Agentic Engineering

The discipline of building AI systems that sense, decide, and act — with purpose, evidence, and human trust.

6Patterns
5Principles
4Layers
8FAQs

Core Patterns

Autonomous Agents

Agents that sense, decide, and act — with human approval gates at every critical juncture.

Orchestration

Multi-agent coordination through typed contracts, not brittle prompts. Each agent owns one job.

Memory & Context

Long-term memory via .yawn files. Short-term context via state graphs. Both are auditable.

Safety Kernel

Policy Decision Points gate every action. deny > escalate > allow. No silent failures.

Evidence Chains

Every agent action produces evidence. Experiments track hypotheses. Nothing is assumed.

Learning Loops

SENSE → MAP → PREDICT → EXPLORE → DECIDE → ACT → PROVE → LEARN. Then repeat.

Architecture Stack

Perception
Trigger DetectionEntity ExtractionContext Assembly
Reasoning
State GraphsPolicy EvaluationApproach Selection
Action
Tool ExecutionCode GenerationAPI Orchestration
Learning
Evidence CollectionPattern DetectionSkill Graduation

Principles

01

Agency with Guardrails

Agents should be autonomous, not uncontrolled.

Human-in-the-loop for high-risk actions. Fully autonomous for low-risk. The kernel decides which.

02

Holonic Architecture

Every part is also a whole.

Agents compose into teams. Teams compose into organizations. Rules inherit down the tree.

03

Evidence over Opinions

Don't guess. Experiment.

Define hypothesis, set success criteria, run the experiment, collect evidence, then decide.

04

Typed Contracts

If it isn't typed, it doesn't exist.

Agent inputs, outputs, and capabilities are TypeScript interfaces. No magic strings.

05

Coherence over Correctness

The system must make sense as a whole.

Individual agents can be wrong. The system catches it through coherence checks and feedback loops.

Frequently Asked Questions

Build Agents That Learn

Start with a yawn. Define the job. Let agents sense, decide, act, and prove — while you stay in control.