Skip to main content
A Project in Avigrah is a scoped deployment of an AI system, created from a template. It represents a real, working instance of AI configured for a specific dataset, context, or business boundary.

​
Project = Scoped AI Deployment

While AI Templates define how a system works, Projects define where and on what data it operates. This means multiple projects can use the same template, but differ in:
  • Data sources
  • Pipelines
  • Integrations
  • Teams
  • Business context (e.g., city, client, region, or segment)

​
Example

You may use the Customer Intelligence template across multiple projects:
  • Customer Intelligence — Ahmedabad
  • Customer Intelligence — Mumbai
  • Customer Intelligence — Enterprise Clients
Each project:
  • Uses the same AI system design
  • But operates on different data and context
  • Produces independent outputs

​
What Defines a Project

A project is defined by its scope, not just its use case. Typical scoping strategies include:
  • Geography (city, region, country)
  • Client or account
  • Business unit
  • Data source separation
  • Testing vs production environments

​
Why Separate Projects?

Projects ensure clean boundaries and reliable execution.

​
Data Isolation

Each project only accesses its own pipelines and data sources.

​
Independent Execution

AI runs separately for each project without interference.

​
Clear Ownership

Teams and access are assigned per project.

​
Safer Experimentation

You can test changes in one project without affecting others.

​
What “Deployed System” Means

Each project should be treated as a fully configured, live AI system. This means:
  • Pipelines are properly set up
  • Integrations are connected
  • Data is flowing correctly
  • AI agents are producing outputs
A project is not just a setup — it is an active system delivering value.

​
Configuration Layers Inside a Project

Each project activates an AI system by combining:

​
AI Template

Defines system logic and multi-agent structure

​
Data Pipelines

Prepare and structure project-specific data

​
Integrations

Connect external systems relevant to this project

​
Teams

Control who can access and use the AI outputs

​
Best Practices

​
Use Projects to Define Boundaries

Create separate projects when data, users, or context should not mix.

​
Reuse Templates Across Projects

Avoid duplicating logic — reuse templates and vary configuration.

​
Separate Production and Testing

Use different projects for experimentation and live systems.

​
Keep Projects Clean and Focused

Each project should represent one clearly defined operational scope.