Composable Prompts
Unleash LLM power: automate, integrate, optimize enterprise processes.. [Paid]
Last verified:
What is Composable Prompts?
Composable Prompts is an enterprise platform designed to help organizations leverage Large Language Models (LLMs) in their business processes and workflows. It provides an end-to-end solution to govern, secure, and orchestrate LLMs, enabling enterprise teams to design, deploy, and operate LLM-powered tasks that augment and automate business processes and applications.
Key features include an Interaction Designer for defining tasks LLMs perform, a Prompt Library with reusable prompt templates, an API Service for executing interactions, automated training capabilities, and persistence for storing execution runs. The platform supports an extensive selection of both open-source and proprietary models like Claude, LLama2, Mistral, GPT4, and AI21, running on various providers such as OpenAI, Google VertexAI, Amazon Bedrock, Replicate, TogetherAI, and Hugging Face. Its unique abstraction and virtualization layer allows enterprises to effortlessly leverage these models and providers while optimizing for performance and cost efficiency.
Composable Prompts is built for enterprise teams, developers, and organizations deploying LLM technology who need governance, fine-grained security, collaboration, and orchestration. It is particularly suitable for businesses wanting to transition from rudimentary text-based interactions to structured, domain-specific approaches for content processing. The platform was founded in 2024 and raised $4 million in seed funding led by Elaia Partners and Illuminate Financial.
Composable Prompts pricing
Pricing model: Freemium
Composable Prompts uses a PAID pricing model with custom enterprise pricing. The platform does not offer a free plan or free trial for production use. According to available information, the platform is free for developers to test on their app.at composableprompts.com testing environment. Pricing is custom and negotiated based on enterprise needs, likely varying by number of users, execution volume, and required features. The platform is B2B SaaS targeted at enterprise customers.
Composable Prompts pros
- Supports multiple LLM providers including OpenAI, Google VertexAI, Amazon Bedrock, Replicate, TogetherAI, and Hugging Face
- Works with diverse models like GPTx, Claude, Cohere, AI21 Labs, Titan, Llama2, and Mistral
- Unique LLM virtualization engine for abstraction from direct model management
- Enterprise-grade governance for LLM deployment
- Fine-grained security controls for enterprise needs
- Collaboration tools for team-based prompt development
- Orchestration capabilities for complex LLM workflows
- Interaction Designer for visually defining LLM tasks
- Prompt Library with reusable prompt templates
- JSON Schema support for structured LLM responses
- JavaScript Template engine with control blocks (for, if, else) and array functions
- Version control for interactions with draft and published states
- Public interaction sharing allowing other organizations to fork
- REST API and JavaScript SDK for integration
- Execution Run tracking with status monitoring (created, processing, completed, failed)
- Tagging system for organizing execution runs
- Multiple prompt format support (openai, llama2, claude, GenericLLM)
- Cost and performance optimization without provider management complexity
Composable Prompts cons
- No free plan available for production use
- No free trial offered
- Custom/pricing not publicly disclosed
- API availability marked as No on some tool directories
- Primarily targeted at enterprise teams, may be overkill for individuals
- Requires technical knowledge of JavaScript templates for advanced features
- Founded in 2024, relatively new platform with limited long-term track record
- Documentation primarily API-focused, may lack beginner tutorials
Frequently asked questions about Composable Prompts
What is Composable Prompts?
Composable Prompts is an enterprise platform that helps organizations leverage LLMs in their business processes and workflows. It offers an end-to-end platform to govern, secure, and orchestrate LLMs, supporting a large number of execution providers and models with a virtualization approach to operationalize them.
What LLM models and providers does Composable Prompts support?
Composable Prompts supports an extensive selection of both open-source and proprietary models including Claude, LLama2, Mistral, GPT4, AI21, GPTx, Cohere, and Titan. It runs on providers such as OpenAI, Google VertexAI, Amazon Bedrock, Replicate, TogetherAI, and Hugging Face.
What are Prompt Templates?
Prompt Templates are the core building blocks of Composable Prompts. They include JS Template (JavaScript Template engine with string replacement, control blocks like for/if/else, and array functions like map/reduce/filter) and Plain Text (simple format with no variable replacement). Templates are assembled to create interactions.
What is an Interaction in Composable Prompts?
An Interaction is the core concept defining tasks LLMs perform. It consists of prompt segments (template list), an optional JSON Schema for structured responses, and configuration (environment, model, execution parameters). Interactions can be tested in the Playground before production use.
How does versioning work for interactions?
When you create an interaction, it starts as draft at version 1. Publishing creates a read-only copy at the current version and increments the draft version. Published interactions can be made public, allowing other organizations to fork them. The draft version is always what you work on; published versions are frozen in time.
What execution environments are supported?
Composable Prompts supports OpenAI (GPT3.5, GPT4), AWS Bedrock (Anthropic Claude, Cohere, LLama2, A21's models, AWS Titan), Replicate (Llama2 and others), and Huggingface Inference Endpoints (GPT2, Llama2, etc.).
What prompt formats are natively supported?
Composable Prompts natively offers four prompt formats: openai (array of plain text messages with role attribute), llama2 (string with special syntax for tagging segments), claude (string with 'Human:' and 'Assistant:' tags), and GenericLLM (string with User and System tags for Mistral, AI21, Cohere, etc.).
How do I integrate Composable Prompts with my application?
You can integrate by invoking the Composable Prompts REST API with the interaction ID, data input object, and optionally specific execution environment configuration. JavaScript applications can use the JavaScript SDK. First create an API key in project settings.
What are Execution Runs?
Execution Runs are the execution of an interaction, representing both the request to and response from the generative model. Runs have statuses: created (not yet started), processing (executing), completed (successful), or failed (error in error field). Runs can be tagged for easy finding and analysis.
Is Composable Prompts suitable for enterprise security requirements?
Yes, Composable Prompts is designed for enterprise needs with fine-grained security, governance measures, and collaboration tools. The platform equips businesses deploying LLM technology with security and governance measures to ensure proprietary data remains secure and LLM technology can be leveraged safely.