Now that the GenAI agent hype is almost tangible, it’s a good time to refocus on practicality and value creation. What needs are agents best suited for, how do they actually work and how do you get started?
To start with the basics, an agent is a computer program designed to perform tasks autonomously by learning from past experiences, observing its environment, making decisions, and executing actions. It is built to achieve specific goals or complete tasks without requiring human intervention. At its core, an agent can be simplified as the intelligent chaining of a large language model’s (LLM) responses through a feedback loop, enhanced by the integration of various tools and the use of effective prompt engineering. So no, this isn’t Skynet but building these tools and solutions requires expertise.  That said, the efficiency they bring to automation is on a completely new level, delivering an incredible leap in productivity. But instead of diving into deep technical jargon, let’s explore how things work in practice.
Concrete Business Case - Sales Team Automation
What better concrete example case could there be than one from sales? In sales, people play the central role, making it an excellent context. So, how can we use GenAI agents to take the performance of a sales team to an entirely new level?
Understand the problem, the real problem
During my career, I’ve already seen far too many projects where the end-user wasn’t listened to and solutions were built without understanding the root cause of the problem. Everything starts with talking to the team members themselves — where do they feel the biggest bottlenecks are in their daily work, where is the most time spent, and what do they consider the most valuable tasks that create the most business value? These are fundamental questions, but they are all too often forgotten. And as team sizes grow, it’s a good idea to use advanced data collection methods, such as process or task mining. These methods reveal bottlenecks and anomalies in operations based on data, rather than intuition. Once these are identified, it’s time to move to the next phase — determining what can and should be automated.
As you can see, there are quite a few tasks that a salesperson needs to handle — and this is just mentioning a few! Without the right support, their workdays will inevitably stretch long, as they often already do. This is where automation can step in to make a difference by streamlining repetitive tasks and freeing up time for higher-value activities.
In today’s information-driven world, customers expect salespeople to do their homework. Generic, copy-paste messages or impersonal calls to schedule meetings often leave a poor impression — quality always trumps quantity. For this reason, customer outreach is something that shouldn’t be overly automated. That said, there are still many tasks where AI agents can add significant value. Here’s an example of what their role could look like:
Agent 1) Customer Analyzer
Salespeople rarely have the time to go through all the available data on a customer—they need the information immediately to provide personalized service. This is where an agent can help, with functionalities designed as follows:
Fetch Relevant Data: The agent retrieves all relevant data about the selected customer from the data platform.
Data Processing and Analysis: The agent processes the data, analyzing it based on predefined criteria (e.g., average purchase value compared to the general average, how long they’ve been a customer, last purchase date, etc.).
Compare Current vs. Target State: The agent evaluates the customer’s current status against the target state, guided by the rules of the customer care model.
Provide an Executive Summary: The agent delivers a concise, data-driven summary, complete with actionable recommendations for the salesperson.
Agent 2) Email Writer
This one’s a bit of a double-edged sword, but sometimes starting the writing process is the hardest part. Building on the output of Agent 1, Agent 2 could draft a complete email, personalizing the message based on the customer’s data. This way, the salesperson doesn’t have to start from scratch but can focus on refining and tailoring the email to make it more engaging and customer-centric.
Agent 3) Meeting Location Optimizer
Everyone knows that scheduling client meetings on opposite sides of the city on the same day isn’t the smartest way to spend your time. Here, an agent could step in by analyzing the office location and the existing meeting locations in the calendar. If there are already meetings scheduled in the same area, it could naturally suggest scheduling new meetings for those days as well. This way, travel time is minimized and overall time management becomes much more efficient.
Agent 4) Sales Material Creator
Creating sales materials is a vital yet repetitive task. While some personalization typically happens, it’s rare to build entirely new solutions from scratch for each customer. That’s why parameterized sales materials are the answer. An agent can generate personalized sales materials based on specific requirements, saving the salesperson from having to manually tweak them day after day. This frees up valuable time for more impactful tasks, boosting overall efficiency.
Every success story begins with architecture
Naturally, a solid GenAI architecture requires rock-solid governance and a strong foundation, and that’s where Databricks comes into play. It was alarming to read a recently published study revealing that as many as 80 % of AI projects never make it to production. This is why the end-to-end process must be streamlined to be as agile as possible, making the building, testing, and maintenance of agents as seamless as possible. Databricks excels in this area, offering everything you need within its ecosystem. The closer you can keep your GenAI agents to your data, the easier your life will be. So, if you haven’t yet started transitioning your GenAI development to Databricks, I highly recommend doing so as soon as possible. For more details, check out my earlier article, where I covered governance made possible by Mosaic AI Gateway in today’s landscape: Simplifying GenAI Architecture with Databricks Mosaic AI Gateway.
One agent at a time, big picture in focus
Teams and roles evolve alongside technology. I recently had an excellent lunch conversation where suggested an exciting new idea for a future business role — the Agent Team Lead. As agents become part of teams as virtual employees, it’s only natural that someone will need to oversee them. This person would ensure the agents' functionality and performance from a business perspective. I was instantly inspired by this idea, so naturally, I’ve shamelessly "borrowed" it to share and spread the message further.
To move from planning to action, it’s important to break the bigger picture into manageable pieces. This is a significant transformation, including culturally, so taking overly large steps in haste is not advisable. A good approach is to build an architecture that serves as the foundation for introducing functionalities one agent at a time. Agents are designed as standalone components, each handling a single task. This way, functionalities can be brought into production faster, and sales team members can gradually adapt to using agents in their daily work. This approach not only facilitates the evolution of work tasks but also creates new career opportunities along the way.
It's good to keep in mind that not everything can or should be automated and delegated to agents. There are numerous reasons for this, but the most common ones are the complexity or critical nature of the process. An automated process needs to be as robust and reliable as possible, which means that every additional step introduces a new opportunity for something to go wrong. For this reason, we recommend building agents for a single dedicated task and assembling a puzzle out of these individual agent pieces. This approach not only makes it easier to get started by delivering tangible business value right away, but it also increases operational reliability and simplifies the continuous monitoring of quality.
GenAI Agents: The way to take your business value creation to the next level
Now that we’ve covered the business case, let’s dive into the results. Multiple agents are assisting the sales team daily, streamlining tasks and improving efficiency step by step, much like small streams forming a larger river. Beyond automation, customer satisfaction increases with more personalized communication, while salespeople enjoy a boost in job satisfaction by offloading repetitive tasks. Once built, agents shine in their reusability and adaptability — and they can even learn new tricks, seemingly with as much enthusiasm as we do! Best of all, they’re highly cost-effective, delivering a staggering ROI.
Building a data platform has been standard practice for years, but leveraging it from a business value perspective is still lagging behind. To truly capitalize on its potential, it requires the integration of business processes on top of it. Otherwise, data utilization remains limited to reporting, leaving ROI questionable at best. Fortunately, GenAI agents have arrived to save the day and your business value creation. If your data is already centralized in one place, it’s sheer madness not to build automation processes in the same location. Enter the Databricks Data Intelligence Platform. Although we’re still in the early stages of this journey, agents provide an excellent way to implement business processes within Databricks. This creates a centralized hub, ensuring governance, data quality monitoring, and usage remain firmly under control. Ikidata is a pioneer in this field and ready to help you take your first steps toward leveraging agents effectively.
Ikidata is a pioneer in GenAI Agent automation, providing deep insights into this emerging technology from technical, architectural, and business perspectives. We make it simple to bring your ideas to life.
Written by Aarni Sillanpää
The future is yours for the taking
Related Posts
