Big Four firm PwC announced new agentic AI capacities, including a model that proactively identifies areas of value leakage and acts inside the tools teams already use to fix them itself.
The new solution,
The software, which is supported by PwC’s recently-released
The system connects directly into ERP environments, continuously monitors key metrics, and acts inside the tools teams already use. For example, a supply chain agent might detect rising shipping costs and automatically reroute deliveries to reduce spend. Finance agents can spot and correct billing errors before they reach the customer. Clients typically see measurable efficiency gains in the first quarter, with continued improvements over time as the system learns and adapts.
“Too many transformations still rely on one-off pilots and stale data, stretching the gap from insight to impact and suffocating ROI,” said Saurabh Sarbaliya, PwC’s principal for enterprise strategy and value. “Agent Powered Performance flips the economics by distilling PwC’s industry transformation playbooks into AI agents that turn static insights into compounding gains, without rebooting each time.”
Agent Powered Performance is platform-agnostic and built on an open architecture so it can work across different LLMs based on client preference and task-specific needs. It works on major enterprise platforms including Oracle, SAP, Workday, and Guidewire.
Agent OS Model Context Protocol
PwC also
By integrating this standard, agent systems registered as MCP servers can be used by any authorized AI agent. This reduces redundant integration work and the overhead of writing custom logic for each new use case. By standardizing how agents invoke tools and handle responses, MCP also simplifies the interface between agents and enterprise systems, which will serve to reduce development time, lower testing complexity, and cut deployment risk. Finally, any interaction between an agent and an MCP server is authenticated, authorized, and logged, and access policies are enforced at the protocol level, which means that compliance and control are native to the system—not layered on after the fact.
This means that agents are no longer siloed. Instead, they can operate as part of a coordinated, governed system that can grow as needs evolve, as MCP support provides the interface to external tools and systems. This enables organizations to move beyond isolated pilots toward integrated systems where agents don’t just reason, but act inside real business workflows. It marks a shift from experimentation to adoption, from isolated tools to scalable, governed intelligence.
Research Composer
Finally, a PwC spokesperson said the firm has also launched a new internal tool for its professionals called Research Composer, a patent pending AI research agent embedded in the firm’s ChatPwC suite, designed to accelerate insight generation by combining web data with PwC-uploaded content.
Professionals will use the Research Composer to produce in-depth, citation-backed reports for either the firm and its clients. The solution is intended to enhance the quality of client work by equipping teams with research and strategic analysis capabilities.
The AI agent prompts users through a step-by-step research workflow, allowing them to shape how reports are packaged—tailoring the output to meet strategic needs. For example, a manager in Advisory services might use Research Composer to evaluate white space opportunities across industries or geographies, drawing from internal reports and up-to-date market data.
Credit: Source link
