On the latest episode of the HCLTech’s Trends and Insights podcast, we explored a fresh chapter for robotic process automation (RPA).
Covering the insights from a recently released whitepaper, The evolving role of RPA in the autonomous enterprise, authored by Rohit Kumar, AVP, Digital Business at HCLTech, this episode highlights that when RPA is paired with Agentic AI and Generative AI, it becomes the reliable, observable and adaptable execution layer of the autonomous enterprise.
Why traditional RPA struggled to deliver value at scale
Traditional RPA excelled at mimicking clicks and keystrokes but often stalled when rolled out at scale because production systems don’t stay still. As Kumar explains, classic programmes “tried to handle both the execution and the cognition, and that’s where it began to falter.” Bots are superb at deterministic actions, such as copy, paste, click and send, but fragile when asked to decide who to invite, approve or escalate. “The real strength of RPA is automating repetitive tasks…but not necessarily in making decisions,” he says.
Real-world variability compounded the problem. Processes rarely look the same across regions or channels: a complaints workflow might arrive via phone, email or social media, for example. Each deviation created exceptions and maintenance overhead. “Studies have shown…50% of RPA budgets were spent not on innovation, but simply on maintenance,” notes Kumar. Organizations were surprised to find much of their spend diverted to keeping scripts alive rather than extending them to new value.
Common failure patterns:
- High variability across geographies and channels increases breakage
- UI changes and new input formats trigger costly exceptions
- Designs that mix rules with judgment create brittle hybrids
In short, traditional RPA thrived on repetition but lacked the adaptive intelligence to handle edge cases. It was, in Kumar’s words, “waiting for its perfect match;” a reasoning layer capable of context and adaptation.
Reinventing RPA with agentic intelligence
That perfect match has arrived in the form of Agentic AI. The winning pattern, says Kumar, is to separate thinking from doing: let “the brain shift to the agent.” Crucially, this isn’t a rip-and-replace strategy. Existing bots should be “reimagined as tools” that intelligent agents call when needed.
In practice, enterprises expose RPA capabilities — the email bot, the copy-and-paste action, the ERP connector — through interfaces that an AI agent can reason over. A user asks, “Summarize my reports,” and the agent decides which tools to invoke, in what order and with what parameters. “It’s not RPA or agentic,” emphasises Kumar, “it is really about the combination of the two.” RPA returns to its sweet spot, which is fast, deterministic execution, while the agent supplies judgment, planning and error recovery. The result is a system that scales because the cognitive load moves out of fragile scripts and into an adaptive reasoning layer.
New opportunities for customer and employee experience
With a reasoning layer in place, two value pools open rapidly.
1. Customer experience: Speed and responsiveness jump. “Processes that used to take days or hours can be completed in minutes with AI in the loop,” says Kumar. He cites loan processing moving “from anywhere from 12 hours to 30 minutes,” a dramatic reduction that translates into higher satisfaction and conversion. In insurance, he points to “80% automation of first notice of loss claims,” where agents extract signals from photos, emails and forms to triage faster and cut overhead. Consistency improves too: codified standard operating procedures (SOP) mean an agent can follow the same steps every time, whether issuing a refund or handling suspected fraud.
2. Employee experience: Agentic RPA becomes a practical copilot that removes routine tasks, such as data entry, form filling and moving files, which frees employees to “focus more on what matters,” from empathy to complex problem-solving. Kumar recalls building a legal copilot that answered natural language queries from a regulatory body, allowing field teams to “reduce document drafting time by 90%.” The impact goes beyond productivity and improves engagement and retention as roles shift toward higher-value work.
What leaders should ask before investing
“Set realistic expectations,” advises Kumar, because early over-promising erodes sponsorship when ROI takes longer than hoped. Bring business stakeholders in from day one: without their process knowledge “you’re guaranteed to ensure failure.” He also cautions against automating waste. If candidate shortlisting is broken, scaling interviewing capacity with bots merely accelerates a sub-optimal process.
Five due-diligence questions:
- What exact business outcome are we targeting, and how will we measure it?
- Are we automating a broken process that should be redesigned first?
- How will we maintain oversight, including observability, audit trails and alerts, once agents act?
- Where must we keep a human-in-the-loop or require approvals for sensitive actions?
- What is our plan for ongoing training, model updates and governance of bias and risk?
Operational safeguards matter as agents gain powerful tools. Anything “super critical like delete or send an email” should demand higher accuracy, escalation paths or human sign-off. Ethical and regulatory risks must also be anticipated, especially in high-stakes decisions that accept or reject people. And because foundation models evolve, teams must plan for continuous validation: “You cannot assume that...I don’t need to do any verification.”
Measuring ROI with intent
To avoid vague outcomes, Kumar recommends a three-layer approach to value:
- Operational ROI: Tangible efficiency, including time saved, throughput, cycle-time reduction and direct cost impacts. For instance, person-hours eliminated per month and minutes saved per ticket
- Experiential ROI: Stakeholder outcomes, including customer satisfaction, NPS, retention and employee engagement. Automating tedious work can reduce burnout and attrition
- Strategic ROI: Adaptability and innovation capacity, including speed to launch new products, ability to respond to new fraud patterns or regulatory changes and talent retention as roles upskill
Baselining is vital. “Down the lane when people start asking where the ROI is…the baseline doesn’t really exist,” he notes. Teams should capture current performance before automation, use A/B tests where possible and agree on timelines for results. Some efficiencies arrive immediately and others, like strategic gains, compound over months as models, prompts and policies improve.
The adaptability dividend
Looking forward, agentic RPA directly increases organizational resilience. In a world of rapid change, systems that require hard coding will always lag. By contrast, “agentic RPA systems can adapt to new inputs, new situations without explicit reprogramming.” Standard operating procedures can be provided as instructions rather than translated into brittle scripts, so adding a clause does not require a multi-week engineering cycle.
As underlying models keep improving, organizations that ride this curve gain capability without rewriting everything. Agents can reason over multimodal inputs, combine tools in novel ways and escalate appropriately. “Often it is a matter of updating a model or swapping modules,” says Kumar. The result is operations that are robust against disruption and a culture that treats adaptability not as a threat but as a competitive advantage.
A future of reusable tools in intelligent workflows
Traditional RPA remains the best way to execute precise tasks across legacy systems, but it needed a thinking partner. Agentic AI supplies that partner, elevating bots into reusable tools inside intelligent workflows. Instead of a wholesale replacement, the path forward is a disciplined combination of discovery, guardrails and measurement. Organizations should repurpose what they have and put agents in charge of orchestration. In this reality, autonomous capabilities stop being an unachievable ambition and start becoming muscle memory that enables adaptable processes, resilient customer journeys and empowered employees.
