Docs Welcome to KanoScaleWelcome to KanoScale
Introductory How-to Guide
Exciting News: Introducing KanoScale
Section 1: How to Use Sherpa AI
KanoScale is made to be accessible for virtually anyone! Simply describe your desired configurations for the environment you'd like to provision in natural language. Feel free to include and specific technical or business requirements, such as specific services you want to use in your architecture, or budget controls you'd like to enforce.
Start a new project by interacting with the Sherpa AI assistant, or by selecting a blueprint. For this example, we'll start by describing a specific architecture we want to create to Sherpa AI.
Here we prompt for a complex Kubernetes architecture we will use for a ML pipeline. We will need to wait up-to 5 minutes for a response in the current beta version of the product. Please be patient, we are working on optimizing this latency.
FAQ: Why does it take so long to generate a response?
Answer: KanoScale uses a toolchain LLMs for semantic analysis and code generation. Model generated code is put through an extensive validation process consisting of static code analysis and error-correction.
Sherpa AI is designed to generate the most reliable infrastructure-as-code possible, but sometimes we can make mistakes. In our current beta version, code that is generated will require human input and correction up to 60% of the time. As we improve our models, the human input required to correct or validate code will reduce dramatically.
Click on the Manual Action Required notification to make the necessary changes. This will bring up the code editor.
You will be able to easily toggle between the files of your project and find the file that has an error in it. As we can see here, the error is a result of the module not being installed. After we've addressed the issue we can click the Save button. If our changes aren't valid, we'll get the following notification:
If your changes were successful, you will now be able to see architectural diagram of the generated solution. Here you can click on any low level component of your infrastructure to adjust specific configurations or parameters.
For example, we can adjust the <strong>SECURITY GROUP ID</strong> parameter of the AWS Security Group.
Once you're done adjusting configurations and parameters, you can export your Terraform code with the Code Export button. And just like that you are ready to deploy your infrastructure-as-code!
NOTE: Not every solution will require manual action, some solutions can be generated with no errors.
Section 2: Configuring architecture through No-Code UI
The No-Code UI within KanoScale makes IaC maintenance and management a breeze. To enter the No-Code UI, click on a component of the architecture diagram. Here, you can easily modify any low level configuration within the resource or module. Let's try to modify the type of compute instance in our solution by clicking on the orange icon representing our compute instance. We can see the No-Code Panel displayed now. Let's change the instance type to an m4.large.
We should get a confirmation that our change was valid as shown below.
Once the change is successful we can instantly see the estimated cost of our solution go up considerably now that we changed the compute instance type. Make use of the No-Code UI to easily modify and iterate your architecture without worrying about the underlying code!
Section 3: How to import existing IaC to KanoScale
You can easily import existing Infrastructure-as-code files to quickly make changes to your infrastructure without having to directly modify the code base. Once you've imported your project into KanoScale, you can make any necessary changes and even identify opportunities to optimize the cost, utilization or performance of your deployment.<br></br><br></br>To import existing IaC, click on the My Projects button.
Once in the Projects menu, navigate to the right side and click the +Add New button to import an existing project.
You will then be asked to upload your files.
Your code will now get validated. If it passes all checks, the architectural diagram will be generated. If the passes fail, the code editor will be brought up so you can make the required modifications in order to fix the code..
What can I do with KanoScale?
The possibilities are endless. Here are more examples of frequently used architectures that can inspire how you design your next infrastructure-as-code project.
Important Tip
Try to explain your desired configuration in simple, single declarative sentences. I.e., “Create an RDS database called analytics_data”
I want to deploy an environment on OCI that sets up Oracle E-business suite machine Image.I want to deploy and environment in OCI that sets up Oracle JD Edwards machine image.Deploy a serverless application infrastructure on AWS.Deploy a DevOps CI/CD environment in Azure.I want to deploy a Kubernetes cluster on top of Amazon EC2 Kubernetes Service (EKS.)Create a scalable fleet of Microsoft servers on Azure that should load balance the ingress traffic.Create a bastion server on AWS that can serve as a jump-host to a Postgres DB in a private subnet.Deploy a best practice VPC on AWS.Deploy a compute instance on OCI that is HA (High Availability.)Create a multi-node GPU cluster in OCI.Deploy a Windows VM on Azure.Create an Azure serverless function that can write data to Azure SQL DB.Create a serverless datalake solution on AWS.What's next for KanoScale?
We are constantly refining our AI models to become more reliable at IaC code generation and in understanding intended semantics and configurations through natural language instructions. In our public release, our model will allow users to engage in multi-turn chat dialogues where users can make iterative modifications and changes to a generated solution. The model will also be capable of reliably producing more complex, composite architectures.
Check out our roadmap for more information on upcoming product features.
FAQs
Q: Does KanoScale support prompts for any cloud?
Yes, KanoScale will generate solutions for most major Cloud Service Providers (AWS, GCP, Azure, OCI)
Q: Does Kanoscale use ChatGPT to generate code?
No, KanoScale fine-tuned multiple code-generating LLMs with proprietary data. We've also invested heavily in a pre and post-processing pipeline for validating model generated code.
Q: Can other cloud be onboarded to KanoScale?
This feature is coming soon, check out our roadmap for more information
Q: Can KanoScale be used with on-prem deployments?
This feature is coming soon, check out our roadmap for more information
Q: How do I properly phrase prompts?
The Private Beta version of KanoScale performs best when prompted with simple, single declarative sentences. In the Public Release version the model will be capable of handling complex/ambiguous instructions, multi-turn dialogues, and will generate reliable Infrastructure-as-Code for complex composite architectures.