From Copilot to Collaborator: When AI Starts Doing the Work


Solutions
Industries
Software
Location
AI is changing the relationship between people and software, moving beyond tools that assist with individual tasks toward intelligent agents capable of performing more complex work. For software developers and tax professionals alike, this evolution promises more than greater efficiency. It creates opportunities to spend less time managing processes and more time applying the expertise and judgment that deliver meaningful results.
The first generation of generative AI tools gave software developers a powerful new assistant. Developers could use AI to help write code, troubleshoot errors, explain unfamiliar functions, or solve problems more quickly. Developers still directed the process and remained responsible for most steps, but AI made those steps faster and more efficient.
The next stage of that evolution may be more significant. AI is now moving from responding to individual requests to completing sequences of work. Instead of asking AI to help with one part of a task, developers can increasingly give an AI agent an objective, establish the parameters within which it should operate, and have it work through multiple steps before returning with a result for review. Software development is now shifting from “help me do this” to “do this for me.”
The distinction matters because it changes the relationship between the developer and the technology. A coding assistant typically responds to a specific instruction. An AI agent can potentially be given a broader assignment and determine the steps needed to complete it. For example, emerging agentic workflows can take a bug ticket, identify the relevant service, investigate the problem, propose a fix, and run tests before presenting the work to a developer.
By no means does this make the developer less important, but it changes where the developer’s expertise is applied. Software development involves much more than producing code. Developers need to understand how individual components fit within larger systems, anticipate how a change could affect other applications, consider security and architectural implications, and determine whether a proposed solution actually addresses the underlying problem. As AI becomes capable of handling more of the execution, developers can devote more attention to defining problems, providing context, evaluating solutions, and deciding what should move forward.
This transition is increasingly relevant within Ryan’s own development environment. AI tools are already being used to increase engineering capacity, helping development teams complete fixes and updates more efficiently. The larger opportunity, however, extends beyond speed. It creates an opportunity to reconsider which parts of the development process require people to perform the work directly and which can increasingly be delegated to AI with appropriate oversight.
In that environment, the developer’s role starts to shift. Rather than being responsible for producing every individual input, developers can spend more time mapping out objectives and requirements, giving AI the context it needs, reviewing what its work, and determining whether the result is appropriate. AI takes on more of the execution, but the developer remains responsible for understanding the problem and evaluating the solution. The emphasis from a developer’s standpoint, therefore, shifts from focusing on inputs to analyzing outputs.
That distinction becomes particularly important as AI takes on more complex work. AI systems are probabilistic, drawing on patterns and existing information to produce their outputs. A result can appear convincing without necessarily accounting for every architectural, security, or business consideration surrounding it. Human expertise remains essential because someone still needs to understand the larger context in which that work will operate.
This shift in the development environment has a close parallel in Ryan’s tax software. Tax technology has long helped people perform work more efficiently by organizing information, standardizing processes, performing calculations, and making large volumes of data easier to manage. Even with those advances, users have traditionally remained responsible for many of the inputs required to keep those processes moving.
AI creates the potential to change that relationship. Across tax functions, significant amounts of time can be devoted to collecting and ingesting information, reviewing documents, extracting data, completing forms, performing due diligence, preparing filings, and meeting compliance requirements. AI and agentic capabilities can increasingly perform portions of this work directly rather than simply giving the user another tool to perform it.
There are already practical examples of that progression within Ryan software. Intelligent document processing can use AI to extract text and information automatically, reducing the manual effort required to turn documents into usable data. As agentic capabilities develop, software may also be able to take appropriate actions based on the information it processes.
The greatest opportunity lies in changing how tax professionals spend their time. If technology can handle more of the repetitive work required to get a process moving, tax professionals can focus their expertise on the outcome. Instead of spending as much time collecting information, entering data, and moving work from one step to another, they can focus on reviewing results, identifying exceptions, analyzing outcomes, and determining what those results mean for the business.
Human judgment remains essential as tax processes become more automated. Filings and payments ultimately reach government entities, compliance requirements have real consequences, and many decisions require an understanding of circumstances that extends beyond the information available to an AI system. The goal is to use technology to reduce routine work so tax professionals can focus on applying their expertise.
That also provides a useful test for how AI should be evaluated. AI capabilities should be assessed based on the tangible value they deliver, including time savings, reduced effort, and improved efficiency. As AI takes on more work, important questions follows: What should AI be allowed to do, under what conditions, and where should people remain in control? How that balance is built into the development lifecycle will be the next challenge.

Access data sets across jurisdictions, market value comparisons, and consulting leads.

Manage personal property tax classifications, depreciation, and compliance in one dashboard.
Transform tax into a strategic advantage with deep expertise and innovative technology—explore our services or request a software demo.
Contact Us