A copilot for anything
Embed agents in your app: chat UX, real tools, and explicit human checkpoints before side effects.
Examples
Tactus is designed for executable, verifiable workflows. Examples are the fastest way to learn: read a procedure, run it, then iterate with validation, specs, and evaluations as your guardrails.
Use cases are the quickest way to see what you can build with Tactus. Start with the first two walkthroughs, then browse the rest.
Embed agents in your app: chat UX, real tools, and explicit human checkpoints before side effects.
A simple, high-signal workflow: route messy text into an explicit label set with guardrails.
Browse the complete list, including business process automation and Level Zero Operators.
Runnable, tested examples organized progressively from basics to advanced topics. Each example includes embedded specifications and can be executed directly.
Welcome to Tactus! This chapter introduces the fundamental concepts of the Tactus programming language through simple, self-contained examples.
5 examples
This chapter focuses on working with AI agents in Tactus - the core abstraction for interacting with language models.
3 examples
Real-world classification patterns: route messy text into a small, explicit set of labels.
These examples are designed to run in mock mode (fast, deterministic) and also in real mode (with API keys and/or optional ML dependencies) so you can iterate safely.
5 examples
Patterns for integrating human decision-making into AI workflows. These examples show how to request human input, approval, and feedback during procedure execution.
2 examples
This chapter explores Tactus's built-in standard library - utilities and primitives that make it easier to build robust AI workflows.
3 examples
This chapter covers behavior-driven development (BDD) with Tactus specifications - embedded tests that validate your AI workflows.
3 examples
This chapter explores Tactus evaluations - quantitative assessments of AI workflow performance across datasets.
3 examples
This chapter covers advanced Tactus capabilities for building sophisticated AI workflows with tools, composition, and external integrations.
0 examples
Clone the repository, run examples locally, and see the continuous integration tests in action.
If you're new, start with Guiding Principles, then learn the guardrails stack: validation, behavior specifications, and evaluations.
Start with a minimal workflow, then add guardrails as you iterate.