Host managed LangChain Apps in seconds.
Async compute, data storage, embedding search, and custom endpoints built-in.
Hosted LangChain Examples
Click for a Replit you can clone and deploy
Self Ask with Search
Host a self-ask with search endpoint.
Persistent Chatbot
Host your own LangChain-based chatbot that remembers user input over sessions.
Summarize YouTube Videos
Summarize the audio track of YouTube videos with LangChain and Steamship.
Question Answering with Sources
Use persistent embeddings to answer questions with sources.
Simple LangChain Prompting
Example of a LangChain-based prompt API on Steamship.
How does hosted LangChain work?
- Swap
steamship_langchain
classes for selectlangchain
counterparts to get fully-managed hosting. - Wrap your code in Flask-style endpoints to deploy a live, multi-user API.
Send us a PR!
Supported Swap-ins
- LLMs
- Swap in Steamship's
OpenAI
class to gain cloud logging and key management. - Buffer Memory
- Use Steamship's
ConversationBufferMemory
class for multi-user, cloud-hosted caching. - Windowed Buffer Memory
- Use Steamship's
ConversationBufferWindowMemory
class for multi-user, cloud-hosted caching. - Google Search
- Swap in Steamship's
SteamshipSERP
class for cloud-managed Google Search. - Embedding Search
- Use with Steamship's built-in embedding search plugins to create and search vectors.
Hosting
- How do I host my API?
- Just run
ship deploy
to deploy your API to Steamship's cloud. - How can people use my API?
- Steamship auto-generates endpoints protected by key-based authentication.
- How can I publish a demo?
- For certain project types, we will auto-generate a web page where others can try it.
- Can I use any Python?
- Just about. Please be reasonable and add your dependencies to requirements.txt.
- How do I debug?
- Debug like regular Python on Localhost or Replit. If you're stuck, just reach out to hello@steamship.com.
- Where are my logs?
- Create an instance of your deployed package click the View Logs button from web page.
Limits
- Are there usage limits?
- Our free trial gives you enough room to experiment, then visit your account page to sign up for plan.
- What happens if I hit usage limits?
- You'll get an error message from our API telling you what happened and how to view our plans.
- What's include in a paid plan?
- Steamship's paid plans include models from OpenAI, Cohere, HuggingFace Replicate, Assembly, Microsoft, Google, and more. As well as fully-managed embedding search, data management, and endpoint deployment. View the plan page for more!
Response Types
- How do I return JSON?
- Just return a Python dict object. This recipe shows how.
- How do I return an image?
- Just return raw bytes plus a MIME type as shown by this recipe.
- How do I return audio?
- Just return raw bytes plus a MIME type as shown by this recipe.
- Install our Python library
- From the folder containing
api.py
, runpip install steamship
. - Run 'ship deploy'
- Installing our python library makes the
ship deploy
command available. Your first time, you will be walked through a few questions to configure your package. - Share the results
- When deployment succeeds, you"ll see a URL that lets anyone try your API on the web or use it over HTTP.