Qualtrics Unveils 'Experience Agents': AI-Powered Customer Service Revolution

Qualtrics bets its new 'empathetic' AI agents can fix customer service

They're called Experience Agents, and aim to provide personalized, empathetic responses in real time. AI agents are the next frontier of AI assistants, going beyond answering questions to performing tasks. Qualtrics is implementing agents so its technology can interact directly with customers to improve their experiences. On Monday, ahead of Qualtrics' X4 2025 Experience Management Summit, which kicks off on Tuesday, the company unveiled Experience Agents, assistants who can step into customer interactions to learn more about and resolve existing issues. Experience Agents offer speed, empathy, and resolutions, according to the company. Also: OpenAI wants to trade gov't access to AI models for fewer regulations "Our approach here was how can we expand on what customers are already using us for, to enable them now to act on every single interaction," said Brad Anderson, president of products, user experience, and engineering at Qualtrics, to ZDNET. Experience Agents can interact across various customer touchpoints, including surveys, call center chats, online reviews, and other online interactions. The agents will try to close the customer feedback loop by instantaneously providing an answer tailored to individual requirements. Also: Why you should ignore 99% of AI tools - and which four I use every day Qualtrics gave the example of a sports fan leaving feedback about slow food service at a game. The Experience Agent can interact with the fan, learn more about their issue, and respond in a personalized manner. In a briefing before the announcement, I had the opportunity to watch the agent in action. An airline user provided feedback on an issue they handed into a Qualtrics survey. I was impressed by the speed and quality of the AI agents' responses. Not only were the responses immediate, but they were also empathetic, a quality in customer service often missed by traditional chatbots and even professionals. A standout feature is that Experience Agents interact directly with customers to ensure they have a better experience rather than just streamlining internal processes. In the future, Qualtrics expects the agents to anticipate client needs based on previous interactions with employees and the company's industry expertise. Also: Worried about DeepSeek? Turns out, Gemini is the biggest data offender When asked about hesitations companies might have about having AI interact directly with their clients, Anderson said Qualtrics values accuracy and has a team working on AI ethics and guardrails. To help reduce customer hesitancy, the company also makes many research findings available on its Trust Center for clients to explore. We will keep you informed as more news comes out of the conference.

Qualtrics AI Chief - human sentiment data could enable business value beyond traditional experience management

While Qualtrics announced its Experience Agents at this week's X4 Summit in Salt Lake City, the company's ambitions could stretch even further beyond agentic AI, surveys and sentiment analysis in the future. According to Gurdeep Pall, President of AI at Qualtrics, the company's repository of 'human understanding and preference' data could enable it to deliver business value that broadens boundaries of what we would consider traditional experience management. From its origins as a provider of online surveys measuring customer sentiment, Qualtrics has evolved considerably over the past two decades. What began as a data collection platform, sourcing information on customer and employee views, has transitioned into a system that captures nuanced human behavior across multiple touchpoints. This potentially places the company in a unique position to combine its behavioral insights with artificial intelligence to enable new and interesting categories of business applications. It's early days and Pall was keen to state that nothing is set in stone, but it's clear that Qualtrics is evaluating how the vast amounts of data that relate to human behavior could benefit other areas of the business. He told diginomica: Experience is our bread and butter. But you've got to ask yourself, if you really understand humans, what else can you do that is of higher business value? For example, can you do pricing design? Packaging design? Messaging design? Now you're getting into a very interesting space. This obviously represents an expansion of Qualtrics' current, official strategic ambitions - but the line of thought is logical and could well see the vendor enter market segments largely dominated by other system of record providers. Those providers hold data on their areas of expertise, which aim to provide agentic AI solutions in areas such as HCM, CRM, etc - but human behaviour is a much broader proposition. Pall, who joined Qualtrics in 2023 after tenures at both Microsoft and Meta, frames the company's AI strategy within what he describes as an evolving "multi-agentic world" - he rejects the notion that enterprise AI will be dominated by single, monolithic agents. Instead, a more realistic scenario - despite certain vendor attempts - will be a landscape where multiple specialized agents will work together, each handling specific functions. He said: The first idea of agents was that there's this one thing, and it does everything and it's transactional. But actually it's going to be very, very, very different. What's going to happen is things are going to be broken up into multiple agents... each of them does a specific function very well. This multi-agent architecture forms the foundation of Qualtrics' aims, with executives at the event this week telling diginomica that they hope that open standards and architectures will allow for easy agent-to-agent systems in the future. But Pall did indulge us with some future gazing, where he said that if the company can execute effectively on its agentic AI ambitions, there's no reason that Qualtrics couldn't become a "simulator for business", thanks to its data assets. He explained: If you're flying a drone, you understand the physics, you understand the motion, you understand other forces that are working. The same idea applies to digital business. You are able to understand the dynamics of the business. You are able to understand causality. This now becomes a simulator for the business on which I can run any question I want answered. This 'business simulator', he suggested, would allow companies to test strategies and scenarios virtually before implementing them in the real world - extending experience management as a base towards enabling strategic advice: I can say these are 10 candidates, give me two of them which I really should try out, you can really play with them. There is an opportunity there. The potential for Qualtrics' to move beyond experience management into broader business applications is ultimately because of what Pall sees as the company's key competitive advantage: its distinctive data assets. Qualtrics' repository of human sentiment data allows for pattern recognition. Pall said: Because of the biggest database of human sentiment and preference, we are able to train, we were able to find patterns. If I know nothing about you, I will still have a pretty good shot at how you might respond to something. I know it's freaky, but guess what? This is what Google does all day long and makes a lot of money with it. Building on this is what Pall describes as Qualtrics' longitudinal, cross-channel view of user journeys: This is the power of XM, where in an omni-channel world, we are able to capture user journeys across many different touch points. We have this longitudinal view of the user, which is very important context. When a user wants to interact the next time, you can probably predict what they came for. The third critical component, according to Pall, comes from operational data that drives an understanding of causality, not just correlation. Pall illustrated this with a personal example: Today I flew on Delta from Seattle. They asked me the same quick survey they usually do when I landed. If I hit three [out of five] - and if they didn't know that there was a delay of 45 minutes on the tarmac, they wouldn't know why I'm hitting three. These three data assets - human sentiment patterns, cross-channel journey data, and operational causality insights - form what Pall sees as the company's competitive data moat, in what is an increasingly crowded enterprise AI landscape: I think it makes us pretty uniquely situated, because we have signal here, signal there, we can correlate that over time to find causality. These are the things, I think, which give us an interesting place in this world. Alongside its product strategy, Qualtrics is placing a significant bet on open standards for agent interoperability. The company recently announced a partnership with LangChain to develop Experience Agents and work toward an open-source framework for agent-to-agent communication. Pall emphasized the importance of this approach in a market where major players may pursue proprietary strategies: We want the agentic world to evolve much more like the web, with some of these key protocols for agent to agent communication... because Microsoft will not play in this, at least initially. Salesforce will not play, initially. This stance positions Qualtrics as an advocate for openness. It's probably fair to say that this is driven from a recognition that Qualtrics doesn't have the reach of other vendors (yet) across the enterprise to try a walled garden approach, but it's also a move that could appeal to enterprise customers wary of vendor lock-in: Many of them have been burned by lock-in with certain vendors and they don't want that again. We are thinking about how we can create a movement around it, and then create a de facto standard. However, despite its ambitious vision, Pall acknowledges that challenges remain. He said that there is still concern among potential customers about AI reliability and its potential brand impact, stating: I think there's still a lot of concern about this hallucinating AI, and what it can do to the brand. So that's one ongoing concern. In addition, adoption pathways represent another hurdle, with Pall suggesting incremental approaches may be necessary to build confidence. This is why Qualtrics is focusing its initial agentic AI use cases within surveys, intervening with agents to 'close the loop' on customer or employee complaints. By validating the agents effectiveness in a situation where users are used to interacting with the software - surveys - this creates appetite for expansion. Pall said: A chatbot on a website could be a very interesting way to soft launch an agent. So that over time, if more and more people start to use this, this becomes bigger and bigger, the rest of the stuff pushes away, and pretty soon you just have an agent, right? In addition, I put it to Pall that one of the biggest hurdles will be rebuilding trust with users that have been underwhelmed by the ineffectiveness of chatbots that have been placed all over websites. These chatbots have been positioned as tools to get problems solved without having to pick up the phone and talk to someone, but ultimately they fall down as soon as a request doesn't follow a certain set of rules. Agentic AI has the potential to change that, but trust will have to be gained again. Pall agreed and said: Trust has been diminished by chatbots. We'll have to work with our customers on how to do that. The main thing you want to do is you want to, perhaps, visually have something that holds a different promise, compared to the old promise, and how do you go about doing that? We have got some ideas. Qualtrics' vision of expanding from experience management into broader business applications would be an ambitious pivot for the company down the line. This is very, very early days and Pall was very cautious to say that this is a way off yet - but, if successful, it could significantly increase the company's potential and position it more centrally in enterprise decision-making processes. However, several significant challenges lie ahead. In the competitive landscape, Qualtrics will face opposition not just from other CX vendors like Adobe and Salesforce (who have their own customer data repositories), but potentially from enterprise AI platforms and specialized business intelligence tools if it expands into areas like pricing and product design. The company will need to clearly articulate how its approach differs from traditional analytics solutions. That being said, I do think Qualtrics has an advantage with its data. An understanding of human behaviour is less rigid than some of the more process driven datasets out there, held by the other vendors. If it can gain trust and show results with its agentic AI solutions, there will be a willingness to expand this understanding into other business areas and to address other problems. Qualtrics' strategic direction represents a logical extension of its core strengths. By leveraging its human sentiment data in new ways, the company is attempting to create unique value in an increasingly crowded enterprise AI market. It offers an interesting counterpoint to more operational AI implementations. While platform vendors often focus on automating existing processes, Qualtrics is betting that understanding human behavior at scale will unlock new approaches to achieving business value. I for one looking forward to seeing how it executes on this over the coming months and years.

Qualtrics bets on Experience Agents in competitive enterprise AI market

Qualtrics today announced the launch of Experience Agents at its X4 Summit in Salt Lake City, with CEO Zig Serafin saying that the vendor will be providing capabilities that go beyond more transactional agents being deployed by other system of record companies. This claim is supported by what Serafin sees as more 'human-based data' across the Qualtrics X4 platform, with customer and employee feedback data being collected and stored by its customers for a number of years. Qulatrics hopes to convert this repository of 'human understanding' into automated, AI-driven actions, with agents proactively intervening across customer and employee interactions with an organization. Speaking in Salt Lake City this week, Serafin said: Today, we are introducing a radical transformation of the XM platform: Experience Agents. These are specialized, autonomous digital workers. Think of them as an extension of your workforce, and they bring a whole new level of intelligence and decision making and action that's tailored to the thousands of experiences that take place across your entire business. And unlike transactional agents or rules-based chatbots, experience agents are deeply rooted in Qualtrics' unique ability to help organizations understand people. Serafin told diginomica last year that organizations face challenges in maintaining meaningful connections with customers across an increasingly complex digital landscape: We've built channels to enable convenience, but in doing so, we added a lot of complexity, especially in the last 20 years. The only solution is to truly get to know every single customer and every employee, and to deliver what they expect at every touchpoint and every interaction, even as they move across a plethora of channels. Serafin clearly sees Experience Agents as being key to tackling this complexity. Qualtrics is attempting to differentiate its offering from what the CEO terms "transactional AI agents", offered by competitors like Salesforce and Microsoft. According to Qualtrics, while these competitor agents handle routine tasks like content generation or invoice processing, Experience Agents tap into Qualtrics' database of customer preferences and feedback to enable more contextual interventions. An example was provided where a customer tried to buy a pair of headphones on a website, but when trying to fill in the credit card security information, the website failed. Noticing the customer's frustration, an Experience Agent intervenes and completes the transaction via a chatbot - whilst also waiving the delivery fees as an apology for the frustration. Serafin said: Experience agents are now a new defining layer of that entire platform, and it's uniquely possible for us to actually innovate on how agentic AI comes to life, in experience management, because of the centerpiece of the platform, which is actually understanding human beings. It's what drives human connection, being able to understand what drives preferences, what drives interests, what might then indicate future behavior as a result of understanding friction. Experience Agents have a whole realm of capabilities. Capabilities like being able to improve in the moment, being able to fix things, anticipate, personalize - and you'll see a variety of different solutions that are being built on the system. It redefines what the art of the possible can be for experience management. To enable these agents, Qualtrics is working with LangChain, a platform for building applications using large language models. The company will use LangChain's LangGraph Platform to develop its Experience Agents, while simultaneously working on an open-source framework for agent interoperability - a tacit acknowledgment that organizations already have multiple AI agents from different vendors that need to work together. This partnership signals Qualtrics' (correct) recognition that the agent ecosystem is fragmenting, creating integration challenges that could hinder adoption if not addressed proactively. I've carried out interviews with LangChain and Qualtrics' AI President, where we touched on this topic - more will follow later this week. It's still early days on the Experience Agents front for Qualtrics, with no real products being released this week - rather an intention to build slowly and carefully gather feedback from customers. I expect we will see tangible agents announced this time next year. However, Qualtrics did announce a number of AI-enabled features for its XM platform this week, which include: Core to Qualtrics' AI strategy is its customer and employee feedback data, which as we've already noted, the company believes provides a unique foundation for creating more empathetic AI interactions compared to general-purpose models. Serafin has previously described three components that form the foundation of Qualtrics' AI approach, each building on the company's historical strengths while extending them into the AI domain: This technology stack allows Qualtrics to position Experience Agents as having access to what Serafin calls "patterns, journeys, and brand" - essentially claiming that customer behavioral data gives Qualtrics an advantage over general-purpose AI agents in understanding the nuances of human preferences and responses across different contexts and touchpoints. Whilst it's still early days, one use case that Qualtrics highlights as optimal for Experience Agents is real-time intervention during survey completion - building on the company's core survey technology, but shifting it from a data collection point to an active service recovery opportunity. Serafin explained: Imagine someone is taking a customer experience survey, and in the near future, Experience Agents are going to step into that survey moment. They're going to engage directly and solve and react to a problem as it's happening. The same thing can happen in an online product review or on your website or in response to digital frustration, wherever your customer is, the Qualtrics experience agents can fix. They can improve the experience right there and then. So there's no more waiting for customer feedback to learn about an issue, and by the time your customers finish their survey, their problems are resolved. Qualtrics also plans to deploy these agents to monitor digital touchpoints like product reviews and websites - another expansion beyond a traditional survey-based approach. According to Serafin, these agents will not rely solely on explicit feedback but will proactively engage based on behavioral signals: Experience agents won't always wait for a survey response to be able to engage, they'll also listen to omni-channel feedback. So let's say, for example, customers are frustrated with some part of the service, right? Experience agents will anticipate what a large number of customers might want, and they can tap into customer history right within the XM platform, so they can deliver truly personalized service. It's as if you have a team of some of your best company representatives standing by to proactively check in and then address the needs of all of your customers, building loyalty through moments of delight so they can turn frustration into a slam dunk. On the employee experience side of the house, Qualtrics envisions its agents operating in an advisory capacity, guiding team leaders on where things could be improved. Serafin said: Experience Agents are also going to be able to help you with your workforce. For example, let's say that you have a new manager that's having a change in the business, an experienced agent would coach them and would help them confidently drive connectivity in their team. Additionally, Serafin outlined how agents could also play a role in using the human-based data to guide strategic decision making: And then, on a whole another level, Experience Agents will help you build your business for the future. They'll design strategies that enable faster decision making, to be able to best act on market trends that help you stay on top of your game within your industry. Qualtrics' Experience Agents announcement comes at a time when virtually every enterprise software vendor is incorporating AI agents into their offerings, creating both opportunities and challenges for customers considering their adoption strategies. It's an increasingly complex landscape and the company's differentiation attempt centers on its experience data, positioning it as the "empathy layer" for AI interactions - a strategy that seeks to capitalize on growing concerns about AI's potential to depersonalize customer and employee experiences. This does, however, pit Qualtrics against established CX players, like Adobe and Salesforce, who have their own customer data repositories and are rapidly expanding their AI capabilities. The challenge for Qualtrics will be demonstrating that its experience data provides a sufficient enough advantage to justify specialized agents, rather than extensions to existing operational tools that customers already use. Either that, or Qualtrics could seek to push its capabilities further into the backend to encroach on what some of these system of record vendors currently do (more on that later this week). But ultimately Qualtrics' strategy hinges on three key bets that will determine its success in this increasingly competitive landscape. First, as already noted, its repository of experience data provides a competitive advantage over general-purpose AI models. Second, that customers will prefer specialized agents for experience management over extending their existing AI investments from platform vendors like Microsoft, Salesforce, or Google. And third, that it can develop AI agent capabilities fast enough to remain competitive. However, let's remember that for most organizations, AI agents are still somewhat of a pipedream. My belief is that organizations are still grappling with cloud complexity and are testing carefully with some of the more basic generative AI use cases. AI agents still largely operate within single workflows and much of the full benefit will come from inter-agent compatibility. That is a way off yet and customers are waiting for proof points. The key, however, I believe, will be to what extent Qualtrics will achieve this integration with other platforms in the enterprise to support agent-to-agent work. It won't be able to do it all, hence its work and support for LangChain. Open standards and protocols across the market will work to Qualtrics' advantage, but it depends whether the larger cloud platforms will be willing to play ball. That being said, I do think there's an argument to be made for Qualtrics to work as a key facilitator across the enterprise AI agent landscape, given that its platform is often at the front-end of engagement for customers and employees. There's a reason ServiceNow paid a decent sum for Moveworks recently - to get closer to the user. I will be finding out more over the coming days, as well as speaking with customers. Keep an eye on diginomica for further analysis and updates.

Qualtrics says its new AI agents can satisfy most customer complaints without human guidance - SiliconANGLE

Qualtrics says its new AI agents can satisfy most customer complaints without human guidance Qualtrics Inc. today used its X4 Experience Management Summit conference to unveil a set of specialized artificial intelligence agents it says can address most customer interactions automatically and without human intervention. Experience Agents are trained on the company's database of over 15 billion questions and answers derived from customer interactions. They are designed to interact directly with customers and employees in personalized, proactive and empathetic ways, the company said. Experience Agents can be deployed across any channel and respond instantly to address complaints, including offering compensation if approved by company guidelines. They're currently targeted at what Qualtrics calls "moments of friction" but will become more proactive in the future. "Historically, customer satisfaction has been addressed post-interaction, and organizations only close the loop on a small fraction of what comes in," said Brad Anderson, president of product and engineering at Qualtrics. "Our unique value is the ability to understand human emotions and to turn listening systems into action." Qualtrics claims to be the world's largest survey platform, with an average of 50,000 survey interactions per minute. In addition to asking for feedback, the software can detect evidence of digital frustration, such as "mouse thrashing," in which users move a mouse back and forth rapidly, usually in response to frustration. An AI agent can intervene and step the user through the process. The difference between agentic AI and more common forms of call center automation is that "agentic AI is autonomous," Anderson said. "It understands emotion and intent and takes the action that is most appropriate." Qualtrics codeveloped the agents with about 10 customers. Training typically lasted less than 48 hours. Customers can feed a set of runbooks and policies into the system to guide its behavior and specify which actions can operate autonomously and which require a manager's approval. While each Qualtrics customer customizes its surveys to some extent, most ask a basic set of similar questions, Anderson said. Qualtrics used anonymized questions and answers to create demographic and behavioral categories that guide agent responses to the most common complaints. "We've had enough customer conversations that we're comfortable that customers trust us to use that data," Anderson said. Most of the agents run on self-hosted Llama models in a Qualtrics data center but customers can choose whichever large language model they prefer. Customer data "is never used to train a general-purpose public LLM," he said. Model training was reinforced by extensive human review. Reports on agent activity are aggregated in a dashboard called the Location Experience Hub. Managers can create and review responses for agents to deliver. Agents are expected to be available early in the second half of this year.

Adobe Summit 2025 - a week when agentic AI took center stage. Here's why

Some people and companies may wonder if agentic AI is all hype, but there are already some interesting applications in the tech industry -- Adobe is no exception. Amit Ahuja, SVP of Experience Cloud, Platforms, and Products, Digital Experience at Adobe, walked me through the firm's new Experience Platform Agent Orchestrator and several of the new Experience Platform Agents to demonstrate how Adobe is embedding agentic AI throughout the Platform. If you follow Adobe, you know it's been on a journey to use AI to transform its products and platform for many years now, with generative AI being the newest AI capabilities. Last year, it introduced the AI assistant, which enabled conversations over enterprise data. However, Adobe's customers want more than a Q&A. The aim is that AI takes on and finishes complete tasks and does them smarter and faster. This is where the Experience Platform Agents and agentic reasoning come in. First, let's understand how Adobe defines an agent because, let's face it, it means different things to different companies. In Adobe's case, an agent is: Adobe's agent orchestration works within the Adobe Experience Platform, co-ordinating the actions of one or more agents. Ahuja says the firm looked at where the bottlenecks were for marketers and focused the agents on more complex use cases. It also works across the broader ecosystem. Adobe is committed to agent interoperability and helping others extend and customize experiences and workflows based on their specific needs. Ahuja pitches the three pillars of Adobe's agent strategy as: (1) Helping practitioners do more Streamlining tasks, automating processes, and providing intelligent recommendations are ways the agents work for practitioners. Ahuja talks about Audience Agent, Journey Agent, and Experiment Agents, all of which create, optimize, and manage the audience, journey, or experiment. The agents work on a combination of fine-tuned LLM and Adobe's proprietary knowledge. The Audience Agent can be used to create a new email campaign. The agent is prompted with a goal and responds with options and a plan. It can recommend a new audience or suggest an existing audience based on inputs from the marketer. It can do propensity modeling and predict how much of that audience is likely to convert. The agent works with connected datasets, including the CDP, customer journey analytics, and more. At every point, the marketer has the ability to select the audience, request changes, move forward, or even move backward and have the agent re-evaluate. At every decision point, the agent clearly explains its reasoning and recommendations, including the data sources used, how the model was trained, and what factors drove predictions. Ahuja says the core constructs of agentic reasoning -- planning, verification, and backtracking -- are built into all agents. He said it is necessary for humans and AI to work together. (2) Transforming customer experiences This pillar focuses on helping marketers create engaging, personalized customer experiences using agents. The Brand Concierge (Phil Wainewight talks about this here) is an example of an agent in this pillar. There are two parts to the Brand Concierge agent. There's the setup of an experience and the delivery of that experience to the customer. The marketers defines a goal and based on connected data the agent recommends strategies and content. It can also create the experience, both visual and content, following brand guidelines, and it can personalize the experience of customers using known information. The marketer can adjust what the agent creates at every point. The second part is the actual delivery of that experience - a virtual assistant (agent) the customer uses to help them do something. The example provided here was a virtual assistant that helped a customer customize their vacation trip. Adobe provides APIs and tools for developers to create their own agents and interoperate with the Adobe agents. The Adobe Marketing Agent is a great example. It integrates with Microsoft 365 Copilot and is available in applications like Word, Teams, and PowerPoint, essentially embedding Adobe marketing capabilities directly into the 365 workflow. For example, after a team meeting for a new campaign, inside Office 365, a marketer can request that the Marketing agent draft a creative brief, provide audience recommendations, suggest content, and create a project workflow in Adobe Workfront. And the agent can do this by analyzing the transcript, and integrating with the necessary Adobe applications. Marketers can be in Office Copilot during and after the campaign and request performance reports, analyze journey data and recommend improvements, and draft PowerPoint presentations for reporting. Rapha Abreu, VP of Global Design, and Shekhar Gowda, VP of Global Marketing Technology at The Coca-Cola Company, joined Ahuja to share some insights on Coca-Cola's AI journey (Madeline Bennett's write-up of the session can be found here). Ahuja said the company has some of the most agile thinking about implementing AI. Abreu's design team is responsible for ensuring the brand storytelling remains iconic worldwide, balancing timelessness with timeliness across every touchpoint. So, his job is to develop brand guidelines that define how global brands should look or visually behave worldwide. When generative AI came in, they were in the process of shifting from static to more dynamic guidelines. But they needed to do more. They needed a design system that allowed for creative flexibility to support a global presence but infused local relevance but remained consistent with the brand. Brand consistency is paramount for us, and the more we scale, the greater is the risk of misinterpretation of brand guidelines that can lead to loss in brand fit and ultimately a bad quality execution. [...] We needed an AI-enabled system that didn't just replicate designs, but deeply learned and truly understood what makes Coca-Cola look and feel like Coca-Cola, right? That's what Project Fizzion does. Powered by Firefly services and a custom model, Project Fizzion lets designers create intelligent, dynamic design systems from original designs. Designers create their layouts in Adobe Creative Cloud, Fizzion learns everything they are doing and codifies it into a new style ID and applies it across different applications of the design. The agent isn't replacing designers; it's ensuring their vision is fulfilled, executing it across potentially thousands of applications. Gowda said scaling is important for the company, but it was essential to do it in a way that maintains the integrity of the brand across everything. He also doesn't believe that AI replaces creativity - it enhances it. Coca-Cola believes in a human-led approach where designers lead, and AI follows. Another interesting point Gowda made was o emphasise the truism that AI is not a silver bullet. You must start with a clear problem to solve and define AI's role carefully. What I saw related to Coca-Cola's use of AI and design was interesting, and the way the Audience Agent reasoned out and built a new audience was pretty cool. But, the Marketing Agent embedded in the Office 365 CoPilot caught my attention the most. As companies start to dive into using agents and tech providers build agents on top of their products and platforms, it's the interoperability that I haven't seen much of yet. That's partly because the tech providers are still trying to figure out how to create agents that work within their own systems. And some are creating multiple agents that don't talk to each other. Trying to figure out how to design them to talk with agents outside their ecosystem is a bigger, broader initiative. Adobe is there, working with partners like Microsoft to build interoperability between systems and agents. Plus, you're working with an agent in line with the tools you use every day. You don't have to move between tools to get work done - that is a key productivity hack. Tech providers can build all the agents they want, but if they are still requiring multiple interfaces and multiple agents that don't talk to each other, they're missing the point.

Oracle's AI Agent Studio is out - but what changes for customers? Behind the news with Oracle's Steve Miranda

Yesterday, at Oracle CloudWorld London, Oracle announced its new AI Agent Studio. My London-based colleague Phil Wainewright was on the case: CloudWorld London - Oracle debuts its AI Agent Studio and pledges $5bn for the UK. How does this news advance Oracle's AI narrative? Wainewright: Until today, there had been no word of tooling to allow customers to build and customize their own agents. That changes as the CloudWorld caravan rolls into London, where Oracle is taking the wraps off its Agent Studio this morning. This is the same type of Oracle agent tech, just exposed externally to customers (and partners). Wainewright adds: The new AI Agent Studio, available at no extra cost to Oracle Fusion Apps customers, allows users to create new agents or modify pre-built agents, test them and then deploy and manage them across the enterprise. It's based on the tooling already used by Oracle product teams to create the [50+] pre-built agents announced so far. Another notable detail: via the AI Agent Studio, Oracle's customers (and partners) can not only build their own agents, they can do it with a choice of external LLMs (check Wainewright's piece for more announcement info). This matters for AI agent governance as well as creation. As readers know, I have pretty strong views about what agents are good for, and what they are not - and how vendors should approach this with customers. How would that clash with this news? Time to put Steve Miranda, EVP, Oracle Applications Development, in the hot seat once again. Miranda told me this announcement marks another phase in Oracle's AI timeline: A year and a half ago, at the 2023 Vegas CloudWorld, we announced our first set of generative AI-based features - and we announced 50 use cases that were coming. We subsequently delivered 100 and counting. Essentially every place within our applications where you generate text or could generate text, we have an 'AI Assist' button - everything from creating job posts to creating item descriptions, from draft emails to prospects within CX, to narrative reporting on financials. So that's a handful of the 100+ use cases that are out there, and that will continue every time we have any kind of text-to-creation... If you go back to last year at CloudWorld, roughly six months ago, we announced that we'd be building out the next step, which is agents. So 50+ agents, going from a tactical generative AI use case to a step in a business process. So we demoed AI agents around benefits. We have AI agents around supply chain planning. We demoed the AI agent around Document IO for automation of finance and payables, a general ledger agent for inquiry and drill down and audit - a host of these agents. And then, on stage in London, Chris Leone announced the next step in that evolution, if you will, which is the Oracle AI agent studio. With Oracle's AI Agent Studio, you can build teams of smaller, discrete agents, adding human steps where needed - either as a precaution, or a more permanent workflow feature. Miranda explains: I think the term that's popular today is that you create teams of agents. Basically, these are the discrete agents, and you're allowed to put them into connective steps, either via workflow or with the step that's now called 'human in the middle,' where an agent does step one; a human does step two; an agent does step three. In addition to that, we have extensibility. That's where customer choice in LLMs factors in: This comes in two forms: the ability to for our customers or partners to take an agent that we've delivered, and that we will continue to deliver, and to modify that meaning, add prompts, or choose a different Large Language Model. There's sometimes settings within a Large Language Model on how aggressive or non-aggressive you want the Large Language Model to be. Or you can build a completely new agent. For example? Let's take the case of a recruiting process. You can break down a recruiting process into several steps: source candidates, schedule interviews, schedule follow ups, negotiate an offer, do a background check, give an offer. Let's suppose that background check is something outside of the Oracle system. In fact, it probably is. Now, our customers, or our partners, on behalf of our customers, can build an agent that connects to a third party system, does a background check, and comes back into the AI Agent Studio to orchestrate a workflow. Progress? Definitely. But how is Oracle's approach to AI agents differentiated? Miranda makes the case. First, rewinding to the basic message you're hearing. Lots of our competitors nowadays talk about the need for centralized information in order to run AI. You'll hear terms like data lakes, and a lot of different offerings to bring data together. Again, we've been quite proud of the fact that we're the only SaaS application that really gives you end-to-end input into your data. We've always had it for business reporting. Now you have it for business process automation and/or better, AI, because all of your data is, in fact, in one place. Miranda says training the leading models on OCI infrastructure can't be underestimated: Also notably: we're the only application that there's also in the technology business. As the application vendor sitting on top of that, we have direct access - so we call and test against Llama, against ChatGPT, against Cohere, and then, depending on which Large Language Model is best at the time for the use case, we choose that. Miranda cites another differentiator for agent building: a proven/compliant stack. This isn't agents (and Agent Studio) built on top of a generic area. It's natively integrated to the Fusion Applications. So Agent Studio incorporates the agents that we build. It's native to the security model, the UI model, everything to help you best orchestrate your business processes. Miranda has never seen an announcement that energized Oracle's partner community as much as this one. If he is right, that should be an asset to customers, because - let's face it - customers have a big learning curve here. Even if agents are easy to build, deploying them brings a ream of new considerations, from risk management to agentic evaluation to outcome metrics. So, Mr. Miranda, how should your customers get started? When I meet with customers today, what I tell them is, start using the 100 use cases that we've already given you now. Are those like, tremendous business payback? I'm not saying that, but it allows them to get used to what the AI is like. It allows them to go through all of their internal questioning on: what's your AI security posture, what's your AI privacy posture, what's your AI ethics posture? Pre-built tools push users into hands-on mode: This allows the end users to see, 'Here's my use case' - and get used to that. 'Well, wait a minute, if I press the button again, even though it's the same thing, it's going to generate slightly different text for me, because it's, a probabilistic model.' Getting their way of thinking around that certainly helps that transition process. Frankly, the way we built it is improving those use cases, improving those agent use cases. Now we'll improve the end-to-end workflow. Fellow diginomica contributor Brian Sommer and I have an ongoing skirmish on practical versus imaginative AI use cases. Sommer is looking for the truly compelling AI scenarios, well beyond out-of-the-box starters like job description generation. Sommer isn't wrong, but with AI, customers do need to get used to a different approach to app building and use case design - not to mention risk management. Where does Miranda stand? I hear from customers. 'Hey, when am I going to have my close-your-books agent?' Let's say that's what we want to build. Let's say that's what we want to announce. How are we going to do that? Well, there's 1,000 steps in closing your books. You're paying invoices; you're matching the POS; you're creating journals; you're doing allocations; you're doing revaluations, currency adjustments. We're going to build that by building 1,000 discrete steps of these different agents, and then hooking them together. That's how it's going to come about. Oracle's inclusive AI pricing and 'OCI-powered' aspects are different than most enterprise software vendors. I also don't hear the typical over-emphasis on "autonomous agents." For now, building human supervision (and human steps) into agentic workflows is an important option. That said, customers should take risk profiles into account when getting started with agents. Recruitment is a good example of where the consequences of over-automation can be concerning, if not liability-inducing (granted, many companies have already taken the plunge with algorithmic screening tools, perhaps beyond where they should, so in that case, agents don't necessarily introduce a new risk factor. Still, that's no excuse for moving too fast in an area with higher consequences). On the other hand, getting agents involved in sales and service support/interactions is an example of a lower risk profile area where early lessons can be derived. Oracle's emphasis on smaller, specialized agent roles is smart for use case design and accuracy. However, even those specialized agents are not always going to do what the user expects. That's life with probabilistic technologies, not to mention users getting a comfort level with prompts and agentic workflows. This is where the use case selection and design comes in, as well as building in verification and auditing steps (LLMs as auditors are one aspect of this). Oracle has a chance to detail more on how they help customers measure and improve LLM and agent accuracy. It's a topic I will return to with just about every vendor, so Oracle hasn't heard the last of my questions on that. Miranda quipped about how we've moved on from "hallucinations" to "probabilistic." I'm not sure if that's true in general, but it's certainly true for me. Hallucinations conjure up images of ChatGPT issuing a viral, outrageous response. Enterprise vendors can largely control that type of absurd output though narrower models, specialist agents etc. Whereas "probabilistic" is a characteristic you can mitigate, but not eliminate. Ultimately, you line up the right use cases, and avoid ones where the tech doesn't fit or the data quality is low. LLM agents work quite well when fused with more deterministic workflows, the ones I'm constantly told are legacy forms of automation. That's a big reason why the supposed "death of SaaS" is on the wrong track, but that's an argument for another time. Meanwhile, customers need advisory on the best business outcome metrics. Miranda shared several conversations with customers on how they arrived at their metrics, from financial services to HR/talent/recruitment. There isn't one right way to go about this; the point is to have these in place before you roll out. Miranda pared it down: Is AI moving the ball forward on the business metric that you care about? He says to start there; you'll get no argument from me on that.

How Bela Stepanonova imagines and re-imagines the future of AI as Apollo.ai's CPO

Bela Stepanova started coding when she was little. It led her to get an engineering degree. After college, she took a job in technology consulting and worked with companies on Wall Street building training and wealth management platforms. Then, she had an opportunity to work with a small boutique company that hired engineers who also had a business mindset. This job gave her the opportunity to work with customers directly. Looking back, she loved that she could not only think about architecture and how to solve problems but also talk to customers and understand what they were trying to do and why: I wanted to be closer to the problem because I think tinkering with the problem itself and truly understanding is really what drives the right technology and building the right thing for the customer. And that sort of customer obsession set in with me pretty early, and I love that part. In terms of her current day job as Chief Product Officer (CPO) at Apollo.ai, Stepanova explains that a CPO's role is to represent the customer, understand them and the market space, and translate all that into technology. The CPO must inspire engineers to understand the problem and find the right solution, she says. This requires many skills, including data science and data understanding, the ability to research and understand qualitative data, and understanding UX paradigms and AI technology. Stepanova reckons you are constantly learning, and the role is constantly evolving, which makes it fun for her. In here time she has worked for many companies, including as SVP of Product at Iterable, a martech platform, and Senior Director of Product Management at Box, a secure collaboration platform. There is a transferable skillset, she argues, although the platforms are very different. For Stepanova, working on scalability and data is what she loves. Box has grown to support millions of customers, and Iterable has billions of customers in a database; Apollo is also working at scale with data, she contends: For me, personally, it's really fun to take a sort of slightly different angle in a new industry, but yet it has a sort of common foundation. If you work in sales and marketing, you probably know about Apollo. It's a sales platform that helps you identify leads, engage prospects, and close deals. Apollo is designed to help salespeople grow their businesses, which is hard to do today. Companies need to figure out who their customers are and how to reach them. Then, they need to take what they know about them and turn it into a repeatable motion that the entire sales team can replicate. What excited Stepanova about joining the company was thinking about how to democratize best practices in sales and take them to the next level with AI. Most salespeople use up to ten different tools at the same time and spend more time working with the tools than actually selling, she notes. She wants Apollo to remove that challenge so sellers can focus on building relationships and selling. That may start with the user experience. Stepanova loves everything about user experience and wants to help transform how people work. For her, technology must become a way to do the work you need to do easily but in better ways: How do you infuse all that to not have 10 different platforms, not to spend 70% or whatever of your time just figuring out technology. But technology is easy. It's actually there to help you, to guide you. It's proactive, it's contextual, it's dynamic. How do we re-envision that? And that's obviously where AI will come in in the future as well. The people at Apollo were another reason for joining the company. She has worked with and advised many founders and describes the the Apollo team as sharp, ready to take risks and invest a lot more in R&D than a typical company of its size. All that said, what then is Stepanova's vision for the Apollo platform? There's a heavy and long-standing AI focus, of course. She argues that Apollo invested in AI early on and launched some of the first AI agents, but that's only the start. AI agents have become the 2025 buzzword, but they are more than that. Stepanova pitches that AI is the how, not the what, and that the how can help re-imagine and transform the industry. We are at the first phase of AI agents and assistants, she believes, where you get intelligent insights in context at your fingertips and task automation that helps reduce busywork. Asking what AI features a platform has is the wrong question, she insists - you have to go back to the idea of being customer-obsessed: We're not ultimately trying to build AI features. We're trying to solve a problem, and so can we re-imagine how we can solve it? And I think that the UX portion really has become about what's easy to use, and AI will play a huge hand in the amount of UX, and as well as context. What does Stepanova think is coming next? Understanding how AI can help with strategy is important, she suggests, positing that marketers and salespeople are like conductors of a symphony - give AI the music to play, and tell it your goals and what you are trying to do and the AI will find all the ways to help you do that and help you design and execute that strategy. To get from where we are now to there requires two key things: data and context. And Apollo has both, she says: AI is not just a check box to check to have AI. It has to be relevant, has to be simple, has to be quality answers, right? It has to be perceived as accurate by a customer as well. So data in the context really come into play. There also needs to be transparency. It's one thing to build features and another to bring the user along the journey so they trust it, but: How do you really think about transforming the user experience? How does it become contextual right when you need it? How does it become configurable for your business and what you need to do. How does it become dynamic and proactive in what it does? So, really re-thinking the interaction patterns versus just bolting on UX, I think, is the second part to context. And I actually think we're all still building that muscle. Stepanova isn't seeing a lot of best practice use cases yet, but they are starting to emerge, and she sees Apollo out on the curve. The other thing Apollo has going for it is its ecosystem and partnerships. As we start to look at interoperability, figuring out how agents talk to each other is just the beginning. Stepanova says it goes to the strategy phase she spoke to earlier with all these systems talking to each other in a bigger way. Stepanova recently wrote that we have to re-imagine how we build things from the ground up. That's what she's referring to when she says it's not enough to have an agent bolted on to an application; it's about embedding AI throughout the user experience and everything you do. No-one is talking about that right now, she suggests. Today, it's about agents that sit on top of the platform or application and take over some of the interface capabilities. But do you continue to evolve the UX of the application? Or will the application itself go away because you have these agents that will do all the work for you, and you don't need that UX anymore? Those are questions we'll need to start asking soon, she concludes: "Right now, it's click, click, click, and the agent that says, 'Okay, I will do the clicking'. But technically, the clicking is still there. But how do you really re-imagine it, where AI becomes more dynamic and contextual based on what you do? I think there is a really important part for being able to interact back and forth with an agent because that's how we interact with a coworker. I don't think that solves everything. I think this still has to be task-specific, also very proactive. [...] So really mimicking that. If you think of the interaction, building more dynamic feeling into the app, building more contextual, building more proactiveness into the app. I think over the next year plus, we will all have to really rethink what the modern UX looks like. When we were talking about AI and content creation, I was thinking about what's next. How will tech providers embed AI in the marketing (or sales) tasks to make work easier for the marketer or salesperson? We're there now with AI agents and automation, trying to figure out how it will all work. But as much as it's helping, it's also raising many more questions, like how do agents talk together when they aren't on the same platform? How do they share the data essential for them to work cohesively? What happens to the application UI when people spend all their time with the agents? The pace of change is exciting and scary at the same time. And while companies work to figure out the technology part, they must also recognize the impact on workers. AI is changing things, and companies must help their employees understand what that means for them and take them along with them, involving them in the change every step of the way.