Download RuntimeTalk to Aegora
Runtime Download Guide

Download, configure, and run Aegora with the setup that fits your machine.

This page is the customer-facing handoff for the Docker-first Aegora Runtime bundle. It explains what is in the package, how to start it on Mac and Windows, what memory to plan for, and how to use a lower-memory option without Ollama.

Download Runtime BundleView Setup Options
Package Options

Choose the runtime mode that matches the customer machine.

The current handoff stays Docker-first, but customers can choose between the full local AI path and a lower-memory runtime path.

Standard runtime bundle

Starts the full Docker-first MVP runtime with LocalLLM through Ollama, the platform infra layer, the platform core services layer, and the admin console.

Lower-memory runtime bundle

Starts the runtime without Ollama. If an external provider such as OpenAI, Azure OpenAI, Gemini, Claude, Cohere, or another compatible endpoint is configured, ai-assist stays available; otherwise the governed runtime and admin console still come up with reduced AI-assisted features.

Memory Guidance

Plan memory before you start.

These are practical sizing guidelines for local testing and customer trials. Full local AI mode needs materially more memory than the lower-memory runtime path.

Lower-memory mode

Recommended for laptops with limited memory. Plan for roughly 12 GB to 16 GB available to Docker. With an external provider configured, core AI-assisted runtime features can stay available without local model hosting.

Standard mode with Ollama

Recommended for customers who want the out-of-box local AI experience. Plan for roughly 20 GB to 24 GB available to Docker, especially after local models are pulled.

Comfortable full-mode target

For smoother local trials and demos, 32 GB system memory is the comfortable target for the full runtime with local models.

Important: the memory difference is mainly driven by local model hosting. If the customer does not need bundled local AI immediately, the no-Ollama mode is the safer default on smaller machines.
Bundle Contents

What the package includes and what the customer needs to configure.

The bundle is designed so customers can start the runtime without building from source or learning the service graph first.

What is in the package

Preloaded Docker images, compose assets, startup and stop scripts, verification scripts, admin console sources, and the MVP runtime README.

What the customer configures

For the default MVP path, most customers can start first and configure later. For external model providers or production-like integrations, update the included environment file before startup.

What the start script does

It brings up Platform Infra Services, Platform Core Services, and the Admin Console together. Customers do not need to start PostgreSQL, Kafka, Temporal, or the UI individually.

When to skip Ollama

Use the lower-memory mode when the machine cannot comfortably host local models, or when the customer plans to connect OpenAI, Azure OpenAI, Anthropic, Gemini, Cohere, or another external provider instead.

How external AI works in lower-memory mode

If the bundle .env file or shell environment includes supported external provider credentials, the lower-memory launcher starts ai-assist without Ollama and enables external fallback automatically.

Direct bundle link: Download the current Runtime Bundle
Mac Setup

Mac instructions

Use these steps on macOS. Docker Desktop is the expected customer runtime for the current MVP package.

  1. Download the Runtime Bundle.
  2. Extract the archive to a local folder.
  3. Open Terminal in the extracted folder.
  4. Run ./load-runtime-images.sh.
  5. For full mode, run ./run-mvp-minimal.sh.
  6. For lower-memory mode, run ./run-mvp-minimal.sh --without-ollama.
  7. Open http://127.0.0.1:3000 and sign in with the bundled admin credentials.
  8. Use ./stop-mvp-minimal.sh when you are done.
Standard start
./load-runtime-images.sh
./run-mvp-minimal.sh
Lower-memory start
./load-runtime-images.sh
./run-mvp-minimal.sh --without-ollama
Windows Setup

Windows instructions

Use these steps on Windows PowerShell. Docker Desktop is the expected customer runtime for the current MVP package.

  1. Download the Runtime Bundle.
  2. Extract the archive to a local folder.
  3. Open PowerShell in the extracted folder.
  4. Run .\load-runtime-images.ps1.
  5. For full mode, run .\run-mvp-minimal.ps1.
  6. For lower-memory mode, run .\run-mvp-minimal.ps1 -WithoutOllama.
  7. Open http://127.0.0.1:3000 and sign in with the bundled admin credentials.
  8. Use .\stop-mvp-minimal.ps1 when you are done.
Standard start
.\load-runtime-images.ps1
.\run-mvp-minimal.ps1
Lower-memory start
.\load-runtime-images.ps1
.\run-mvp-minimal.ps1 -WithoutOllama
Verification

Validate the runtime after startup.

Customers can do a lightweight runtime check first, then run the deeper MVP proof when they are ready.

Runtime deployment verification

Run the included lightweight runtime verifier after startup to confirm the main services and console routes are reachable.

Four-scenario proof

Run the deeper four-scenario MVP verifier when you want to prove the guided governed operating-path journey end to end.

Mac
./verify-runtime-deployment.sh
./verify-four-scenarios.sh
Windows
.\verify-runtime-deployment.ps1
.\verify-four-scenarios.ps1
Runtime Next Step

Start the runtime that matches the customer machine.

Use the full bundle when local AI is part of the trial, or use the lower-memory runtime option when the machine cannot host Ollama comfortably.

Download Runtime BundleOpen Runtime Overview