Sanjay Gidwani is Chief Operating Officer for Copado, a leading DevOps and testing platform for enterprise SaaS deployments.
Agentic AI has been front and center recently, with evangelists such as Marc Benioff of Salesforce leading the charge. But what exactly does it mean, and does it truly provide a return on investment?
Here is a brief definition: Agentic AI refers to AI systems that can operate autonomously to pursue goals, make decisions and take actions without requiring constant human intervention.
But how will agentic AI transform enterprise application software and impact your business? Let’s examine the factors that make up the investment side of the equation before looking at the potential return.
What Agentic AI Investment Looks Like
The most obvious cost component is that of the software and services that deliver this new agentic capability. Most vendors are looking for a return on their investment, so you can expect an increase in license fees for the software. You can also expect a change in the billing structure from a “per seat per month” fee to usage-based or outcome-based.
Next is the cost of process transformation. Adopting agentic AI will likely require minor retooling to enable 100% automation. Sadly, some companies still use humans to integrate systems, such as the so-called swivel chair integration. They put off investing in API-based integrations, but will be forced to make those changes to take advantage of AI agents.
Then, as with any change in process, human employees will need training. Their roles will transform from doers to directors, auditors and advisors. It is unlikely that, in the short term, companies will trust agents completely. Humans will still be in the loop, but they will need to be trained on how to work in this new environment. How do you correct an agent and set them on the right track?
There will also be costs due to “agentic mistakes.” Ideally, the adoption of agents will reduce the number of errors in your process, as noted in the return section below. In the short term, plan on a bumpy ride as you dial in the new process and tools.
Finally, remember: Automating a bad process can result in more mistakes in a shorter period. We may see a new type of malpractice insurance for agents.
The Return On AI Agents Investments
Many expect a massive productivity improvement and reduced costs, but perhaps a moderate improvement is a better expectation. Agentic AI vendors have focused on agents boosting productivity, not eliminating jobs. Companies have been running lean for the past few years, so it is possible that a reduction in labor costs will not be a significant contributing factor.
There is always a small percentage of underperforming employees who have been retained due to the cost of finding a replacement. The productivity boost from agentic AI will enable companies to finally eliminate their poor performers, resulting in some cost savings. The productivity gains of the remaining team members will mean these workers will not be replaced.
In software development, DORA metrics are a set of four metrics used to analyze performance. These concepts are a great way of summarizing the productivity improvement factors that will lead to a return on the agentic investment:
• Lead Time: How long does it take to complete a process? In the case of a bank loan approval, for example, the less time it takes to approve a good loan, the faster that loan starts generating interest revenue. If done right, agentic AI should significantly improve lead time.
• Throughput: Agentic AI should increase the number of loans made in a given period, increasing top-line revenue. This assumes an unconstrained pipeline of demand, but if you use agents in your demand-generation process as well, then you’re all set.
• Fixing Bugs: In software development, finding bugs before they are released saves money. The same is true for every process in your organization. Mistakes in accounts receivable and accounts payable can have a huge impact on the bottom line. Agentic AI promises to reduce human error, improving the percentage of good outcomes.
• Restoring Services: You will still have failures, both human and agentic. The question is, how fast will you be able to recover from it? In enterprise software, monitoring systems are a key practice to ensure that defects are discovered and corrected quickly. Agents can be used to monitor your process outcomes as well and alert you when something appears to be off. Processes should be designed with the idea of rolling back issues quickly when discovered, providing another opportunity for an agent.
Conclusion
Agentic AI has already demonstrated substantial value in initial use cases—particularly in customer service—and will inevitably permeate all business functions. As with any significant digital transformation, careful prioritization and strategic investment in change management, prompt engineering, validation systems and seamless workflow integration will determine your success.
By proactively embracing these practices, you position your organization not just for moderate productivity gains but for transformative business outcomes.
Be realistic about costs, patient with adoption and vigilant in learning from early-stage mistakes. This approach ensures not only a solid initial return but also ongoing growth in value as your organization becomes adept at harnessing the full potential of agentic AI.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
