Superflows lets you level up your UX with an AI copilot, without months of engineering time. Just connect your API, test in our playground, and add to your product in a few lines of code.
Hey PH 👋🏼
We’re building Superflows to enable large language models to control powerful software products. Our goal is to give everyone an expert assistant in the software they use every day.
Modern business software is incredibly powerful, with large feature sets and the flexibility to adapt to a wide range of use-cases. This makes the software complex, decreasing usability.
For product managers of these products, this results in low activation and retention. They know the features that drive the most value, but struggle to get users to that point.
We experienced this first hand from both sides and asked ourselves: What if you could just type in a box and have the actions taken for you? And what if software products could add this functionality in a few hours?
With ChatGPT’s release and GPT4, it’s now possible to have AI assistants for every aspect of your job. This would be super useful, but these LLMs aren’t connected to the data and software you already use.
When exploring the problem with companies trying to build an AI assistant like this, we heard a lot of shared problems:
- Prompt engineering LLMs is challenging and unlike other software engineering - a prototype quickly gets to 60% accuracy, then hits a wall. Months of engineering time gets wasted
- Product Managers lack control over AI assistant behaviour because you have to get deep into prompt engineering and code to really move the dial
Our toolkit uses LLMs, chain-of-thought reasoning and fine-tuning to create a fast, reliable, customisable AI assistant for your product.
Here’s how it works:
- Connect your API
- Configure in a no-code dashboard
- Test in our playground
- Deploy in a few lines of code
Our product roadmap is public so you can see what’s coming down the tracks and suggest new features: https://docs.superflows.ai/blog/...
We’d be really grateful for your support and to answer any questions you have!
Cheers,
Matthew
Correct me if i am wrong. I can use this as assistance for my product? its not limited to UI/UX (as you mentioned)? suppose i am building a product say whatsapp then how can i use it in my product?
@adamriha Thanks Adam!
We're really excited by the roadmap.
We plan to fine-tune open source models to let companies avoid sending user data to third parties. Accuracy is always the limitation here, but we have some promising early results.
We're also looking into plot generation, the ability to guide users around the product we are embedded in based on their query, and support for adding documentation the AI can reference
@a007mr exactly what we're trying to solve Anton!
Product managers know the features that drive value for users, but there is often a steep learning curve - especially for powerful products.
We hope to help them make products easier to use
@matthewsuperflows , this gives them the information of the data in a curated format. However how would you be able to train the users to use those complex features?
@san_geetha1 our goal is to make these complex features easier to use. Rather than learn a complex interface, you can now get started with a simple question in a text box
One thing on our roadmap is the ability to direct users to useful interfaces in the product, and eventually guide them about how to use it
@laurent_malka Hey Laurent, so the API the demo I showed is connected to is real -- just the interface is faked.
We have some demos on public APIs coming soon that we will host on our website
@sentry_co It's like having a ChatGPT plugin within products. It can answer questions based on the data inside the product, and take actions for you within the product.
@matthewsuperflows That's cool! What about escaping the playground? Like a companion app in the OS. Interacting with apps. like Spotify, Bloomberg terminal, slack etc?
Learney