    Formal models ready for implementation

  1. Research note · Cognitive systems

    Memory as infrastructure

    What changes when memory becomes a stable component of digital architecture.

    In preparation12 min

  2. Architecture paper · AI and governance

    Governance as execution architecture

    Policy, authority, and evidence as native properties of intelligent systems.

    In preparation15 min

  3. Research note · Knowledge

    Preserving the unknown

    Why systems should preserve uncertainty instead of manufacturing confidence.

    In preparation11 min

  4. Framework · People and technology

    Intent before navigation

    Software experiences organized around purpose, not a hierarchy of pages.

    In preparation10 min

  5. Architecture paper · Enterprise

    Intelligence above systems of record

    Decision intelligence without replacing enterprise systems.

    In preparation14 min

Some questions must remain open.

Research is not about producing certainty. It is the disciplined work of narrowing what we do not yet understand.

Research partnershipsExperimentsArchitecture