Reliably Assistant helps you answer what is often the most difficult question: “What now?”
After each experiment execution, you can ask the Assistant questions related to your current concerns. It will give you advice on how to mitigate common weaknesses, and recommend new experiments ideas based on your system. The Assistant can even help you write new experiments.
The Reliably Assistant is still a work in progress, but we feel it's one from which you can benefit immediately. The assistant is already available, and we will deploy new features as soon as they've been proven useful.
In the meantime, we'll continue exploring new ways to use LLMs to make your reliability journey more exciting.
Reliably Starters embed questions about the experiments. Those questions are passed to ChatGPT, which provides you with answers and hints about what to do next.
Embed your own questions in your imported experiments to let ChatGPT give you more specific answers about your system.
When using the experiment builder, Reliably Assistant suggests new activities that go along the ones you've added.
Assistant UI Coming Soon
Add new questions directly from the UI. No more JSON editing!
GPT-4 and Experiment Companion Coming Soon
We're currently testing GPT-4's ability to build new experiments for you. Describe your system, explain what you would like Reliably to do, and get even the most complex experiments ready to run in minutes.
Let's explore together!
We're open to user feedback: submit your ideas and requests.
Frequently Asked Questions
Is sensitive information sent to OpenAI?
Currently, the only data sent to OpenAI is the questions you ask as part of your experiments. It means the experiments themselves are not shared with any third parties.
What if I don't want to use Reliably Assistant?
The Assistant must be enabled with the Reliably OpenAI integration. If you don't activate the integration, no data is ever sent to OpenAI. You can also choose to use the integration for some experiments, and not for others.
Are the provided answers safe to use?
As everything that involves Large Language Models and chaos engineering, separately but even more when used in conjunction, you should always proceed with caution. A good rule is to never do or run something you don't understand.
Do I have to provide my own OpenAI API key?
Currently, yes. It will be stored by Reliably as an encrypted secret.
Does using Reliably Assistant introduces additional fees?
The Reliably OpenAI integration requires a valid OpenAI API key. While we won't be charging you, each question you ask will use tokens that you will be billed for by OpenAI.
Reliably Assistant is an alpha feature. Like everything that involves chaos engineering or tempering with your infrastructure, it is to be used with caution. We hope you'll have fun with it, and it will help you win time, but always remember to take what it says with a pinch of salt.