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 programs to reinforce AI’s potential to reinforce enterprise processes and drive innovation.
However, as with each expertise, figuring out the place to begin is fraught with difficulties. “When all you will have is a hammer, all the pieces appears like a nail”—or so the expression goes—however in the case of enterprise challenges, not each downside warrants an agentic AI method.
You might have decided candidate areas for agentic AI utilizing an analogous method to that which we described when discussing quickly ideating on worth in a earlier weblog. Nevertheless, how have you learnt if it actually warrants an agentic method, after which, when you’re assured that it does, how do you establish the worth it’s going to convey on your group?
This weblog goals to supply steering on the right way to deal with these areas to empower you to make knowledgeable choices and unlock the complete potential of agentic AI.
Enterprise and technical standards
Primarily based on our expertise working with retail and shopper items corporations throughout the globe, there are some frequent tendencies that may be thought of as standards for figuring out if a particular course of—or a part of a course of—is an effective use case for agentic AI.
These aren’t thought of to be “exhausting and quick” standards that should be adhered to—they’re merely pointers.
- 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, due to this fact, it’s more likely to be way 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 signifies that the precise course of factor 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 might have a worthwhile position making certain these paperwork are right.
- Interplay. A course of that interacts with a number of programs. For instance, updates, reads from, or consolidates knowledge between totally different programs. Processes the place customers should assessment, or consolidate, content material from a number of programs are prime candidates for the applying of agentic AI. Typically known as “swivel-chair integration,” these kinds of 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 knowledge, paperwork, or programs. That is sometimes work that’s mundane and repetitive. Agentic AI can assess and spotlight gaps, variations, or anomalies. It may 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 factor 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 susceptible—which regularly happens with repetitive, mundane human operations. Extra importantly, one the place any errors or points through the course of execution trigger opposed downstream penalties equivalent 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’s an extra requirement, albeit one which should be thought of when architecting an answer. This pertains to knowledge availability.
It’s crucial to make sure that the information required for the agentic AI utility is on the market and accessible with out inflicting challenges elsewhere. It’s frequent that agentic programs must seek advice from knowledge to help decision-making. For instance, it might 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 might be that the agentic answer causes efficiency points within the transactional system. Architecturally, this problem could be averted by extracting this knowledge into an information lake or different knowledge 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. In relation to evaluating the enterprise worth of any expertise funding, one of many first factors to contemplate is figuring out the precise drivers of worth. As well as, understanding the way you’ll measure that is equally necessary.
From the work we have now completed regarding agentic AI, worth sometimes falls into three areas:
- Productiveness. You possibly can 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 extra actions is crucial to delivering worth from agentic AI. For example, one retailer was in search of to liberate 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 increased 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 may typically be seen as cliché. Nevertheless, on this occasion, the main focus is the discount of errors or points. Particularly, those who have a detrimental consequence downstream throughout the group or provide chain. For instance, promising stock that doesn’t exist will adversely influence buyer satisfaction scores and will effectively lead to future misplaced gross sales.
Measurement is essential
For every of those worth driver areas it’s necessary to determine the metrics or KPIs that that is more likely to influence in your particular case. The graphic above provides some examples, however that is the place the worth of agentic AI actually comes into power.
For the productiveness worth driver, liberated time can be utilized to determine extra income producing alternatives, which may 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, nonetheless, is the place it turns into attention-grabbing. Figuring out the downstream detrimental penalties of a delayed or misinformed resolution could be tough, however it’s worthwhile. One method to contemplate is to make use of Microsoft Copilot to assist ideate on this, asking for strategies as to what the detrimental downstream penalties of errors in a selected course of could be. This may occasionally not yield the precise reply for your small business, however follow has proven that it often evokes a brand new thought or perspective that pertains to your small business.
Microsoft Cloud for Retail
Join your prospects, your individuals, and your knowledge.
Transferring on worth
Choosing the correct use instances for agentic AI requires a radical understanding of each the factors for implementation and the drivers of worth. By specializing in high-volume, error-prone processes that require vital human effort and interplay with a number of programs, organizations can determine essentially the most promising areas for AI utility.
Moreover, defining and measuring the worth of AI investments by way of productiveness, course of effectivity, and high quality enhancements will be sure that organizations can unlock the complete potential of agentic AI. With these pointers, 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
Oliver Man, World Trade Architect, focuses on serving to enterprise leaders innovate and compete extra flexibility and effectively. For greater than 25 years he has delivered worth for retail and shopper items corporations throughout the globe with expertise led change. Oliver is a acknowledged Retail Expertise Influencer by Retail Expertise Innovation Hub (RTIH) and can also be a RetailWire BrainTrust panelist.
Felice leads Provide Chain and Sustainability Technique for Microsoft’s Worldwide Retail and Shopper Items Trade Group collaborating with prospects and companions to reimagine knowledge and AI options, and leveraging expertise to drive innovation and higher enterprise outcomes. She has in depth expertise in shopper merchandise, retail, and world manufacturing. Felice has been an angel investor in early-stage startups, the founding father of Delvv, a machine studying studio that created AI-driven interface expertise to reinforce the smartphone consumer expertise and is an advocate for consumer-centric expertise within the cell area.
As a Manufacturing trade architect, Paul works with senior enterprise and IT leaders to assist them perceive the right way to apply Microsoft’s applied sciences and accomplice options to digitally remodel their firm. Leveraging over 35 years of producing trade expertise, Paul works with shoppers to construct their transformations technique, contemplating 4 key pillars of digital transformation: buyer expertise, operational excellence, workforce transformation and product-as-a-service.
