// epigenetic intelligence v0.8

AI THAT
REMEMBERS
EVOLVES
COMPOUNDS

Helix is a persistent intelligence layer that sits between you and your AI assistant — learning from every session the way biology learns from experience, without rewriting a single line of its own code.

0
Atoms in Pool
0
Live Symbols
0
MCP Tools
Scroll
the problem

Every Session Starts From Scratch

The model has no access to the context that makes you productive: the patterns you've built dozens of times, the naming conventions you've settled on, the architectural decisions you made last week. Every session begins at zero.

Repeated Explanations

You re-describe your stack every session. The AI asks what FastAPI is. Again.

Pattern Drift

The AI proposes solutions that contradict conventions established three weeks ago.

Context Collapse

Long sessions lose early decisions. Short sessions have none to lose.

No Compounding

The work done yesterday doesn't make today faster. Every session has the same ceiling.

The origin of Helix is not a design document. It's a conversation that started with the question:

Why does every session feel like the first time?
the principle

Epigenetic Intelligence

Biological epigenetics describes how an organism's experience leaves marks on gene expression without changing the underlying DNA. A cell exposed to repeated stress develops methylation patterns that make it respond faster next time. The organism adapts by annotating, not rewriting.

Helix applies this principle to AI infrastructure. Every record has two regions.

CORE IDENTITY (DNA)

id:atom_a7f3x9
name:verify_token
fingerprint:sha256:a4f...
created_at:2026-03
[ NEVER CHANGES ]

META (Epigenetic Marks)

structural:{ ... }
domain:{ ... }
co_occurrence:{ ... }
quality:{ ... }
[feature_N]:{ ... }
[ GROWS FOREVER ]

Atom Occurrence Counting

Every code block is tracked for frequency. Patterns that appear repeatedly are promoted to the permanent library. No human decision required.

→ DNA methylation in biology

Convention Inference

Coding style observed across sessions builds a personalized style guide. The system learns your naming patterns, your indentation, your architectural preferences.

→ Histone modification
§

Compression Profiles

High-frequency phrases are tracked, scored by token savings, and promoted to a personal symbol dictionary. Your vocabulary compresses your own context.

→ Chromatin accessibility maps

Scanner Tier 3 Heuristics

After enough LLM classifications of the same language, the scanner generates its own rules — and stops paying for API calls. It learned the pattern itself.

→ Epigenetic memory across generations

Anomaly / Nudge Lifecycle

Unresolved risks persist across sessions. A flagged decision that was never resolved stays active and resurfaces when relevant context appears again.

→ Stress-induced marks that sensitize

Pattern Auto-Registration

Structural fingerprints that match across sessions auto-register as templates. The system generates diversity from proven segments — no curation needed.

→ V(D)J recombination
the system

Full Architecture

Five biological components. One unified intelligence layer.

Membrane
Receptors
Intake capture, fetch interception, provider detection
Enzymes
Compression proxy — reduces token energy cost
Synapse
Context injector — bridges backend and LLM
Sensors
Tool parser — extracts structured tool usage from stream
Cortex
Intake
Single endpoint, type-based routing
Mitochondria
Three-tier scanner + Haiku classification
Chromosomes
Storage — SQLite + ChromaDB
Tap
Output APIs — context, search, compression
DNA (Forge)
Atoms
Individual code blocks — smallest unit
Molecules
Functional chunks — groups of atoms
Organisms
Complete compositions
Conventions
Inferred coding standards
Fingerprints
Structural hashes for matching
Scanner
Tier 1
Tree-sitter — 100+ languages
Tier 2
LLM analysis — Haiku
Tier 3
Self-generated heuristics
Editor
Expressions
Living pattern instances
Observer
Session logger
Nervous System
Anomaly detection
the atom pool

Patterns That Emerge From Use

The Atom Pool doesn't contain templates someone wrote. It contains templates that emerged automatically from observing what was built. Frequency is the signal. What you do repeatedly is what you care about.

Atoms

Smallest Unit

Individual code blocks, function signatures, config entries. 2–15 tokens. The base vocabulary of your infrastructure.

990 live
Molecules

Functional Chunks

Groups of atoms that combine into working subsystems — auth middleware, DB modules, health checks. 15–100 tokens.

113 live
Organisms

Complete Compositions

Full MCP servers, service stacks, API routers assembled from molecules. 100+ tokens. Ready to instantiate.

growing
// How Patterns Auto-Register
📥
Code Enters
via Intake
🔬
Scanner
Analyzes Structure
#
Fingerprint
Generated
2+ Matches
→ Auto-Register
the loop

Information Flow

Every user message flows through Membrane, fans into four parallel processing channels, converges at Context Builder, passes through compression and injection — then the LLM response loops back to start the cycle again.

T1

Tree-sitter Universal Parsing

Structural analysis of any code in 100+ languages. Runs on every file at near-zero cost. Identifies structure without understanding meaning.

Cost: ~$0.00 / file
T2

LLM-Powered Analysis

When Tier 1 needs deeper understanding, Haiku classifies the code block — purpose, type, quality, reuse potential. Only runs when Tier 1 is insufficient.

Cost: ~$0.001 / file
T3

Self-Generated Heuristics

After N Tier 2 analyses of the same language, the system generates its own rules. Those heuristics run before the LLM on future analyses — eliminating the API call entirely.

Cost: $0.00 (learned, no API call)
adaptive symbol schema

A Private Language That Grows With You

High-frequency phrases are promoted to symbols in your personal dictionary. The LLM always receives the full phrase. Your context window carries the compressed form. Token savings compound with every session.

// before compression
"authenticate_request_with_bearer_token"
"FastAPI router with dependency injection"
"Docker compose service definition with
  Traefik labels and health check"
"Helix Cortex intelligence backend"
"MCP server registration protocol"
// after promotion
§authtk
§fapirt
§dcsvcdef

§hxcort
§mcpreg
400+
Active symbols
Daily
Promotion scheduler
Keyspace growth
0
Atoms in Pool
0
Molecules
0
Live Symbols
0
MCP Tools
0
MCP Servers
0
Live Containers

Ready to Build
Intelligence Infrastructure?

Helix is in active development. We're looking for early builders, researchers, and teams who want to compound their AI workflows before everyone else does.

Request Early AccessRead the Architecture
3 deployment paths: Free (extension) · Paid (hosted) · Self-Hosted (Docker)