Principal Conversational AI Lead

Job Title

Principal Conversational AI Lead

Job Description

AI Operations owns reducing the cost of serving IG’s clients, using AI, automation and channel change as its levers, while protecting the quality of the client experience. Automating well-defined procedures with a platform partner takes us part of the way. Going further requires an agent that knows the client, one that can explain a specific fee, a specific closed position, a specific delayed withdrawal. That is an integration and architecture problem before it is a product one.

This role owns that problem. You lead the API integration roadmap that exposes account and transaction data to our conversational AI safely and accurately, you set the technical direction for the estate, and you own the framework within which automated decisions are made and evidenced. You build the function, starting with the Conversational AI manager who owns live performance and the platform partner’s day to day.

Key responsibilities

• Set technical direction for conversational AI across servicing: what we build, what we buy, where we go deeper with a platform partner, and what we do not automate at all
• Lead the API integration roadmap that makes personalised, account-aware automation possible: read paths for account, position, transaction, fee and funding data, write-back of outcomes and audit information, latency and freshness guarantees, and defined behaviour when data is stale or unavailable
• Own architecture and access design with InfoSec and Data Protection: least privilege, nothing exposed beyond what a procedure requires, and decisioning that can be reconstructed after the fact
• Own the governance framework for automated decisions — how thresholds are set and reviewed, how accuracy is evidenced, how changes are approved — robust enough for Internal Audit and light enough that changes still ship
• Secure and hold engineering capacity on roadmaps you do not control, by building the case in the receiving team’s own terms
• Lead expansion into channels not yet live, and orchestration across voice, back office and outbound
• Own the senior relationship with the platform partner: performance against commitments, delivery capacity, scope extension, and escalation of anything the AI Operations Lead cannot resolve
• Transition each new capability to live ownership at go-live, against client requirements and accuracy standards agreed with the AI Operations Lead before build starts
• Own the economics of the programme and defend them to an evidence standard Finance signs off without argument

What you’ll bring

Essential

• A genuine engineering and product hybrid: you have owned technical architecture and made product calls, and you are credible in a design review and in a board paper
• Delivered API or platform integration work in a regulated financial services environment, at a scale where access control, data protection and auditability were real constraints rather than checkboxes
• Practical understanding of how LLM-based agent systems behave in production: tool use, retrieval, evaluation, guardrails, latency, and why confidently wrong answers are the central risk
• Experience selecting and managing a strategic technology partner, including holding one to its commitments and recognising when a partnership model is quietly shifting work onto you
• Demonstrated ability to influence roadmaps you did not control, in a matrixed or decentralised organisation
• Awareness of IG’s regulatory context, including Consumer Duty outcomes and the needs of vulnerable clients, as it applies to automated decisioning
• People leadership, or clear evidence you are ready for it

Desirable

• Contact centre or client servicing transformation experience
• Trading, brokerage or another leveraged product background
• Experience on the vendor side of an enterprise AI deployment

What success looks like

By the end of year one:

• Personalised, account-aware automation is live across several use cases, with defined behaviour when data is missing or stale, and a measurable effect on resolution
• Engineering capacity for this work is committed rather than negotiated case by case, and the integration roadmap is sequenced beyond the first release
• The governance framework has survived Risk and Internal Audit scrutiny without slowing delivery to a stop
• Automation has extended into a channel that was not live at the start, and the economics of the programme can be evidenced to a standard Finance accepts

Number of openings

1