Developers and architects are more often than ever wanting to build AI-powered tools that can assist them with their development journey, but finding time and resources can be a challenge. This is where AEP (AI Enhanced Platform) steps in, offering a platform that includes ready-to-use Generative AI workflows along with a flexible framework to help you and your team easily experiment and innovate with your own AI solutions.
AEP has been developed by a small team at Kainos consisting of myself, Joshua Knight & Matthew Ricketts – in this blog post I will give you an overview of what AEP has to offer.
Why should you integrate AI into your development workflows?
A question that often arises from AI discussions, why should you think about integrating AI into your development workflows? Here are a few reasons as to why:
- Increased Productivity: Automating repetitive tasks within the development workflow can free yourself and team up to focus on the more complex and creative work – boosting developer productivity! There has been numerous research studies showing how using and integrating with AI can suggest that developer tasks can be complete sometimes up to twice as fast when leveraging AI – how awesome is that? Read more of how within this article
- Enhancing overall code quality: Using AI can certainly assist you in enhancing overall code quality including notifying you of potential security vulnerabilities, identifying potential bugs and even code optimisation suggestions – with AI assisting in all this will lead to improving overall code quality and reducing errors within your development workflows
- Faster Development time: Automating with AI various parts of the development workflow will lead to a faster development time, such as using AI to summarise pull requests, asking for feedback of the pull request etc can significantly reduce time to development, allowing you to deploy new features faster
There is of course other reasons as to why you should integrate AI into your development workflows but these three areas of increasing productivity, enhancing overall code quality and faster development time, I certainly feel are core
Tackling the Challenges and seizing the opportunities
Of course, integrating AI isn’t without its challenges. Here are some hurdles you might face and how AEP can help:
- Reducing Adoption Friction: We aim to make it easier for development teams to adopt new AI tools
- Cost-effective Solutions: AEP provides budget-friendly AI tooling
- Mitigation Risks: We address concerns related to accessing and using AI technology
We see the bleeding edge as the point where challenge meets innovation opportunity
Matthew Ricketts – https://www.kainos.com/insights/blogs/using-generative-ai-in-the-development-process-where-challenge-meets-opportunity
Addressing the Main Challenge: Friction in Development
- Usability Focus: No matter how cost-effective or secure a tool is, it must be easy for developers to use effectively. That’s why AEP has been developed to integrate AI seamlessly into development workflows
AEP has been developed to integrate AI into those development workflows!
AEP Platform incorporates a suite of Agents that harness the capabilities of Azure Open AI (& additional Large Language Models (LLMS) in the future) to enhance both GitHub and Azure DevOps development with Generative AI workflows. What does this mean? This integration allows AEP to seamlessly augment your development processes with advanced AI-driven functionalities… more on that later!
By leveraging the above, AEP offers a flexible and scalable solution that can be easily adopted by you and your teams that are already familiar with common CI/CD tools and practices.
The tree pillars of AEP are:
- Pre-built Templates and Tools: A pre-built set of templates, tools and examples to get started
- Guidance and assurance: We provide you with the assurance, messaging and guidance to help your engineers and customers understand the tools
- Framework for Innovation: Build and share your own AEP-integrated tools and templates
We see the journey like this:
Current functionalities of AEP
With AEP version 1, we have rolled out some exciting features:
1. AI-driven pull request summaries (Supports both GitHub & Azure DevOps)
This feature automatically generates a concise summary of the changes within a pull request using Open AI
Benefits:
- Saving time for reviewers by providing a concise overview of the pull request
- Improves understanding of pull request intention and context
2. AI-driven pull request review (Supports both GitHub & Azure DevOps)
This functionality provides AI-generated insights on the changes within a pull request
Key Aspects:
- Provides recommendations for improvements
- Offers advice various areas including: best practices, security and code quality
- Posts updates to the pull request conversation thread
This feature can potentially catch issues early, improve code quality, and reduce the workload on human reviewers. However, it’s crucial to remember that AI reviews should complement, not replace, human expertise
3. AI-driven Pull Request Review with Board context (Azure DevOps only)
Currently an Azure DevOps only feature, it will analyse both the PR changes and associated work items from Azure Boards with capabilities:
- Assesses the completeness of linked work items
- Determines the “done” status of tasks
- Provides a holistic view of the PR in the context of the overall project management
By connecting code changes to project management items, this feature can help ensure that pull requests align with planned work and meet completion criteria. This integration can significantly improve project tracking and delivery
How is AEP built?
AEP is a Python application that uses the FastAPI library, deployed in Azure and uses Azure OpenAI as its AI of choice, below is the high-level user flow design & journey:
Application Journey Flow
With the HLD above showing the deployment in Azure, lets look at how AEP works its magic with the below diagram. Also it highlights what happens when someone throws a curveball into the mix with a prompt injection. You know, when someone tries to be clever and slip in an unexpected input? We utilise Azure AI Content Safety to assist with this:
I’m interested in using AEP, can I? Yes!! AEP is Open Source, the call to action:
At Kainos, we manage hundreds of build systems across many projects, and we know that this can lead to unnecessary repetition when it comes to building common tools. Siloes can be a barrier to experimentation, and with generative AI, the opportunities to experiment are seemingly endless.
That’s why we built the AI Enhanced Platform (AEP) and made it Open Source, we are excited for you to join in on our journey. With AEP, you have a platform that provides easy-to-use Generative AI workflows and a framework to experiment with your own.
Following this hosting guide, you can deploy AEP within your organisation
How AEP can help your organisation
AEP is designed to help your organisation in the following ways:
- Consistency: Provide a consistent way to interact with large language models and community tooling so you can get started right away.
- Cost Efficiency: Reduce the time and resources needed to build AI-backed DevOps tooling, with an average cost per developer at £2 per month.
- Experimentation: A framework to experiment with your own Generative AI workflows, enabling your organisation to move quickly.
- Support: Support your adoption by using our messaging and guidance to help your engineers and customers understand the tools.
- Collaboration: Use the framework to build and share your own AEP-integrated tools and templates.
Contribution & suggestions
We’re actively developing AEP into version 2 and beyond, and we want to hear from you! Your feedback, thoughts, and contributions are invaluable as we take AEP to the next level. Thanks for reading, and I hope this gives you a great overview of the current AEP Platform!
These two blogs I also recommend reading to learn more about AEP: