The Times Tech Podcast – Bonus Episode: Agentic AI Explained – The Next Phase of Artificial Intelligence
Date: February 23, 2026
Host: Katie Prescott (The Times’ Technology Business Editor)
Guest: Lilia Christoffi (Partner, PwC specializing in AI and Data, 20+ years in financial sector)
Episode Overview
This special episode dives deep into agentic AI, exploring what differentiates it from existing AI systems, its real-world applications, and the trust, safety, and governance challenges it presents to organizations. Katie Prescott facilitates an insightful conversation with PwC expert Lilia Christoffi, covering practical examples, deployment timelines, responsible implementation, and the transformative effect agentic AI could have on the workforce and business strategy.
Key Discussion Points & Insights
1. Defining Agentic AI
- Agentic AI vs. Traditional AI:
- Not just about AI "agents" responding to commands; agentic AI involves multiple specialized agents capable of collaborating and coordinating actions independently (01:38).
- “That is not agentic AI. You need multiple agents and an orchestration engine…like a manager with multiple team members, all specializing in different things to create an output.” – Lilia Christoffi (01:41)
- Bots with Initiative:
- Each agent has unique specializations and can interact with others as tools to achieve a larger intent (02:20).
- Example: In banking, a conversational agent interacts with agents specializing in product eligibility or creditworthiness (03:14).
2. Real-World Applications and Potential
- Research and Analysis:
- Agents interact with varied data sources (structured/unstructured), mimicking human deduction (04:04).
- A manager-agent validates and synthesizes outputs from numerous specialist agents (04:50).
- Industry Impact:
- Financial services and legal sectors already using AI for efficiency; the shift will drive new value creation and demand new governance structures (06:56).
- “We are needing to shift our whole economy…to create the value, the foresight and the governance—that’s required…to have a GDP impact…” – Lilia Christoffi (07:17)
- Rollout Timelines:
- Private markets moving quickly, but broad, secure adoption expected in 1–2 years (08:15).
3. Trust, Safety, and Controls
- Testing in Real Time:
- Unlike static digital architectures, agentic AI systems require continuous, real-time testing and self-healing to prevent issues like model drift and hallucinations (10:12).
- “If every so often I need to be running a test on production…to see how well the model is performing…and if it’s starting to introduce hallucination…because it will affect my customer.” – Lilia Christoffi (11:06)
- Managerial Re-think:
- Human managers will need support and new escalation frameworks to manage thousands of digital agents (11:58).
- “You’re multiplying risk functions, you’re multiplying security functions…that may have an impact…these are your control framework.” – Lilia Christoffi (12:03).
4. Democratization – or Not Yet?
- Barriers for Small Businesses:
- The cost of building agentic AI systems is high. Until large platforms modularize these services, it will be mainly larger organizations deploying them (13:36).
- “The build cost for the current technology is significant… a lot of it is not a buy, it’s a build decision.” – Lilia Christoffi (13:39)
5. Responsible AI – Real Examples
- Pensions Case Study:
- PwC’s work with a major UK pension provider: agentic AI is rolled out only with trustee board approval, strict responsible AI frameworks, and dedicated AI councils ensuring continued oversight (16:06–17:48).
- “We would never get into an AI transformation…unless we had the trustee board feel that there is certain safety measures in place…” – Lilia Christoffi (16:33)
- Use of “control agents” and real-time dashboards for monitoring and governance (17:03–18:38).
6. Workforce & Organizational Impact
- Jobs & Roles Transformation:
- Risk, HR, and operational roles will need to adapt: from managing human workers to both humans and autonomous agents (19:02).
- “If we’re going to have non-humans, how are you going to performance test those non-humans against humans?… If you don’t build the right measures…you may end up paying more in tech than you would in a human.” – Lilia Christoffi (19:47)
- Significant upskilling required; traditional data roles—like data analysts—will evolve as agents take over lineage and tracking tasks (20:39).
7. Culture and Organizational Mindset
- Inspiration Series:
- Accelerated pace of change—traditional strategies (3–5 years) are outdated, AI roadmaps now updated quarterly or faster (22:06).
- New workforce entrants are optimistic and ready for value-added, tech-supported jobs (22:35).
8. Personal Take: Excitement vs. Fear
- Bias, Fairness, and Opportunity:
- Lilia is optimistic about agentic AI’s societal potential if designed fairly and transparently—control over tech is possible with the right investment, unlike controlling human bias (24:25).
- “If we build correctly, it can create a lot of grounding in terms of fairness and transparency in society, which I think is much, much needed.” – Lilia Christoffi (24:27)
Notable Quotes & Memorable Moments
-
On the real meaning of agentic AI:
“You need multiple agents and an orchestration engine… it’s like a manager with multiple team members, all specializing in different things to create an output.” (01:41, Lilia Christoffi) -
On new business opportunities:
“We are needing to shift our whole economy… to have a GDP impact in our country. And if we do that, we are talking about billions worth of substantial GDP increase.” (07:17, Lilia Christoffi) -
On real-time AI management:
“Your whole testing infrastructure has to change… you have to build testing that is real time.” (10:12, Lilia Christoffi) -
On barriers for SMEs:
“The build cost for the current technology is significant… Until the large tech platform providers actually modularize… it’s not a buy, it’s a build decision.” (13:39, Lilia Christoffi) -
On responsible transformation:
“Every AI transformation really needs to have buy-in from the bottom… but it needs to also be run from the top with very bold ambitions.” (17:03, Lilia Christoffi) -
On workforce evolution:
“We talked about the legal teams obviously using a lot of AI… the change in face of legal, tax and financial functional units are going to change significantly…” (20:15, Lilia Christoffi) -
On optimism for the future:
“I am very excited by it because I think if we build correctly, it can create a lot of grounding in terms of fairness and transparency in society…” (24:26, Lilia Christoffi)
Key Timestamps
- 01:38 — What is agentic AI?
- 04:04 — Real-world examples in research/analysis
- 06:56 — Current organizational use and economic impact
- 08:15 — Timeline for broader rollout
- 10:12 — Shift in testing and controls for agentic AI
- 12:03 — Rethinking managerial roles and escalation frameworks
- 13:36 — Challenges for SMEs and the cost of agentic AI
- 16:06 — Case study: Responsible implementation in pensions
- 19:02 — Impact on jobs, HR, and risk management
- 22:06 — Organizational culture and adoption pace
- 24:26 — Hopes and concerns about agentic AI’s future
Summary
This episode provides a nuanced, expert-led exploration of agentic AI, demystifying its implications for business and society. Lilia Christoffi stresses the transformative potential and the governance rigor required for trust and safety. While the technology promises to revolutionize decision-making and business models, it also demands cultural, operational, and structural adaptation at pace. Adoption is accelerating, but responsibly harnessing agentic AI will require both top-down leadership and bottom-up innovation, especially as organizations race to avoid falling behind in the next AI revolution.