E.V.I. Execution Canvas

E.V.I. stands for Executable Visual Intelligence — it turns any flowchart into a live AI pipeline. You design a flow visually, and Theo walks through each node, executing logic, calling APIs, making AI decisions, and producing real deliverables — all in real time.

What is E.V.I.?

Most flowcharts are static diagrams — you draw them, share them, and they sit there. E.V.I. makes them executable. Every node in your flowchart maps to a skill— a concrete behavior like "analyze this data", "send an email", or "create a presentation".

When you run the flow, Theo executes each skill in order, passing data from one node to the next. You see the progress in real time on the canvas — nodes light up as they run, and you can inspect the output of each step.

E.V.I. is also an open protocol. Flows can be exported as .evi.json files and run by any compatible runtime.

How It Works

1

Design

Build your flowchart in the normal editor — nodes describe what each step does in plain English.

2

Deploy

Click the "Run with Theo" button at the top of the canvas (or from the ⋯ menu). A cinematic transition takes you to the Execution Canvas.

3

Setup

Theo reads your node labels and infers the right skill for each step. Click "Set up my flow" to auto-configure, or configure each node manually.

4

Validate

The graph is checked for cycles, disconnected nodes, and missing configurations. The toolbar shows Ready, warnings, or errors.

5

Run

Click Run to execute. Each node lights up in real time — yellow while running, green when done, red if failed, blue if waiting for approval.

6

Results

The execution log shows a timeline of every step. Click any node to see its output, AI reasoning, duration, and credits used.

The 5 Skill Categories

Every flowchart node is assigned to one of five categories based on its role in the pipeline. Phase 1 ships with 10 built-in skills:

Input (Triggers)

How and when your flow starts running.

Manual Start — Click Run or say "Run my flow" in Telegram.
Scheduled Start — Run on a schedule — "Every Monday at 8am" or "Every 2 hours".

Processing (AI)

AI-powered analysis and decision-making.

AI Decision — Evaluate data and make a yes/no or categorical decision with reasoning.
AI Summarize — Analyze, research, or summarize content. Can search the web for real-time data.

Integration (Connect)

Send data to external services.

Send Email — AI composes and sends an email from your OpenCharts account.
Send to Slack — Post a message to a Slack channel or DM.

Output (Deliverables)

Create real artifacts — documents, presentations, notifications.

Send Telegram — Send a notification to your Telegram.
Create Document — Generate a rich notes document — reports, briefs, meeting notes.
Create Presentation — Generate a slide deck — pitch decks, reports, course presentations.

Gate (Human Checkpoints)

Pause the flow and wait for human approval before continuing.

Wait for Approval — Pauses execution and sends Accept/Reject buttons via Telegram.

The Execution Canvas

The Execution Canvas is the live view at /editor/[id]/live. It has four main areas:

Toolbar — Run/Stop, Test (dry run), validation status, progress bar, export/import, and run history.
Theo Setup Panel — right sidebar listing every node with its assigned skill. Click "Set up my flow" for auto-config, or click individual nodes to configure manually.
Canvas Overlays — colored rings around each node showing execution state (idle, running, completed, failed, waiting, skipped).
Execution Log — bottom panel with a timeline of every step, including duration, credits used, and AI reasoning.

Node Execution States

During a run, each node shows its current state with a colored ring and badge:

Idle

Not yet reached

Running

Currently executing

Completed

Successfully finished

Failed

Error occurred

Waiting

Paused at a Human Gate

Triggers

A trigger determines when your flow runs:

Manual — click the Run button on the Execution Canvas.
Schedule — cron-based, e.g. "Every Monday at 8am" or "Every 2 hours".
Webhook — triggered by an external event (form submission, API call).
Telegram — say "Run my lead flow" in the Telegram bot.

Human Gates

Add a Wait for Approval node anywhere in your flow to pause execution and require human review before continuing. When a gate is reached:

  1. The node turns blue (waiting state).
  2. You receive a Telegram notification with the data summary and Accept/Reject buttons.
  3. If you approve, execution continues to the next node. If you reject, the run is cancelled.
  4. You can also approve from the Execution Canvas directly via the /api/evi/approve endpoint.

Export & Import

E.V.I. flows are portable. Use the toolbar buttons to:

Export — download a .evi.json file containing the graph, skill bindings, trigger config, and execution metadata.
Import — load a .evi.json file to restore a flow with all its configuration.

The .evi.json format is an open protocol — any EVI-compatible runtime can execute these files.

Test Runs (Dry Run)

Click the Test button (flask icon) to do a dry run. This validates your flow and simulates execution without consuming credits or sending real notifications. Use it to verify your setup before going live.

Start simple — a 3-node flow (trigger → AI summarize → send Telegram) is a great way to see E.V.I. in action. You can always add more nodes and complexity later.
Was this article helpful?

Related Articles

T

Theo AI Overview

Overview of Theo AI — 11 chat modes, creation tools, generative UI, memory, and the sidebar mascot.

S

Skills & Automation

Install skills to extend Theo, automate tasks with hooks and workflows, and manage credit guards.

C

Creating Flowcharts

Learn to create flowcharts with nodes, connections, styling, and the editor toolbar.

A

AI Chat

Conversational AI that creates flowcharts, whiteboards, notes, and presentations from natural language.

Skills & AutomationCreating Flowcharts