Jess Grogan-Avignon is a London-based AI product leader and entrepreneur who leads an agentic-AI portfolio at Accenture. She builds enterprise systems for telecommunications and customer service, concentrating on the operational, governance, and trust challenges that determine whether AI agents reach production.
From behavioral science to enterprise AI
Grogan-Avignon studied psychology at the University of St Andrews, specializing in cognitive and behavioral neuroscience, statistical analysis, and experimental design. In 2012, she co-founded Lightbox Creative St Andrews, a creative agency working across photography, video, design, and social media.
She joined KPMG in 2014, initially advising on information governance, access controls, data migrations, and technology risk before moving into technology, media, and telecommunications strategy. At Accenture, which she joined in 2019, her work expanded into customer-service transformation, cloud data programs, analysis of sales interactions, customer segmentation, and churn modeling. Her enterprise AI portfolio includes telecom network root-cause analysis and automated quality assurance for contact-center calls and chats.
Grogan-Avignon also co-founded Spotty, an aviation AI product, building its mobile interface and front end with React Native, Expo, AWS Amplify, aircraft data, and cloud-connected machine-learning services. Building the application reinforced her view that AI-generated code accelerates development but cannot replace architectural understanding. During a 2024 sabbatical, she deepened her technical grounding in Python, embeddings, semantic search, vector databases, tool calling, security, and privacy.
How she approaches enterprise agents
Governance at machine speed. Coding agents increase software output, but manual security reviews, fragmented data ownership, approval queues, and underdeveloped CI/CD pipelines can prevent working systems from reaching production. Grogan-Avignon argues that enterprises need stronger engineering automation alongside enforceable controls.
Progressive autonomy. Agents should begin in shadow mode, observing human decisions without affecting outcomes; advance into advisory roles with human approval; and earn narrowly bounded operational authority through accumulated evidence. Evaluation, explicit safeguards, and kill switches make trust an engineering requirement.
Hypothesis-driven agent delivery. Because agent behavior and useful applications emerge through experimentation, fixed scope, guaranteed performance, and premature return-on-investment projections create misleading certainty. She favors short evaluation cycles, statistical confidence, and portfolios of AI investments.
Feedback as competitive advantage. Customer interactions, edge cases, corrections, and domain-specific behavior give enterprises signals that competitors cannot readily reproduce. Her writing on enterprise AI products emphasizes workflow integration, operational monitoring, and continuous learning over isolated model comparisons.
At AI Engineer Europe 2026, Grogan-Avignon and fellow Accenture presenter Jack Wang examined how these technical and organizational constraints keep enterprise agent projects trapped between promising prototypes and reliable deployment.
