Skip to content

Home

Jump to bottom
Vladyslav Horbachov edited this page Jul 16, 2026 · 6 revisions

DigitalBrain

One living workspace where you and AI agents work together — safely.

Everything addressable is a Neuron. That is not a metaphor.

  • The Login page is a Neuron.
  • The Chat screen is a Neuron.
  • An Approve button is a Neuron action on a Neuron.
  • Gmail is a Neuron.
  • Salesforce is a Neuron.
  • A Feature you install (or write yourself) is a Neuron with real logic.

People and agents share one brain. The app only shows live state. Anything that changes the outside world needs your approval.

One picture: everything is a Neuron

flowchart TB
  You["You<br/>Flutter app"]
  Agent["AI agent<br/>Claude · Codex · Grok · …"]
  MCP["DigitalBrain MCP<br/>the only doorway"]

  You --> UI
  Agent --> MCP
  MCP --> Brain

  subgraph Brain["Your DigitalBrain — one graph of Neurons"]
    direction TB

    subgraph UI["UI Neurons — screens, pages, buttons"]
      direction LR
      Login["Login page"]
      ChatUI["Chat page"]
      FeatUI["Features page"]
      ConnUI["Connections page"]
      ActUI["Activity page"]
      ApproveBtn["Approve / Reject<br/>actions"]
    end

    subgraph Core["Work Neurons — memory, skills, control"]
      direction LR
      Chat["Chat"]
      Feature["Feature<br/>your programmable skill"]
      Effect["Effect<br/>prepared outside change"]
      Approval["Approval"]
      Session["Session"]
    end

    subgraph World["World Neurons — connectors"]
      direction LR
      Gmail["Gmail"]
      Salesforce["Salesforce"]
      More["… other systems"]
    end

    UI -. projects / invokes .-> Core
    Feature --> Gmail
    Feature --> Salesforce
    Feature --> Effect
    Effect --> Approval
    Approval -->|"you said yes"| Gmail
    Approval -->|"you said yes"| Salesforce
  end
Loading

Read it left-to-right in your head:

Layer Examples What it means
UI Neurons Login page, Chat page, a button What you see and click — not a separate app model
Work Neurons Chat, Feature, Effect, Approval Memory, skills, and control
World Neurons Gmail, Salesforce, … Real systems, same Neuron kind as Login

Same kind of thing. Different jobs. One brain.

Why this is powerful (and still simple)

1. One kind of thing

You do not learn five product taxonomies (“screens” vs “tools” vs “agents” vs “plugins”). You learn Neuron.

2. The app is a live window

Screens project Neurons. They do not keep a private copy of truth. When an agent talks in Chat, your Chat page updates because both touch the same Chat Neuron.

3. Connectors are Neurons too

Gmail is not a hidden plugin. Salesforce is not an agent pack. Each is a Neuron with health, capabilities, and (for writes) the same approval rail as everything else.

4. You can program the brain

Because UI contracts and connector contracts are both Neuron contracts, a Feature can:

  • read and write the same surfaces you see
  • call Gmail / Salesforce (with grants)
  • run complex multi-step logic
  • prepare Effects that wait for your Approve / Reject

That is the full power: compose UI + logic + world in one model, without dual brains or silent side channels.

flowchart LR
  A["1 Connect world Neurons<br/>Gmail · Salesforce"] --> B["2 Abilities unlock"]
  B --> C["3 You or agent work<br/>Chat · Features · UI"]
  C --> D["4 Outside change?<br/>Effect → Approve / Reject"]
  D --> E["5 Activity shows the trail"]
Loading

The idea in 30 seconds

Old way DigitalBrain
Chatbot here, apps there One workspace of Neurons
Screens vs tools vs plugins Same kind: UI, skills, connectors
Agents with private tools & memory Agents share your brain
Paste tool URLs / install agent packs Connect once → health + unlocks
Automations that surprise you You Approve / Reject outside changes
Only eng ships new skills Product can grow — Features you approve (or author)

One sentence: DigitalBrain is a personal OS made of Neurons — pages, buttons, chat, Gmail, Salesforce, and the skills you build are the same kind of thing; people and agents share one brain; you stay in control.

What you can do

  1. Talk and work in one place — Chat is real work the rest of the product can see.
  2. Work with an agent — agent acts; your screen updates live. Same Neurons.
  3. Connect real systems — Gmail, Salesforce, … each a World Neuron with clear health.
  4. Stay in control — sends and CRM updates wait for Approve or Reject.
  5. Grow skills on purpose — Features: propose → you decide → install (or write richer logic yourself).
  6. Build your own Neurons — UI contracts + connector contracts mean Features can own complex behavior inside the brain, not in a shadow app.
  7. See history — Activity is the honest trail.

Simple rules (do not dilute)

  1. Everything is a Neuron — Login, Chat page, Gmail, Feature, Approval: one ontology.
  2. One brain — no separate “app world” vs “agent world.”
  3. App = live window — UI Neurons project state; they are not a private database of truth.
  4. Agents are guests — they enter through DigitalBrain MCP, not a tool mall.
  5. You approve outside changes and new Features.
  6. Program on the same rails — custom Features use the same UI and world contracts; they do not bypass approval.

What we are not building

  • Agent marketplaces / “install the Gmail agent pack”
  • Dual brains (UI copy vs agent copy)
  • Silent auto-changes to Gmail or Salesforce
  • “Paste any MCP URL” as the integration model
  • Multi-agent swarm as the core product
  • A second programming model beside Neurons

Day in the life

  1. Login Neuron — sign in; enter your workspace.
  2. Open UI Neurons: Chat · Features · Connections · Activity.
  3. Connect Google → Gmail Neuron becomes Healthy → abilities unlock.
  4. An agent helps in the same Chat Neuron; UI updates live.
  5. Something needs Salesforce → Effect Neuron → you Approve or Reject.
  6. A new Feature Neuron is proposed (or you author one with real logic) → you decide if it becomes part of the brain.

Words (tiny glossary)

Word Meaning
Neuron Any real, addressable thing: a page, a button’s target, Chat, Gmail, a Feature, an Approval…
UI Neuron A screen or destination (Login, Chat page, Connections…) plus the actions on it
World Neuron A connector system (Gmail, Salesforce, …)
Feature A programmable skill Neuron — can use UI contracts + world Neurons with grants
Workspace / brain The shared graph of Neurons you and agents use
Connection Linked account + health for a World Neuron
Capability An unlocked ability (“read mail”, “update CRM”)
Effect A prepared outside change waiting for you
Approval Your decision on an Effect or Feature
Activity What ran, waited, finished
MCP How agents join this brain (doorway, not shopping mall)
UI contract The typed shape of a UI Neuron — so Features and agents can use the same surfaces safely

What’s real vs vision (honest)

Strong today (stack running) Still maturing
Shared workspace + live app Every label saying “Neuron” in the UI
Agent action → UI can update Fewer, simpler agent verbs
Connections health + unlocks Every Gmail/Salesforce path everywhere
Approvals for risky work & Feature install Zero leftover chrome in demos
Neuron model as product axiom Full “build any Feature with UI contracts” surface polished for everyone

More pages

In one line: Pages, buttons, Chat, Gmail, Salesforce, and the skills you build are all Neurons — one brain, live window, you approve what touches the world.

DigitalBrain

Clone this wiki locally