Product Manager

AI Product Manager, IntentAI

Role Context

This role sits within the Product & Tech organisation at Intent and partners closely with Data Science,Engineering, Client Services (to design and maintain quality standards) and other Product Managers.

IntentAI is the inference layer that turns raw behavioural and device signals into the brands, topics, moments and, eventually, forward-looking predictions (move, buy, churn) that make IntentOne valuable.

IntentAI includes an Intent Semantic Layer that selects the right context for each goal and agent at scale, and is our core technical moat.

Today that moat is real but under-explained. We build strong models, but the quality and value story is still fuzzy, and the field cannot yet connect the smarts of the platform to the customer promise. This role owns that gap: setting the model north star, raising the bar on model quality and measurement, and making the intelligence legible to buyers, sellers, and the team.

Role Summary

This role is responsible for:

• Owning the product roadmap and north star for IntentAI: what our models predict, how good they need to be, and how that quality compounds into commercial value.

• Harmonising the Intent Semantic Layer across device and cloud signals, so behaviour captured

on the device and in the cloud resolves into one consistent, trustworthy view of human context.

• Setting and operationalising the quality bar for models and enrichments (coverage, precision,

recall, lift, freshness, drift, and cost at scale) so model performance is measured, visible, and

improving. This role will be heavily data- and metric-driven.

• Owning the product story of the semantic layer: explaining clearly why it is differentiated, how it makes sense of human context against a goal, and why it is hard to replicate.

• Translating model and signal capability into outputs customers trust, with explainability and

confidence built in rather than bolted on.

• Partnering with Data Science and Engineering to prioritise the model roadmap, close evaluation and monitoring gaps, and turn research into shipped, reliable product.

Core Responsibilities

1) Own the model north star and vision

• Define what “good” looks like across the model portfolio (brand and topic affinity, intent

segments, moments of need, propensity and churn), with a clear understanding of how it ladders to customer ROI.

• Maintain a prioritised model roadmap grounded in commercial impact, not research novelty, and make the hard calls on where signal and model depth create real advantage.

• Articulate a durable vision for how the intelligence layer compounds, with better signals and

scale driving better models at lower unit cost.

2) Raise and operationalise model quality

• Set the evaluation standard for inference quality, coverage, cost, and scalability, measured by

lift, precision, recall, and proxy outcomes against ground truth and in-market results.

• Work with QA to close known gaps in evaluation, drift detection, and performance monitoring so quality is continuously observed, not discovered in incidents.

• Define maturity indicators, confidence thresholds, and suppression rules so we never ship

outputs that are misleading or unexplained.

3) Explain the intent semantic layer

• Own clear, layered explanations of the semantic layer for technical and non-technical audiences, covering how it harmonises device and cloud signals, selects relevant context per goal, and keeps the platform accurate and cost-efficient at billions of events.

• Turn the moat into language Sales and Solutions can use to connect the smarts of the platform

to the customer promise, without overstating what the evidence supports.

4) Make outputs trustworthy

• Ensure outputs are grounded in model evidence, with explainable scoring and clear confidence, so enterprise buyers trust how they are generated and how their own rules are handled, removing the “black box” objection.

• Translate model needs into clear product requirements and reusable platform capability with

Data Science and Engineering, and own the tradeoffs across model richness, latency, cost, and

time-to-value.

What you’ll bring

• Deep fluency across the modelling stack, from supervised learning, propensity and uplift, to

representation and embedding models, alongside modern LLM-based systems. You understand

how predictive models are trained, evaluated, and shipped, not just prompted.

• A metrics-first instinct: fluent in precision, recall, lift, AUC, drift, and the economics of running

models at scale. You explain complex intelligence simply and hold a high bar without stalling

delivery.

• An exceptional flair for data translation and storytelling. You can turn complex models, signals,

and analytical outputs into clear, compelling narratives that make the intelligence intuitive,

• credible, and actionable for technical and non-technical audiences alike.