Skip to content

Everything is a Neuron

Jump to bottom
Vladyslav Horbachov edited this page Jul 16, 2026 · 4 revisions

Everything is a Neuron

This page is the product idea behind DigitalBrain. No engineering manual — just the model everyone can use.

The axiom

Everything addressable in DigitalBrain is a Neuron.

Differences are what a Neuron does (show a page, hold chat, talk to Gmail, run your skill). They are not different product kinds with different rules.

If you can point at it in the product, it is a Neuron (or an action on one).

Same thing, three everyday jobs

flowchart TB
  subgraph one["One kind: Neuron"]
    direction LR
    UI["UI<br/>Login · Chat page · button"]
    Work["Work<br/>Chat · Feature · Effect · Approval"]
    World["World<br/>Gmail · Salesforce · …"]
  end

  UI -->|"projects & invokes"| Work
  Work -->|"uses with grants"| World
  Work -->|"outside write?"| Effect["Effect + your Approval"]
  Effect --> World
Loading

UI Neurons — what you see and click

Example Neuron?
Login page Yes
Chat page Yes
Features / Connections / Activity pages Yes
Approve / Reject on a card Action on a Neuron (still the same model)

The app is not a second database. It renders UI Neurons and invokes them. Routes are how you navigate to a Neuron address, not a separate app tree.

Work Neurons — memory, skills, control

Example Role
Chat Conversation work you and agents share
Session Who is signed in
Feature Installable / authorable skill with real logic
Effect Prepared change to the outside world
Approval Your decision

World Neurons — connectors

Example Role
Gmail Mail system as a first-class Neuron
Salesforce CRM as a first-class Neuron
(next systems) Same pattern: connect → health → unlocks

Not “agent packs.” Not “paste an MCP URL.” Connect once, see health, unlock abilities.

Why Login and Gmail are the same kind of thing

It feels surprising the first time:

  • Login feels like “UI”
  • Gmail feels like “integration”

In DigitalBrain both are Neurons:

  • Login has contracts (authenticate, project the sign-in surface)
  • Gmail has contracts (read mail, propose send, …)
  • Both have identity, state, and clear actions
  • Both can be described, read, and invoked through the same brain

UI is not a special snowflake. Connectors are not a special snowflake. That is what makes composition possible.

Program the brain your way

Because UI contracts and world contracts live in the same Neuron model, you are not stuck with only the built-in screens.

A Feature is a Neuron that can hold complex logic and, with your approval and grants:

  1. Use UI Neurons — project status, offer actions, fit into Activity / Chat surfaces
  2. Use World Neurons — Gmail, Salesforce, … under capability grants
  3. Create Effect Neurons — never silent writes; you Approve / Reject
  4. Compose multi-step workflows that stay visible in Activity 
You (or an agent) propose a Feature
        ↓
You decide (approve install / grants)
        ↓
Feature Neuron runs inside the brain
        ↓
It may call UI + Gmail + Salesforce contracts
        ↓
Any outside mutation → Effect → your Approval

That is the full power in plain language:

You can literally program DigitalBrain the way you want — not by bolting on a side app, but by adding Neurons that speak the same UI and connector contracts as the product itself.

One brain, two doors (not two copies)

Who Door What they touch
You App (sign-in) UI Neurons → same brain
AI agent DigitalBrain MCP Same Neurons by address

If the agent updates Chat, your Chat page updates. There is no “agent memory” that the UI cannot see.

Trust stays simple

Action Rule
Read inside the brain Shared Neurons, your workspace
Change Gmail / Salesforce / … Effect → Approve / Reject
Install new skill logic Propose Feature → you decide
Grow the product Same rails — not a shadow runtime

What this is not

Not this Instead
Five taxonomies (screens / tools / agents / plugins / packs) Neuron
UI brain vs agent brain One graph
Gmail agent marketplace pack Gmail Neuron + grants
Silent automation Your approval on Effects
Programming only “outside” the product Features inside the brain on the same contracts

Tiny mental checklist

When you look at anything in DigitalBrain, ask:

  1. What Neuron is this? (page, chat, Gmail, Feature, …)
  2. Who can invoke it? (you, agent, another Feature — with grants)
  3. Does it touch the outside world? (if yes → Effect + approval)
  4. Is the app just projecting it? (yes — live window)

If those four are clear, you understand the system.

In one line: Login, buttons, pages, Chat, Gmail, Salesforce, and the Features you write are all Neurons — one programmable brain with a live window and a human approval gate.

Next: How it works · Demos · Back: Home

DigitalBrain

Clone this wiki locally