Steamship Packages: full-lifecycle language AI for software developers
Today we’re excited to launch the beta of Steamship Packages.
Steamship Packages are full-lifecycle language AI solutions that developers can add to their code in minutes. They feel like regular software libraries to use, but they run in the cloud on an auto-managed stack designed for natural language processing. That means you can scale from 1 to 10 to 10,000 separate instances without ever managing infrastructure.
Today we’re launching three packages in a private beta:
- ticket-tagger lets you classify support tickets with a zero-shot classifier, and then roll it over to a trained AutoML classifier seamlessly when you have enough data. Here’s a Streamlit demo and here’s the onboarding form.
- chat-analyzer lets you find natural threads, intents, and sentiments in unthreaded Slack-style chat logs. Companies like Ab.bot are using it to build better support tooling around chat. Here’s a Streamlit demo and here’s the onboarding form.
- podcast-searcher lets you build an index around podcasts and videos, and then search it for topic mentions, sentiment, and quote summaries. Here’s a Streamlit demo and here’s the onboarding form.
All you have to do is use them as if they were regular code:
In the coming month’s we’ll be launching self-serve signups, more packages, and an SDK that lets anyone build and share their own. It’s part of our broader plan to build Heroku for language AI — read our launch blog post here.
We built Steamship Packages to make it easier for developers to weave natural language AI into their code. In our language AI work at prior companies, we noticed a huge disconnect between the ease of building a prototype and the complexity of putting it into production.
Packages solve this bottleneck. You can build a prototype in minutes, and you can scale it to production with no changes. The same few lines of code in your prototype are already production ready, because they’re already powered by a production-ready stack in the cloud.
We think this is the missing link to scaling AI into all software.
Take a look at the demos above, and then reach out to us using the onboarding links if your software development team is interested. We’re just getting started and are onboarding over Zoom for the best experience, but we promise it doesn’t take long.
Builders to builders: we’re eager for your feedback and can’t wait to see what you make.