Get Started
Introduction
Your Re-Factor TL;DR
Key Components
Re-Factor is designed to enable business teams, data scientists, and engineers robustly design, evaluate, and subsequently implement AI within their business processes. We have a few core concepts that will help you make best use of the platform.| Component | Description | Key Features |
|---|---|---|
| Resource | Our general term for any piece of unstructured or structured data your AI-driven processes will utilize. This can include documents of any type, webpages, emails, and more. | Resource Repository |
| LLM | A LLM is an AI model that is capable of generating text based on a prompt or context. LLMs are used in various forms of AI-driven business processes. | Integrations with LLM Providers |
| Prompt | A prompt is the vehicle for delivering context, instructions, and embedded resources to a large language model. | Runnable Studio |
| Tool | A tool is an interface to an external system. Tools provide a means for your LLM to access external resources, enabling LLMs to go beyond generating text and perform real work. | Tool Registry, Tool Designer |
| Completion | A completion is the basic unit of response from an LLM and can be used to generate text or structured data. | Runnable Studio, Runnable API, Runnable Dashboard |
| Flow | A flow is a workflow that executes a series of tasks in the form of prompts and tool calls. Flows give you a way to exert finer control over sensitive or costly processes. | Runnable Studio, Runnable API, Runnable Dashboard |
| Agent | An agent is a highly intelligent AI that is capable of choosing in real time the prompts and tools to use to achieve a desired outcome. | Runnable Studio, Runnable API, Runnable Dashboard |
| Evaluation | An evaluation is an assessment of specific characteristics of the outputs of your AI-driven actions in your processes. | Laboratory, Experiment Log |
| Experiment | An experiment is a vehicle for comparing the performance of different configurations of a prompt, flow, or agent with one another in order to optimize their performance. | Laboratory, Experiment Log |
Using re-factor
We’ve been building AI-enhanced business processes for over a decade. In the LLM era, we AI-driven business process automation has one key challenge: the studio environments for building are unproductive and hard to use. So we designed re-factor to solve everything broken about the current paradigm. Our workflow:- Craft a prompts, flows, and/or agents in the Re-Factor Workbench
- Experiment with different instructions, models, parameters, and more to achieve the lowest latency and cost with the highest accuracy for your workflow.
- Deploy your prompts, flows, and agents to into productive use via our SDK or API.
- Observe the behavior of your prompts, flows, and agents in production to ensure accuracy and cost-efficiency.
- Iterate on your prompts, flows, and agents to continuously improve accuracy and cost-efficiency with assistance from re-factor AI.
Why choose re-factor
We find Re-Factor is best for teams that have a few key goals while developing their AI business processes.- You want total control over every aspect of your prompts, model selection, tool registry, and more.
- You need robust evaluations, versioning, and deployment slots to ensure accuracy in production and continuous improvement.
- You want a dedicated design tool that enables you to iterate far faster to get to production.