Background of the AI Capabilities In Spira
In what now seems like an eternity in the timeframes of AI, we released back in October 2023, our initial Generative AI capability for Spira. Using our SpiraApps plugin extensibility framework, the new plugin let you connect Spira to an OpenAI account and execute key activities within Spira using the productivity benefits of AI. For example, it let you create simple test cases and tasks, as well as identify potential risks. Using the (at the time) groundbreaking GPT 3-5mini LLM, this plugin became one of the most downloaded products on the Inflectra SpiraApp marketplace.
During the next 18 months, we released a series of enhanced versions of this plugin, as well as offering support for additional LLMs such as Anthropic Claude, Meta Llama and Amazon Nova, using the Amazon Bedrock platform to host them.
Now we are ready to take the next major step forward in the Inflectra AI roadmap, with the upcoming launch of Inflectra.ai.
What is Inflectra.ai?
Unlike our existing SpiraApps plugins, the new Inflectra.ai capability is a native feature that will be directly embedded inside Spira (and eventually inside Rapise as well) and makes GenAI features easy to use and seamless, with a single click setup to activate within your instance of Spira. Available for both trial and production accounts, Inflectra.ai lets us take on the hassle of managing the LLM and avoids the need to configure Amazon Bedrock or Azure OpenAI and connect to Spira. Instead, Inflectra will be hosting a variety of different LLMs inside its Amazon accounts (one per region to ensure we comply with data residency requirements) and tailoring the LLM being used to the specific use case.
This means that regardless of your hosting provider (AWS, Azure, GCP, etc.) you will be able to use Inflectra.ai to turbocharge the productivity of your team. It also allows you to have a fixed standard cost for the GenAI functionality, bundled into your existing Spira subscription. This takes away any usage pricing uncertainty or anxiety. So, how will it work in practice...
Using Inflectra.ai
The initial release of Inflectra.ai will provide similar functionality to our existing GenAI SpiraApps with the addition of auto-tagging of all AI generated artifacts as either AI Generated or AI Augmented. This will allow you to clearly distinguish which artifacts in Spira are synthetic and will need human review.
You will access Inflectra.ai by means of a new dedicated icon in the navigation bar at the top of the screen (next to the user profile avatar):
When you click on this icon, it activates the new Inflectra.ai sidebar that will be displayed on the right-hand side of the screen. This sidebar is context-specific and will show you all the AI tasks that are available on the current screen. For example, we are on a Requirements Detail page, so it gives us the options to create test cases, identify tasks, generate BDD scenarios, and deduce risks.
When it completes the requested activity, it will generate the appropriate artifacts and then add commentary in the sidebar, with links to each of the generated items. In this example, it has created 4 test cases from the current requirement.
In this example, we can see how the AI has identified 2 risks associated with the current requirement.
Unlike the current SpiraApps, there is a transcript of the current AI session in the sidebar, so you can keep track of what has been generated.
Currently, the AI will generate the items as a standard action, but we plan on allowing you to give feedback to the AI to regenerate the artifacts or adjust the items. For example, you will be able to ask it to generate test cases that are similar to the ones created, but use a different format, or you could ask it to identify additional risks that are in a different category. One of the benefits of the new Inflectra.ai platform is that it is fully cloud-native and can be updated at a faster velocity than the core Spira application. This will let our product teams innovate faster.
How We Built Inflectra.ai
Leveraging the existing Amazon Bedrock SpiraApp functionality, we have migrated the features (as well as additional ones such as the auto-tagging of synthetic artifacts) to our Inflectra SaaS platform:
However, to build this feature, we were able to leverage the help of our AWS partner team to create this revolutionary new feature that will dramatically increase productivity while at the same time having safety guardrails in place that allows us to ensure that this new Inflectra.ai feature complies with our Responsible AI Principles. We will be using the Amazon Nova LLMs as the primary model for Inflectra.ai, as they offer a great combination of high performance and low cost. If we find use cases that it does not handle well, we can dynamically use other Amazon Bedrock models such as Anthropic Claude or Llama.
We are also using some other key AWS AI services such as Amazon Bedrock Prompt Management and Bedrock Guardrails to help us create the best prompts to handle these defined use cases. This aligns with our vision of how Spira can use AI to solve real business use cases rather than simply implementing a trivial chatbot. By doing this, we avoid the need for the users to understand AI prompts, and Guardrails allows us to ensure no undesirable content gets generated (such as hate speech or biased results).
In addition, we are delivering the Inflectra.ai functionality to our users using the native capabilities of the AWS cloud platform, leveraging API Gateway, Lambda Serverless functions and the AWS CDK and CDK Pipelines to deploy the prompts into our production AWS accounts. By decoupling the delivery of the AI functionality from our core Spira application, it will let us innovate and release updates faster.
What About the Existing SpiraApps?
The existing set of AI SpiraApps will continue to be available for our customers, so they can connect them to their own LLM instances hosted in OpenAI, Amazon Bedrock, or Azure Open AI. However we will be focusing all Inflectra development resources towards improving and enhancing the native Inflectra.ai functionality, so over time we anticipate customers will switch to Inflectra.ai unless they have specific IT security requirements that necessitate using their own LLM. In addition, Inflectra.ai is only available for Inflectra cloud customers, so on-premise customers can continue to use the existing SpiraApps.
Future Plans
This initial release of Inflectra.ai largely delivers feature parity with our existing AI SpiraApps, but that is just the beginning. We anticipate adding additional use cases to streamline common tasks in Spira, including creating requirements from project meta-data and context, refining existing artifacts as well as creating new ones, improving the ease of custom reporting, and performing basic application-wide assistance. However as the following diagram illustrates, we're planning on adding, additional major new AI capabilities in our product roadmap for Spira:
The exact scope and sequence are being developed, but we anticipate using additional GenAI methods such as Retrieval Augmented Generation (RAG) and Agentic AI workflows using the native features available in Amazon Bedrock, including Bedrock Knowledge Base and Bedrock Agents. These advanced capabilities will leverage the rich datasets and traceability information resident inside Spira to product insights and predictive analytics unavailable in any other platform. In addition, a medium-quick win will be to potentially integrate the Amazon Q Index API to better connect other enterprise IT data sources such as Salesforce, Jira, Azure, etc. into Spira allowing product owners and CIOs to have access to information currently stored in different silos.
What About Rapise?
Although this initial release of Inflectra.ai is focused on the AI capabilities available in Spira, we are continuing to innovate with GenAI inside Rapise as well. Our next version of Rapise (8.4) will build upon the existing multiple AI use cases supported in Rapise (converting manual tests to codeless automation, generating synthetic test data, and autonomous testing using UITester and UIRobot) to include AI analysis tools for observing large data sets of automated test runs and generating testing summaries and other QA insights. These features currently will use the same "BYOL" model that the SpiraApps pioneered, but later this year we expect to enable Inflectra.ai to be available for Rapise customers as well.
Frequently Asked Questions
- When will Inflectra.ai be available?
The exact date is still being confirmed, but we anticipate the release to be around the end of April 2025
- What versions of Spira will it support?
The Inflectra.ai functionality will be available for all Spira editions: SpiraTest, SpiraTeam, and SpiraPlan. However you will need to be running at least v8.9 of Spira.
- Do I need to be a cloud customer?
Yes, this functionality is only available for cloud customers. So if you are an on-premise customer, this might be a good time to migrate!
- What is the pricing model?
We have not yet finalized the exact prices, but the pricing will be based on the number of Spira concurrent users you have. It will not be usage-based.
Stay tuned for more exciting information about inflectra.ai!