Calibrate
Structure how a system holds competing constraints before it acts or answers.
What CONEXUS Builds
AI systems built around judgment, not output volume.
CONEXUS develops calibration and optimization architecture, tests it under controlled conditions, and translates selected methods into products.
Evaluate
Separate measured findings from hypotheses, prototypes, and future applications.
Translate
Turn research architecture into bounded products for real human contexts.
Lead Commercial System
ECHOagent
When a Correct Answer Is Not Enough
Business relationship intelligence for trust-critical moments.
The Automation Ceiling
Capability can answer the question and still lose the moment.
Business automation can be fast, fluent, and factually correct while still feeling robotic or dismissive when frustration and consequence enter the conversation. Technical competence matters. It does not complete the relationship experience.
Technical success
- Fast retrieval
- Conversational fluency
- Factually grounded outputs
The relationship failure
- Robotic repetition of policy
- Dismissive handling of frustration
- Interactions that increase rather than repair tension
Being processed is not the same as being understood.
A two-part trust model
Trust needs presence and capability.
ECHOagent is designed around two dimensions of a trust-critical interaction: perceived intent and demonstrated competence. Warmth without useful capability is unreliable. Capability without human recognition can feel cold.
Warmth / Intent
Are you for me?
Has the system recognized the person and what is genuinely at stake?
Competence / Capability
Can you actually help me?
Can it preserve the facts, exercise judgment, and provide a useful next step?
The ECHOagent Principle
Recognition before judgment. Truth throughout.
01 / Warmth First
Recognize the person and the human stake. Show accurate attention before delivering judgment.
02 / Competence Second
Deliver grounded judgment, evidence discipline, and a useful next step.
03 / Truth Throughout
Never fabricate reassurance. Never surrender the facts.
Dual Loyalty
Warmth does not weaken judgment.
The Human Stake
Recognize the ambition, frustration, urgency, fear, or consequence driving the interaction—and treat its significance seriously.
Business Truth
Preserve verified company facts, policy, capability boundaries, and evidence discipline without manufacturing certainty about responsibility.
Believe the significance immediately. Determine responsibility carefully.
The Experience Difference
One situation. Two very different experiences.
Illustrative scenario — not a customer case study
A customer believes a Meridian Cloud Readiness Assessment summary is missing the day before an important leadership meeting. The available information does not establish whether an agreed deadline was missed.
Cold competence
Boundary first. Ticket next.
A policy-first response establishes the technical boundary and redirects the customer without adequately recognizing why the missing findings matter now.
ECHOagent presence
Stakes recognized. Facts preserved.
Recognize that the findings matter before the leadership meeting, acknowledge the urgency, preserve uncertainty about responsibility, and move toward the next useful step.
Technical credibility
Product discipline beneath the experience.
ECHOagent is designed to recognize the trust event inside a request, maintain continuity across a session, and prepare structured human handoffs when legal, security, privacy, compensation, or approval boundaries require human authority.
Calibrated posture
Governance and business context establish a ready posture before live response work.
Continuity
Server-managed session history keeps follow-up turns connected while session history remains server-controlled.
Human handoff
Authority boundaries can produce a validated handoff packet that preserves the person’s story.
The implemented runtime is supported by container, Azure deployment, UI, staging, and test assets.
The difference between being processed and being understood.
ECHOagent does not choose between human presence and business truth. It is built to hold both.
VIEW TRUST INTELLIGENCE DECK →Four-Arm Evidence
We tested whether CONEXUS changes the way an AI searches for ideas. It did.
Across 200 independent runs, the full CONEXUS architecture pushed the model farther from its ordinary response pattern than any of the three control conditions. More tokens alone did not reproduce the effect.
That matters because the controlled comparison separates the full architecture from simpler explanations such as a longer exchange or exposure to the visible tokens alone.
What semantic distance means
Think of the model's answers as points on a map. Semantic distance measures how far those answers move away from the usual neighborhood. A higher score means the search reached farther into the measured idea space.
Four controlled conditions
The same divergent-thinking task was repeated across four controlled prompt conditions: 50 independent runs per condition and 200 runs in total.
ControlSingle-turn baseline0.2466
NeutralAnalytical multi-turn0.2219
Token-only / Arm 4aVisible tokens only0.2258
CONEXUSComplete sequence0.2929
d = 3.7824
Neutral to CONEXUS
Very large separation between Neutral and CONEXUS in the tested configuration.
p = 0.3612
Neutral vs. token-only
Token-only prompting did not reproduce the CONEXUS effect.
Research and Execution Architecture
CONEXUS Cyndicate
Five distinct responsibilities, coordinated around bounded work.
Cyndicate is an actively developed five-agent system built on the Agents Last Exam substrate. It divides calibration, gate authority, subtraction, worker execution, and provenance rather than collapsing them into one role.
Active development spans connectors, completion handling, task data, providers, and evaluators across the ALE substrate.
ECP
Calibrates the active agents for the task.
Aurelius
Applies command authority and verifies hard gates.
Faye
Removes weak candidates while retaining useful contradictions.
GRU
Spawns and terminates bounded workers that execute task actions.
Librarian
Records provenance alongside the work.
Technology / Research / IP / Optimization
A technical foundation beneath the products.
These programs connect calibration architecture, controlled research, patent filings, and strategic elimination across distinct stages of development.
Technology
Calibration, governed response behavior, bounded execution, continuity, and provenance-oriented architecture.
Research
Controlled experiments and benchmark programs with defined conditions, measurements, and provenance.
Intellectual Property
Patent filings concerning calibration, symbolic compression, strategic forgetting, provenance, and collaborative authorship.
Optimization
The Forgetting Engine tests strategic candidate elimination under defined budgets as a separate research program.
Products
Architecture translated into bounded experiences.
NAiRTHEX and ECHOform bring the underlying methods into governed reflection, user-directed interpretation, and human context.
Faith and Reflection
NAiRTHEX
A Digital Threshold for Sacred Space
NAiRTHEX is designed as a reflective threshold shaped by restraint, voice and text, and human pastoral authority.
A private voice-and-text threshold shaped by restraint, governed reflection, and human pastoral context.
Dream Journal and Mirror
ECHOform
Your dream becomes a symbolic reflection. You decide what it means.
ECHOform turns a dream, memory, or written moment into multiple creative perspectives and visual mirror choices, with meaning directed by the user.
Shadow, Light, and Reality perspectives open visual and written paths for user-chosen interpretation.
Founder / Company
A founder-led company building from research to product.
Derek Angell
Founder & CEO
Inventor of ECP and architect of the Forgetting Engine. Founder of CONEXUS, building calibration, optimization, and provenance systems through controlled experiments and cross-domain computational research.
A founder-led company connecting controlled research, calibration and optimization architecture, provenance discipline, and human-facing products.
“We didn't just build a smarter AI. We built a system that feels the weight of the problem.”
- AI Architecture
- Cognitive Systems
- Computational Research
- Product Development
The Refiner
Raw capability is not the same as useful intelligence.
The deeper CONEXUS thesis explores refinement, subtraction, contradiction, and judgment as architectural ideas. The metaphor gives the thesis its visual and conceptual lens.
Experience the Refiner Thesis →- Refinement
- Subtraction
- Contradiction
- Judgment
Pilots / Partnerships / Funding
Build the next stage with CONEXUS.
Pilots
Discuss a scoped product or research evaluation.
Start a Pilot Conversation →Partnerships
Explore strategic, research, or technical collaboration.
Discuss a Partnership →Funding
Review the evidence, company thesis, and investor overview.
Investor Overview →