Over the previous 12 months, buyer conversations have shifted from specializing in generative AI to discussing agentic AI. This evolution displays the rising recognition of agentic techniques to reinforce AI’s potential to reinforce enterprise processes and drive innovation.
However, as with each expertise, understanding the place to begin is fraught with difficulties. “When all you’ve got is a hammer, all the pieces appears to be like like a nail”—or so the expression goes—however relating to enterprise challenges, not each downside warrants an agentic AI method.
You might have decided candidate areas for agentic AI utilizing an identical method to that which we described when discussing quickly ideating on worth in a earlier weblog. Nonetheless, how are you aware if it actually warrants an agentic method, after which, when you’re assured that it does, how do you identify the worth it’ll carry to your group?
This weblog goals to offer steerage on easy methods to tackle these areas to empower you to make knowledgeable choices and unlock the complete potential of agentic AI.
Enterprise and technical standards
Based mostly on our expertise working with retail and shopper items firms throughout the globe, there are some widespread tendencies that may be thought-about as standards for figuring out if a selected course of—or a part of a course of—is an efficient use case for agentic AI.
These aren’t thought-about to be “onerous and quick” standards that have to be adhered to—they’re merely tips.
- Quantity. A course of with excessive volumes or variety of interactions. For instance, a shopper items firm receives many extra orders than an plane producer, subsequently, it’s more likely to be much more relevant to use agentic AI to an order consumption course of in a shopper items firm. That doesn’t imply that agentic AI can not assist an plane producer with this course of. It implies that the precise course of component the place it’s utilized could be totally different. For instance, in putting an order for an plane, a number of detailed configuration paperwork could also be wanted, and agentic AI could have a priceless position making certain these paperwork are appropriate.
- Interplay. A course of that interacts with a number of techniques. For instance, updates, reads from, or consolidates information between totally different techniques. Processes the place customers should overview, or consolidate, content material from a number of techniques are prime candidates for the appliance of agentic AI. Typically known as “swivel-chair integration,” a majority of these processes are each tedious and fraught with error.
- Human. A course of the place a excessive stage of human interplay is required. Maybe involving in search of, studying, contemplating, and reasoning over a number of items of data, paperwork, or techniques. That is usually work that’s mundane and repetitive. Agentic AI can assess and spotlight gaps, variations, or anomalies. It could make suggestions to be evaluated by a human and as such, is designed to work alongside or increase the human by lowering the quantity of mundane, repetitive exercise. The human component is crucial right here—AI permits the human to concentrate on exceptions, strategic evaluation, and complicated choices whereas supporting innovation.
- Errors. Processes which are error inclined—which regularly happens with repetitive, mundane human operations. Extra importantly, one the place any errors or points in the course of the course of execution trigger hostile downstream penalties corresponding to delayed deliveries, misplaced gross sales, compensation claims, or dealing with by a human that incurs value or time. This could be a key space of concern and focus.

There may be an extra requirement, albeit one which have to be thought-about when architecting an answer. This pertains to information availability.
It’s crucial to make sure that the info required for the agentic AI software is obtainable and accessible with out inflicting challenges elsewhere. It’s widespread that agentic techniques have to discuss with information to help decision-making. For instance, it could be essential to look one thing up on a buyer or provider grasp file in a transactional system. The place many of those are required in a really brief time, it could be that the agentic resolution causes efficiency points within the transactional system. Architecturally, this problem could be prevented by extracting this information into a knowledge lake or different information retailer to behave as a reference location.
The AI Benefit: How retailers are shaping buyer experiences with data-driven insights

Defining worth
Developments place agentic AI as a cornerstone for making a extra resilient, environment friendly, sustainable, and autonomous provide chain. On the subject of evaluating the enterprise worth of any expertise funding, one of many first factors to think about is figuring out the precise drivers of worth. As well as, understanding the way you’ll measure that is equally essential.
From the work we’ve performed regarding agentic AI, worth usually falls into three areas:
- Productiveness. You’ll be able to consider this as “agentic liberated time.” This displays lowering the non-value-added time related to human interplay in a course of or course of step utilizing the “liberated time” for value-added actions. Scoping these further actions is crucial to delivering worth from agentic AI. For instance, one retailer was in search of to unlock time for his or her provide chain planners to spend extra time with particular person suppliers planning future promotional inventories. AI brokers can streamline communications with suppliers, monitor contract compliance, and resolve disputes effectively.
- Course of effectivity. This pertains to the elapsed time {that a} course of takes. AI brokers automate repetitive duties and optimize operations resulting in larger course of effectivity ranges and decrease prices. This in flip has follow-on advantages—for instance, lowering the time spent between receiving and processing a buyer order interprets to improved buyer responsiveness.
- High quality. This could typically be seen as cliché. Nonetheless, on this occasion, the main focus is the discount of errors or points. Particularly, people who have a adverse consequence downstream inside the group or provide chain. For instance, promising stock that doesn’t exist will adversely impression buyer satisfaction scores and will effectively end in future misplaced gross sales.

Measurement is vital
For every of those worth driver areas it’s essential to determine the metrics or KPIs that that is more likely to impression in your particular case. The graphic above offers some examples, however that is the place the worth of agentic AI actually comes into pressure.
For the productiveness worth driver, liberated time can be utilized to determine further income producing alternatives, which might improve your income per worker KPI. For course of effectivity, lowering misplaced gross sales could be a related metric if, for instance, you’re automating your buyer order course of.
High quality, nevertheless, is the place it turns into attention-grabbing. Figuring out the downstream adverse penalties of a delayed or misinformed resolution could be tough, however it’s worthwhile. One method to think about is to make use of Microsoft Copilot to assist ideate on this, asking for options as to what the adverse downstream penalties of errors in a selected course of may be. This may occasionally not yield the precise reply for your small business, however follow has proven that it normally evokes a brand new thought or perspective that pertains to your small business.

Microsoft Cloud for Retail
Join your prospects, your individuals, and your information.
Transferring on worth
Deciding on the fitting use instances for agentic AI requires an intensive understanding of each the factors for implementation and the drivers of worth. By specializing in high-volume, error-prone processes that require important human effort and interplay with a number of techniques, organizations can determine probably the most promising areas for AI software.
Moreover, defining and measuring the worth of AI investments via productiveness, course of effectivity, and high quality enhancements will make sure that organizations can unlock the complete potential of agentic AI. With these tips, organizations could make knowledgeable choices and navigate the complexities of AI use case choice, in the end driving innovation and effectivity.
Study extra about agentic AI
