Intelligence needs discipline.

The Founder

Ejentum is built by Frank Brsrk (Franko Luci). Self-taught AI developer. Founded March 2026. One person engineering the 679 abilities, running the benchmarks, shipping the system.

The Logic API runs against his own agents every day. If a new ability can't hold up under that load, it doesn't ship to yours.

The point has always been singular. Push LLM agents to their limits. Get the best. Block the worst.

Partners

Two partners made the realization of Ejentum possible.

Robert Franks funds the research: tools, Frank's time, the runway to keep experimenting. He contributes strategic proposals and supports growth.

Dennis Kassner runs the production infrastructure powering the Logic API.

Product direction stays with Frank.

The Thesis

From how you handle a dirty glass to how you untie a hard knot: every task demands a specific reasoning posture. Simple or complex, the posture matters more than the content. LLMs have the knowledge. They apply the wrong posture.

That is the failure. Not informational. Structural. Every tool in your stack gives the agent more to work with. Nothing tells it how to think about what it already has. Ejentum's 679 cognitive operations are that missing layer: structured postures across reasoning, anti-deception, code, and memory, engineered for agentic inference.

The central discovery: telling a model what not to do is more powerful than telling it what to do. Suppression outperforms amplification. This wasn't theorized. It was measured across 250+ blind-evaluated tasks on the strongest reasoning model available.

What We Built

That thesis produced 679 cognitive abilities across 4 product layers. Not prompts. Not templates. Engineered operations, each held to a standard before entering the graph:

• 1.Must be a cognitive operation: not domain knowledge
• 2.Must be LLM-executable: no external tools
• 3.Must work across subjects: not tied to one industry

An ability is a compact protocol. Its purpose is to shape how the model reasons, not what it knows.

Each one names the failure pattern your task is vulnerable to, provides the reasoning procedure to follow instead, defines the verification criterion that closes the loop, and carries the suppression signals that block the shortcuts your task invites. Retrieved by semantic match to your task. Absorbed before generation. Shape before output, not cleanup after.

Where This Goes

Four harnesses today: Reasoning, Anti-Deception, Code, Memory. 679 cognitive abilities across them. Research is ongoing. Development persists.

Every agent calls Ejentum for judgment the way every web app calls a database for state. Not an optional enhancement. A harness for the power already inside.

How We Work

Every claim on this website traces to a specific benchmark with named methodology. When results disappoint, we publish the regression alongside the improvement.

When our most complex mode underperformed our simplest one, we deprecated it and published why.

When 62% of tasks received the wrong reasoning domain and improvements persisted anyway, we published that too. The finding was more valuable than the embarrassment.

Austere methodology. Adversarial audits. Scientific rigor as a product requirement, not a marketing claim.