Stride Conductor employs collaborative AI agents to mitigate unmanageable backlogs of technical debt, from untested code to static analysis errors
Businesses today are under constant pressure to develop high-quality software and applications rapidly, while keeping costs under control. However, per a recent column in Forbes, 33% of developer time is spent on tech debt-related tasks versus working on new products and features. These tasks are usually mundane and time-consuming for the developers assigned to them, and may cause low motivation and add to high churn rates.
This is where technologies like Generative AI (Gen AI) can make a difference, opening up new possibilities for automating tedious coding tasks and adding capacity to development teams.
Gen AI coding led to a seismic shift in software development. It has propelled the emergence of coding assistance tools, some enhancing developer productivity with intelligent assistance, while others pioneer autonomous coding technologies. This points to a future where AI handles the bulk of coding, transitioning developers from coding tasks to strategic project oversight.
To capitalise on these advancements, Stride has introduced Stride Conductor, a novel solution that uses natural language to direct a team of LLM agents. These agents collaborate to develop, enhance, and test software in the user’s native development environment, essentially augmenting their team. Stride opted for a multi-agent approach because it allows for self-improvement through critique, better decomposition of large tasks, and the creation of inspectable, readable, and explainable code.
With Stride Conductor’s multi-agent approach, technical debt no longer becomes an expensive inevitability in most IT projects
Technical debt mitigation
With Stride Conductor’s multi-agent approach, technical debt no longer becomes an expensive inevitability in most IT projects, nor does it impede progress. Stride Conductor tackles tech debt in the background by taking over tasks such as:
Static Analysis: Stride Conductor facilitates the high volume, low complexity changes related to linting and validation, security audits, accessibility checks, and web performance optimisation.
Automated Code Testing: Stride Conductor enhances test coverage, fixes failing tests, and adds tests to pull/merge requests to ensure that new or changed code performs as expected.
Replatforming: Stride Conductor can also manage much larger tasks like upgrading language versions, migrating to new stacks, or repatterning existing code.
Case study
Stride Conductor completely changes the calculus around technical debt remediation – enabling projects delayed because of high costs and/or high risk to be delivered with confidence. In a recent project involving a public e-commerce company, Stride was tasked with fixing tens of thousands of static analysis errors in test code, which were preventing the team from upgrading their CI/CD test infrastructure.
These types of errors are generally simple to fix, but the sheer volume was overwhelming, and the project had been deprioritised accordingly. This kind of “dirty job” is perfect for semi-autonomous AI, which can be taught about codebase rules and patterns and then apply them to novel situations. Best of all, it’s straightforward to tell when the work is done – when the code patterns look good and the failing tests pass, businesses can move forward with confidence.
Using Stride Conductor, the company was able to create positive ROI for fixing these static analysis errors, bringing a task estimated at two person-years down to a matter of weeks, including human approvals and oversight. Conductor also offers these advantages over other tools which claim to fix technical debt:
Force Multiplication: Unlike tools that require active operation, Stride Conductor functions like an auxiliary development team, executing tasks semi-autonomously. Human developers need only provide high-level guidance and oversight.
Seamless Interoperability: Stride Conductor integrates into a company’s existing environment, minimising process disruption and maintaining compatibility with existing platforms, stacks and best practices.
Traceability and Transparency: Stride Conductor generates code and a comprehensive chain of thought that can be inspected and reviewed. This end-to-end traceability ensures organisations maintain control and accountability in their AI-augmented development processes.