Chatbots have long symbolized digital transformation—polite text boxes on corporate websites and service portals promising smarter, cheaper support. In recent years, generative AI (GenAI) has joined these tools, making conversations feel more natural, but they remain merely automated answer engines. As we approach the end of 2025, traditional chatbots are starting to feel like relics of a bygone era.
A wave of new agent-based AI is emerging: these systems can not only converse but also reason, plan, and act within business workflows. These agents are not just talking assistants; they are digital colleagues that think.
Across industries, companies are redesigning operations to leverage this new capability. They are discovering that agent-based AI is not just an upgrade to chatbots—it is a redefinition of how digital work is done.
1. From Passive Bots to Proactive Partners
Jesse Flores, founder and CEO of web development company SuperWebPros, has witnessed this transformation unfold. He states, “Traditional chatbots are essentially decision trees—if keyword X, then respond Y.” They perform well in FAQs and appointment scheduling, but their world is limited to scripts.
Even when connected to large language models like GPT-5, most chatbots still lack a deep understanding of company data or business context. “They are language-driven responders,” Flores explains. “They can talk, but they don’t think or act.”
According to Flores, intelligent AI changes this dynamic. Each agent has a name, a task defined by system prompts, and a connection to company data enhanced by retrieval-augmented generation. Many of them also utilize tools like CRM, databases, or workflow platforms. “Being an agent is like hiring a new employee who is familiar with your system from day one,” Flores says. “It’s not just about responding—it’s about executing.”
This new collaborative model is also changing how employees interact with technology. Flores notes that his clients often name their agents, treating them as teammates rather than tools. “When the marketing department needs to verify something, they say, ‘Let’s ask Marco,'” he adds. “This naming convention makes adoption easier—it feels more human.”
Moody’s, known for its information services, is gaining further insights. Cristina Pieretti, head of digital content and innovation at Moody’s, states that agent-based AI is changing the nature of how companies can serve their clients.
“With chatbots, you are just chatting about a topic. Agents can actually perform tasks that humans would typically do,” she says.
Pieretti mentions that Moody’s has begun developing agents to directly automate parts of client work—such as generating credit memos and financial analyses. Users no longer need to retrieve one data point at a time; they can configure agents to pull information from multiple systems, assemble the correct parts, and deliver a completed report with just a click of a button.
“This shifts us from being providers of insights to partners in workflows,” she says. The result is that AI not only informs decisions but also helps execute them.
2. Laying the Foundation: Governance and Trust
At IBM, Chief Information Officer Matt Lyteson is applying the same principles globally. His team is embedding agent-based AI into every aspect of the company’s operations—human resources, IT support, procurement, and sales—serving 280,000 employees worldwide.
“Chatbots handle tasks through rigid, step-by-step processes that are prone to failure,” Lyteson says. “Intelligent AI changes this, enabling systems to dynamically reason through processes. This is the direction of future work.”
One of IBM’s earliest success stories is password resets—though less glamorous, it is ubiquitous. Now, two agents collaborate: one to triage requests and another to validate credentials and execute resets, all under the company’s identity and access management system. Each agent has its own digital identity, ensuring audit trails and preventing impersonation. “This is a great example of multi-agent collaboration grounded in enterprise security,” Lyteson says.
These principles now underpin IBM’s broader enterprise AI platform, built on watsonx Orchestrate. The company’s AskHR, AskProcurement, AskSales, and AskIBM systems rely on small specialized agents operating within a unified governance framework. Every IBM employee interacts with these agents daily, likely making it one of the largest agent AI deployments globally.
The returns are impressive. According to Lyteson, IBM’s AskIT system can now resolve 82% of support requests without human intervention, allowing IT personnel to focus on complex issues and enabling IBM to shut down its IT service desk phone lines. “We are now focused on trust and collaboration—humans confidently collaborating with multiple agents,” he says.
3. Responsible Intelligence and the Next Phase of AI
Murali Swaminathan, Chief Technology Officer at IT service company Freshworks, believes that this new era of agent-based AI must prioritize responsibility alongside innovation. He describes the evolution of AI in three phases: traditional chatbots, which are scripted and fragile; agent-assisted systems that index knowledge for humans; and now, proactive AI that can understand context and act accordingly. “It’s like moving from guided driving to fully autonomous driving,” he says.
The company’s Freddy AI platform, launched in 2018, has evolved from chat support to a system that automates end-to-end workflows. For example, in human resources, employees can request leave, and customer service representatives determine which HR system to query, check policies and balances, and then execute the request. “It’s about reasoning and action, not just retrieval,” he adds, noting that clients like the Fraser Group in the UK have shifted about a quarter of their support cases through these agent-based workflows.
Swaminathan emphasizes that responsible AI is not a marketing promise but a technical discipline. He states that Freshworks’ Freddy Trust Framework integrates fairness, transparency, and privacy into every agent workflow.
This framework includes profanity and content filtering, automatic masking of personally identifiable information, and rules preventing customer data from being stored beyond the current session. Clients can also add their own safeguards. “Every deployment is designed to protect user data by default,” he says.
Freshworks has also launched the Freddy Agentic AI Studio, a no-code development environment where businesses can safely build and deploy agents. Templates, pre-configured prompts, and embedded filters make experimentation simple and controlled. “We serve everyone from small businesses to large enterprises,” Swaminathan says. “Simplicity and control must coexist.”
He refers to this philosophy as secure empowerment—democratizing AI while maintaining trust. “Our goal is to help organizations adopt AI quickly and confidently—every step has safeguards, clarity, and simplicity,” he says.
4. Moving from Chatbots to Intelligent AI: A Practical Roadmap
Agent-based AI is not a software upgrade—it’s a redesign of digital work. Every leader interviewed for this report emphasized that success relies not only on culture and experimentation but also on data and governance. Before moving beyond chatbots, IT leaders should not only ask, “Can we do this?” but also, “Where should we start—how can we do this safely?”
Start small—choose the right problems.
Flores from SuperWebPros suggests starting with what he calls a “4 out of 10 pain point”—a mildly frustrating but not fatal issue for the business. “What you want is a PR win, not a huge risk.” A 90-day pilot project should aim to quickly and significantly demonstrate value. Early successes build momentum and cultivate internal advocates.
Pieretti from Moody’s agrees: start with repeatable, well-defined workflows that can deliver measurable value. “Don’t try to boil the ocean,” she advises. “Use generative AI to keep processes consistent, and automation can significantly increase business impact.”
Build on a solid foundation of data and governance.
Lyteson from IBM warns against “AI sprawl”—dozens of poorly coordinated pilots accessing sensitive data. He says, “Start with an enterprise AI platform that enforces identity, access, and auditability from day one.” IBM’s enterprise AI platform gives each customer service representative a unique digital identity that mirrors employee permissions and ensures accountability.
Swaminathan from Freshworks applies similar principles through the Freddy Trust Framework—integrating fairness, privacy, and transparency into every agent workflow. “With great power comes great responsibility,” he says. “Safeguards are not optional; they are a design principle.”
Shape culture, not just code.
Flores points out that human adoption is often harder than technical integration. “People resist change,” he says. “We name agents—Marco, Betty, Harry—to make them feel like teammates rather than threats.”
Pieretti at Moody’s has experienced similar challenges. “The key is to shift the mindset from ‘AI will replace me’ to ‘AI will empower me,'” she says. Training, communication, and co-creation help employees feel they are part of the change rather than victims of it.
Adopt iterative, controlled rollout strategies.
Lyteson and Swaminathan advocate for continuous monitoring and version control—agent 1.0, 1.1, 1.2—each version tested for drift, bias, and reliability. Moody’s Pieretti’s team conducts adversarial “jailbreak” testing before and after deployment to ensure agents perform safely under pressure.
Swaminathan suggests measuring success with hard metrics like deflection rates, resolution times, and user satisfaction. “There’s no plug-and-play AI,” he says. “Start small, measure outcomes, and scale confidently.”
Ask the right preparatory questions.
Before committing to adopting agent-based AI, IT leaders should assess four fundamental elements:
Strategy: Have we identified application scenarios where automation can deliver measurable results?
Data and Integration: Do our systems have comprehensive documentation and secure API or metadata access?
Governance: Do we have clear protections for identity, permissions, and audit trails?
Culture: Do we have internal advocates to model productive, responsible use?
Across these four organizations—SuperWebPros, Moody’s, IBM, and Freshworks—one insight stands out: agent-based AI thrives at the intersection of governance and imagination. Chatbots respond; agents reason and act. But they can only do so safely in an environment built on trust, transparency, and collaboration. Those IT leaders who invest early in these foundations will be the ones transforming AI from a talking tool into a true digital colleague.
If you are interested in AI, we recommend joining a community:
This article is sourced from computerworld.com and translated by Digital Transformation Network. Please note the source when reprinting.
ElectronicCDO、BYDCDO、TCLCDO、HisenseCDO、Chao Wei PowerCDO、HailanCDO、Wuxi Industrial DevelopmentCDO、Jin YuCDO、Jinxi SteelCDO、Heavy TruckCDO、Shandong Dongming PetrochemicalCDO、YagorCDO、NanshanCDO、GoldCDO、Jiangyin Chengxing IndustrialCDO、Yibin WuliangyeCDO、HengtongCDO、Hangzhou SteelCDO、Xinhua UnitedCDO、Jiuquan SteelCDO、GCLCDO、Guangxi Liuzhou SteelCDO、Liaoning Fangda IndustryCDO、Rizhao Steel HoldingCDO、Xinhua United Metallurgical HoldingsCDO、Great Wall MotorsCDO、Wanda HoldingsCDO、Jiangling MotorsCDO、ChuanhuaCDO、Jintian Investment HoldingsCDO、YuedaCDO、Li HuayiCDO、ZTECDO、Yangtze RiverCDO、Inner Mongolia Yili IndustrialCDO、Guizhou MoutaiCDO、ZhengbangCDO、Xuzhou Engineering MachineryCDO、Baotou SteelCDO、SanyCDO、Chenming HoldingsCDO、International Maritime Container(}CDO、Hangzhou JinjiangCDO、TongweiCDO、ZhengtaiCDO