Case Studies
We work with ambitious organisations, from regulated financial services to fast-moving agencies and professional services teams.
AI Workshop
INSURANCE / EMPLOYEE BENEFITS AND INDIDUAL PROTECTION
99 agents designed by the people who own the work. A framework now scaling across EMEA
Challenge:
MetLife UK, part of one of the world's largest insurers, needed to move beyond AI experimentation and build a governed, enterprise-wide agentic AI programme. Like many regulated firms, the appetite was there - but so was the noise. Tech vendors were approaching every week, each claiming to have the next best thing, and it was hard to discern what would genuinely move the business forward. The leadership team wanted more than tools. They wanted an AI strategy grounded in their business - one that worked not just for leadership but for everyone across the organisation, supported their New Frontier growth strategy, and could stand up to FCA and Consumer Duty scrutiny. And because MetLife operates across multiple EMEA markets, each with distinct regulatory environments, whatever was built in the UK needed to be designed to scale.
What we did:
Zygens delivered an agentic AI workshop programme across MetLife UK's Employee Benefits and Individual Protection businesses, built on our Zygens Framework: the domain experts who own the workflows design the agents, not the technologists.
We started top-down, aligning leadership on a five-stage maturity path from Aware to Leading, with clear objectives and metrics tied to MetLife's growth, efficiency, and culture goals including a target of everyday AI usage in every team.
Then we went bottom-up. Frontline teams surfaced and designed agent use cases from their own work, mapped them to MetLife UK workflows, and pitched the strongest Dragons' Den-style - twelve pitches, each grounded in a real pain point, with eight agents built hands-on in Copilot during the programme itself.
Every use case was scored by the team against Value, Risk, and Complexity, producing a ranked portfolio of quick wins, strategy bets, and cases to park. Alongside it, we designed a three-tier technology governance model defining what can be built at the desk, what needs a super team, and what requires full IT development and architecture board approval - so innovation and control grow together.
Outcomes:
99 AI agent use cases designed and road-mapped across Employee Benefits and Individual Protection
23,100+ hours a year of capacity identified from the twelve agents pitched
49 quick wins prioritised, with flagship agents moved into pilot build * A three-tier operating model aligned to FCA and Consumer Duty, defining who builds what and who approves it * 15 AI Champions active across lines of business, with a replicable blueprint now rolling out across EMEA markets
“We see a lot of information - and sometimes misinformation on AI in the market. I wanted to ground an AI strategy which really worked, not just for the leadership of the business, but for everyone, and all of our stakeholders around it.
The biggest thing that surprised me about Zygens was pace. We were able to articulate a joint programme very quickly, enable it, and get quick results - and that builds confidence to go further.
What Zygens brought was a very human-centred approach and the ability to work with all layers of the organisation, while staying really grounded in our business strategy. The big transformation has been around mindset: that we can really go at pace on AI deployment, and look for both quick wins and higher-level strategic plays to transform our business.”
AI Workshop
MELTON BUILDING SOCIETY
From AI Curiosity to a Value Driving Agent Roadmap in Six Weeks
Challenge:
Like most regulated financial services firms, The Melton Building Society could see the promise of AI but wanted more than experimentation. Its leaders were already confident personal users of tools like ChatGPT and Copilot the gap was translation: turning personal curiosity into work application, safely, in a risk- and compliance-conscious environment. The Society had deliberately stepped back from adopting “anything and everything” and asked Zygens for a structured approach: identify high-value, low-risk use cases first, then work out the right technology not the other way round.
What we did:
Zygens delivered its three-workshop AI Adoption programme, built on the Domain-Driven Agent Design (DDAD) framework, in which the domain experts in the line-of-business teams who own the workflows lead the design, and every agent is weighed on value, risk and complexity before anything is built.
Workshop 1 - Foundations and discovery
Seventeen delegates built a shared language for agentic AI, then a practitioner-led exercise to reimagine workflows based on pains and gains surfaced 45 candidate use cases from their own functional areas. Self-rated knowledge rose on every measure; the sharpest gain was spotting practical use cases in day-to-day work.
Workshop 2 - Design and build
Four teams pitched candidate agents Dragons’ Den-style, each grounded in a real Melton pain point. The cohort mapped all 45 use cases across six core workflows plus cross-organisation agents, then built Reg Reader hands-on: a regulatory reading assistant grounded in a Compliance-owned knowledge base, where every answer cites its source.
Workshop 3 - Scoring and prioritisation
The cohort scored use cases into a ranked roadmap, mapped multi-agent workflows with human checkpoints, covered token economics, and finished with hands-on agent building and demos.
Outcomes:
45 use cases, sequenced into quick wins and a pipeline tier
use cases clustered most heavily in operations (17), knowledge and insight (13), and risk and compliance (8), with customer service (5), marketing and finance behind — and the densest clusters in the Customer Journey and Savings Lifecycle workflows, marking where deeper workflow redesign should focusA prioritisation model balancing value vs risk
every use case maps to one of the Society’s six strategic objectives — Risk Control (15) and Constantly Growing (10) leading
among matched delegates, use-case identification rose significantly and the NPS was +53 with no detractors
Reg Reader, built with the cohort, answering PRA and FCA questions with cited sources
“Zygens helped us take a structured, responsible approach to AI adoption. In three workshops our leadership team produced a scored roadmap of 45 agent use cases designed by the people who own the workflows, and weighed on value, risk and complexity. We’re now reimagining how we work, with governance built in from day one.”
AI Discovery Programme
FINANCIAL SERVICES / FINTECH
Building governed AI capability at pace in regulated financial services.
Challenge:
Octopus Money needed to move beyond isolated AI experimentation and build a governed, organisation-wide approach to agentic AI. The risk was not moving too fast; it was moving without coherence or control. They needed board-level confidence, internal ownership, and a clear prioritised roadmap, without compromising governance or regulatory compliance.
What we did:
Zygens delivered a programme across three structured workshops. Workshop 1 aligned the senior leadership team on the Zygens Framework, mapping AI initiatives directly to board priorities before any design began.
Workshop 2 brought cross-functional AI Champions from Sales, Marketing, Product, Data, Customer Support, Coaching, Compliance, and Finance together to design agents using our Pain-to-Gain mapping methodology.
Workshop 3 applied rigorous risk-value scoring and produced a phased implementation roadmap. Follow-on sessions covered practical agent building, evals workshops, and on-demand support tied to real delivery blockers.
Outcomes:
53 fully designed AI agent use cases, each with complete prompts, guardrails, tools, and evaluation criteria
4 priority agents ready for immediate pilot deployment
A cross-functional cohort of AI Champions across 8 business functions
A board-ready, risk-adjusted roadmap with clear recommendations on where to build, where to pause, and where to invest further
Marketing and product teams began building internal agents within weeks of the programme ending
“For any organisation considering how to approach AI adoption strategically rather than reactively, I would wholeheartedly recommend the Zygens approach.
They helped us move from generic enthusiasm about AI to a disciplined, prioritised programme with clear value, manageable risk, and genuine cultural buy-in”
AI Discovery Programme
CREDIT, COLLECTIONS & DEBT MANAGEMENT
From AI ambition to governed action in a Consumer Duty environment.
Challenge:
Perch Capital and ACI UK Ltd needed to move beyond ad-hoc AI experimentation and establish a structured, governed approach to agentic AI, one aligned to board-level OKRs, FCA Consumer Duty obligations, CONC regulations, and the operational realities of managing distressed credit portfolios at scale. Without a structured approach, the risks were fragmented experimentation, growing regulatory exposure, and an inability to demonstrate defensible AI value to the board and investment committee.
What we did:
Zygens delivered an Agentic AI Discovery Programme shaped around Perch Capital's regulatory environment, operational workflows, and strategic timelines. Workshop 1 aligned the senior leadership team on the Zygens Framework and mapped AI initiatives directly to company OKRs, including a specific collections target and cost-to-collect reduction goals, before any design work began.
Workshop 2 brought cross-functional participants from Portfolio Management, Collections Operations, Technology, Risk and Compliance, Data and Analytics, Investment, Customer Experience, and Finance together to design agents using our Pain-to-Gain mapping methodology.
Every agent canvas included explicit guardrails, escalation paths, and clear statements of what the agent must not do. A live Customer Review Agent was built and demonstrated during Workshop 2.
Every agent was then scored using our proprietary framework across Value, Risk, and Complexity dimensions to produce a phased implementation roadmap.
Outcomes:
54 fully designed AI agent use cases covering credit, collections, compliance, and customer servicing
6 priority agents identified for immediate pilot deployment
A 4-phase delivery roadmap aligned to collections targets and cost-to-collect OKRs
A cross-functional cohort equipped to lead AI delivery internally
Two further technical deep-dive workshops commissioned before the programme concluded
FCA Consumer Duty compliance embedded in every agent design from day one
“What impressed me most was how quickly our teams went from cautious curiosity to genuine ownership.
By the end of the second workshop, people across collections, compliance, and technology weren’t just understanding AI; they were designing agents that solved their own problems, in their own language.
We now have a roadmap that’s directly tied to our collections target and cost-to-collect goals, and a team that’s ready to build. ”
AI Discovery Programme
LENDING / FINTECH
From AI curiosity to a clear, compliant delivery roadmap.
Challenge:
Fleximize, an award-winning UK digital lender needed to cut through the AI noise and understand what it actually meant for their business. Lending markets were shifting quickly, competitors were moving, and the team needed clarity on where agentic AI could accelerate decisions, without introducing FCA or PRA compliance risk.
What we did:
Zygens delivered AI education and agentic AI mapping workshops, grounding the Fleximize team in the practical differences between RPA, predictive, generative, and agentic AI, and providing clear market context on why AI adoption matters now in lending.
A process mapping workshop then mapped their chosen lending process end-to-end, identifying repetitive tasks, data-intensive steps, decision points, and bottlenecks.
From there, Zygens designed a multi-agent architecture for bank-statement analysis, defining roles for ingestion, categorisation, calculation, affordability assessment, and anomaly detection, along with the necessary supporting tools.
Outcomes:
End-to-end lending process mapped with AI opportunities identified
Multi-agent architecture designed for bank-statement analysis
A prioritised roadmap built to leverage existing RPA and API integrations
High-value AI use cases identified with regulatory compliance maintained throughout
“We found the Zygens Programme eye-opening. It demystified AI for our team.
They helped us map our lending process end-to-end, pinpoint where agentic AI can make the biggest impact, and build a roadmap that leverages our existing systems while staying compliant.
We now have a clear, actionable plan to integrate AI into our operations and deliver faster, smarter decisions for our customers.”
AI Workshop
DIGITAL AGENCY
Nine people. Three hours. Thirty-two agents.
Challenge:
21 Degrees Digital is a full-service digital marketing agency based in Leeds. They had an AI policy in place. Most of the team were already using ChatGPT. Several had Claude. The CEO, Rory Mason, could code.
By any measure, this was an AI-ready team.
And yet, when Rory thought about where agents could actually change how his business operated, he kept hitting the same wall. He had the strategy. What he needed was his domain experts in the room, thinking in agents alongside him, so they could figure out what to build together.
That is why they called us.
What we did:
Nine members of the 21 Degrees team gathered in their offices for a Zygens half-day workshop. The group spanned every function in the business: PPC, SEO, content, web development, account management, and video production.
We ran the session using our Pain-to-Gain methodology. No slides of AI theory. No capability demonstrations, practical insights to get all the team on the same page. The work started immediately: where are the real friction points in your day? What is slowing down your best people? What falls through the cracks between handoffs?
Within an hour, the group had identified 32 agent use cases across nine business functions.
In three hours, they had a prioritised roadmap and a clear picture of where to start.
Outcomes:
Every participant scored the session 5/5
32 agent use cases identified across 9 business functions in under an hour
The team left with a prioritised roadmap they built themselves
“What I’m really good at is strategy. But I need the domain experts to meet me in the middle. Seeing the team push my ideas forward, thinking about what they could actually build — that means I can now book a meeting with them and say, right, what do you need this tool to do? And let’s go build.
I would absolutely recommend the Zygens workshop to everybody. No matter where you are on your AI journey - whether you have a couple of Copilot licences or you want to start integrating AI across the agency - this is the perfect workshop for you.”
