// AI Product & Engineering
Embedded squads shipping AI features inside your existing platforms, AI-native products from scratch, and the automations and integrations behind them. Same product-led discipline that has defined Propel for a decade — now with AI engineering as the way we work.
0→1
// or 1→n, either way
Spinning up an LLM demo is a weekend. Shipping an AI feature your customers, regulators, and ops team can rely on is a different sport — one that needs product discovery, real engineering, evals, observability, and ways of working that don't break the moment context changes. That's our craft.
// what we ship
We're not picky about the shape — agent, feature, automation, platform — but we are picky about the standard. Every piece of work has to be valuable, commercial, responsible, and engineered.
// 01 · Net-new
From discovery through to a launched product. We validate the opportunity, design the experience, build the platform, and ship it — with the team, evals, and operating rhythm to keep it improving once it's live.
// 02 · In-platform
Ship AI inside your established product without breaking the install base. We embed alongside your team, work to your engineering standards, and lift internal AI capability while we ship.
// 03 · Operational
Internal tools that take cost out and put speed in — pricing agents, document analysis, workflow orchestration, knowledge banks. Built once, audit-able forever, with humans in the loop where they belong.
// 04 · Foundations
The unglamorous work that makes everything else possible — system integrations, data pipelines, retrieval layers, evals & observability, and the platform plumbing that lets your AI work scale.
// the propel bar
The same bar that runs across every Propel engagement. Demos are easy. Software your customers, regulators and ops team can rely on is the work.
// 01
Solves a real problem. We validate before investing in code, and we'll tell you when an idea isn't worth pursuing.
// 02
Cost, latency, ROI and unit economics tested before scale. The maths has to work — including the inference bill.
// 03
Privacy, security, IP, evals, guardrails and human-in-the-loop where it matters. Compliance baked in, not bolted on.
// 04
Production-grade. Observability, deterministic where it needs to be, and the operating rhythm to keep it that way.
// how we work
Our teams sit inside your team. They're full-stack, T-shaped, and built to thrive in ambiguity. We measure them on what changed for your business — not what we delivered.
// 01
Product managers, designers, engineers and AI architects on one squad. No throw-overs, no phase gates. Discovery and delivery happen in the same room.
// 02
Embedded with your team, on your tools, inside your engineering standards. Knowledge transfers as the work happens — not in a closeout deck.
// 03
Our people use modern AI tooling to build modern AI software. Evals, observability, retrieval, agent orchestration, and prompt-as-code aren't extras — they're how the work gets done.
// 04
Every engagement uplifts your internal team. We coach by doing — modelling best practice while actively delivering — so when we leave, your team owns it.
// 05
We validate the opportunity, the experience, and the technical approach before writing the production version. Saves million-dollar mistakes.
// 06
Rate card by default; shared-risk and outcome-based pricing where it makes sense. We back our work — and price accordingly.
// the stack we know
We're partners with AWS, Microsoft, Google and Anthropic — and we'll pick the stack on the merits, not the relationship. Below: a snapshot of the tools we use most, in production.
// proof
A snapshot of production AI work across financial services, hospitality, retail, energy, media and logistics.
// Cameron Harrison · Wealth
Specialised agents for qualitative and quantitative analysis. Real-time market events, earnings & annual report processing, surfaced into analysts' email and workflow.
// Doltone House · Hospitality
Pulls live CRM data, applies all venue, seasonality and lead-time rules, and surfaces a complete pricing recommendation. Junior staff price like seniors. Human approval, always.
// Moving Hub · Outbound contact centre
Real-time natural voice agent built on AWS Agent Core. Verifies identity, detects voicemail, qualifies leads, books appointments — all under human compliance review.
// RMS Cloud · Hospitality SaaS
Dynamic guest profile built on AWS Agent Core + Strand framework. Embedded directly into the RMS UI; smooth adoption with no workflow disruption for hospitality staff.
// Aje · Fashion
Custom prompt-generation app that turns Aje's seasonal creative briefs into Midjourney-ready prompts. Scales the application of AI imagery without breaking design intent.
// Kopa · AdTech
Multi-model PoC validated on cost / accuracy / speed, then productised. Scrapes and tags YouTube videos for ad placement; eliminates the human-in-the-loop bottleneck.
// Redcat · Hospitality tech
Two-week discovery building two PoCs — post-hoc evidence retrieval and live packing monitoring — across 51 structured test scenarios. Honest GO/NO-GO recommendation.
// AGL · Energy
Discovery, feasibility, and end-to-end build of a charger management platform. Full APIs for the consumer mobile app; foundation for AGL's decarbonisation play.
// Netwealth · Wealth platform
Rewiring how Netwealth's product and engineering teams ship AI features inside an established financial platform. Operating model + delivery happening in lockstep.
Propel have been a key partner in delivering a number of solutions to our customers. They bring deep experience and knowledge in product management that ensures we are delivering the most important features first with a customer-centric mindset. Mixed with their ability to deliver in an agile way, cloud technologies, makes them a strong full-service delivery partner.
DARREN SMITH · CHIEF PRODUCT & TECHNOLOGY OFFICER
// ways to engage
Most engagements start with one of these. We'll tell you in the first conversation which is the right shape for the problem.
// 2–4 weeks
Validate the opportunity, the experience, and the technical approach. Honest GO / NO-GO recommendation with the evidence to back it.
Ideal when: the upside is real but the path isn't proven yet.
// 8–16 weeks
Full product + engineering squad to take an AI initiative from validated PoC to first production release. Discovery embedded throughout.
Ideal when: you've validated the opportunity and need to ship.
// 6–12 months
Ongoing forward-deployed squad inside your product team. Ships AI features alongside your roadmap, lifts capability as it goes, and operates the platform once live.
Ideal when: AI is now core to your product roadmap.
// let's talk
Tell us what you're building. We'll come back within a day with a view on whether we're the right squad — and if we are, what the first 30 days look like.