CLAI Survey
Project CLAI aims to help users of the command line (developers, devops, and deployers) by allowing for automated assistance on the command line; as well as AI researchers by providing an integrated application environment on which their cutting-edge AI bots can be deployed. As part of this project, we are collecting data internally on various value propositions related to this project from researchers and developers. Your feedback will be crucial in the shaping the project going forward.
Sign in to Google to save your progress. Learn more
Project CLAI Overview
About you
Please tell us a little bit about yourself.
Have you used CLAI previously? *
Which of these best describe your job role? *
Required
Do you use cloud platforms for you work? *
What percentage of your work requires using a command line interface? e.g. bash, powershell, zsh, etc. *
Why do you use command line interfaces?
Which command line interfaces have you used before?
Which command line interface do you use the most? *
Do you customize your CLI? For example, with scripts, plugins, and other tools. *
How many years of experience do you have using the command line? *
Your e-mail address (your email will not be shared or used for any purpose other than those associated with CLAI)
CLAI for Users
CLAI Interaction Patterns
Once you install CLAI on your command line, you will have access to newer ways of communicating with the terminal. CLAI is completely opt-in and can be turned off. If turned on, most of your life on the bash will remain unchanged. However, CLAI skills may be invoked in various ways. In the following we will take you through some of them.
Automation
Automation: This feature allows you to hand off tasks (achieved by one or more commands) to the command line – e.g. hand the deployment of an application to IBM cloud. Note that this feature does not require you to write scripts to enable it.
An example of automation in CLAI (right-click and "open image in new tab" for a larger version)
This feature would be useful for your command line. *
Additional comments (optional)
In-situ Troubleshooting and Support
This feature brings relevant help from online forums, support communities, and/or man pages directly to your terminal so you don’t have to go looking for them when you hit an error.
An example of in-situ support (1)
An example of in-situ support (2)
This feature would be useful for your command line. *
Additional comments (optional)
Proactive Support
Sometimes, the command line can anticipate errors and let you know about it (or even go ahead and fix it in the background) in advance. For example, it could be that you need to free up space on your cloud before proceeding to deploy. In such cases, CLAI would catch and prevent some errors that you would otherwise encounter on the standard command line.
An example of proactive support in CLAI
This feature would be useful to your command line. *
Additional comments (optional)
Natural Language Support
This feature allows you to speak to your command line in natural language, in addition to existing bash commands that work as usual.

For example, you could ask:
>> how do I extract file.bz2

or tell the terminal:
 >> extract all images from this archive .
An example of natural language support in CLAI
This feature would be useful for your command line. *
Additional comments (optional)
Pedagogy
The new command line can chime in from time to time and help you get better at your task. This could be something as simple as letting you know of new features (e.g. tell you that the new way of running Flask applications is >> flask <file> when you run >> python <file> ); or in the long run, even guide you towards becoming a cloud expert.
This feature would be useful for your command line. *
Additional comments (optional)
Changing the Command Line
It is crucial to you that the look and feel of your CLI does not change. *
I would be comfortable acting on suggestions from CLAI for troubleshooting. (see support & troubleshooting example above) *
I would be comfortable handing over control to CLAI to act on its suggestions. (see automation example above) *
Sometimes CLAI may take some time to respond. How much latency are you willing to bear in order to use CLAI features? *
Tell us what other features or capabilities you wish your command line had. (optional)
How likely are you to use a command line interface integrated with CLAI? *
CLAI for Developers
Have you built plugins for command lines before? *
Would you be interested in making skills/bots for CLAI? *
What would be your language of choice to build CLAI skills/bots? *
CLAI and AI
What branch of AI do you work in?
Have you used simulators to train AI systems before?
The CLAI Environment
We make the bash environment available to an AI developer as a generic Environment API. As a bot/skill developer, you do not have to deal with interfacing to the Bash but can instead focus on building any AI skill you wish. The interface to the Bash environment allows execution of actions and sensing of the result of those actions in a manner similar to the classic AI agent architecture [Russell & Norvig 1995, Sutton 1992].
This makes your job of interfacing to the command line easier.
Clear selection
CLAI Planned Features
The CLAI API lets you intercept as well as execute a callback on every user input on the command line (after the user hits return), and lets you respond appropriately (depending on what skills your plugin provides). Your can use this to:
 
• Do nothing (normal life on the command line follows)
• Do nothing but learn from the user input
• Add something to the user input (e.g. a flag that would make the command work)
• Replace the user input by something else (e.g. respond to a natural language instruction for automation)
• Respond to the outcome (e.g. error) of a user command (for in-situ support and troubleshooting)
• Add something to the outcome (e.g. for proactive support and/or pedagogical purposes)
 
An orchestrator then decides whether to pass on your change based on a confidence score (that you provide), after comparing the confidence to changes suggested by other skills/bots plugged into that same bash session.
I would like the API to give me access to user inputs before they hit return as well, e.g. for auto-complete functionality. *
I would like the API to allow me to simulate my bot/skill on the terminal offline, e.g. to train a reinforcement learning agent. *
I would like the API to allow me to simulate my bot/skill on the terminal online, e.g. to anticipate effects of the actions in real time. *
What other features do you want to see in the CLAI API? (optional)
Submit
Clear form
Never submit passwords through Google Forms.
This content is neither created nor endorsed by Google. Report Abuse - Terms of Service - Privacy Policy