Case Studies & Patterns
Real engagements, sanitised, reusable. This module is a living library — grows as we deliver, shared across the Practice.
Welcome
Every other module in this programme gives you theory, frameworks, and language. Module 12 is where we ground all of it in what we've actually done. Real customers, real engagements, sanitised so we can share them openly across the Practice.
This module is deliberately a living library. It will never be "finished". Each new engagement that teaches us something — good, bad, or qualifying-out — should land here in the standard template so the next person can reuse it. If you're reading this early in the programme's life, you'll see placeholders. That's honest: we'd rather ship the shape now and fill it in as we deliver than pretend we have a back catalogue we don't.
By the end, you'll know how to read a case study, find the pattern that fits your current opportunity, and contribute the next one.
Why this module has placeholders
Real case studies only land in this library once they come from a completed, sanitised Kainos engagement with Account Director sign-off. We don't invent customer detail, outcomes, or metrics to make the module look fuller than it is. What is live and reusable today is the shape — the six-part template, the four recurring patterns, and the contribution process. Use those now; replace the placeholders as engagements complete. If you've just delivered an engagement worth sharing, jump to How to contribute.
Learning objectives
How to read a case study
Every case study in this library follows the same six-part template. Same headings, same order, every time. This is deliberate — you should be able to scan a case in two minutes and know whether it's relevant to the deal in front of you.
Customer context
Industry, size, geography, Workday footprint. Sanitised — never identifying.
Trigger
What prompted the conversation — a board ask, a renewal, a specific pain.
Approach
How we applied the Kainos AI Navigator — Use, Adapt, or Build.
Agents deployed
Which GA, Early Access, or custom-built agents ended up in the solution.
Outcomes
What changed for the customer. Evidenced, not speculative.
What we'd do differently
The honest retrospective. The part that makes this reusable.
The last section is the one most worth reading. Outcomes sell; retrospectives teach. A case study without a "what we'd do differently" is marketing, not Practice knowledge.
Case study 1 — HCM at scale
Engagement story pending CoE Lead review — contribution guide below.
1. Customer context
[to be completed with real engagement] — Indicative shape: a global HCM-led customer, tens of thousands of employees, existing Workday tenant in steady state. Central HR operations team; regional HRBP network.
2. Trigger
[to be completed with real engagement] — Indicative shape: renewal conversation surfaced executive pressure to "show an AI story". HR ops buried in tier-1 case volume; backlog on policy queries from the frontline.
3. Approach
[to be completed with real engagement] — Indicative shape: Kainos AI Navigator discovery workshop mapped pain to the GA agent catalogue. Decision: Use first, Adapt where needed, Build nothing in phase 1. Sequenced rollout by population, not by agent.
4. Agents deployed
[to be completed with real engagement] — Indicative shape: Self-Service for the knowledge-worker population, Frontline for deskless workers, Recruiting for the talent team. All GA. No custom build in scope.
5. Outcomes
[to be completed with real engagement] — Indicative shape: to be evidenced — deflection rates, time-to-hire, employee satisfaction deltas. Do not populate with speculative numbers.
6. What we'd do differently
[to be completed with real engagement] — Indicative shape: we expect the honest retrospective to centre on change management starting earlier, content ownership being named explicitly, and a first-90-days adoption plan treated as a workstream not an afterthought.
Case study 2 — Finance & contracts
Engagement story pending CoE Lead review — contribution guide below.
1. Customer context
[to be completed with real engagement] — Indicative shape: finance-led Workday customer, mid-to-large commercial operation with material revenue-contract volume. CFO office sponsoring the AI conversation.
2. Trigger
[to be completed with real engagement] — Indicative shape: audit findings on contract obligation tracking, plus pressure on the legal team's review throughput. Discrete, measurable pain — the right shape for an agent conversation.
3. Approach
[to be completed with real engagement] — Indicative shape: Use, Adapt, or Build assessment concluded Use with light Adapt on process. Resisted the temptation to over-scope into supplier contracting before the revenue side was bedded in.
4. Agents deployed
[to be completed with real engagement] — Indicative shape: Revenue Contract, Contract Negotiation, Contract Intelligence. All GA. Supplier Contract noted as roadmap and deferred to a phase 2 conversation rather than forced into phase 1.
5. Outcomes
[to be completed with real engagement] — Indicative shape: to be evidenced — cycle time on contract review, obligation coverage, audit close-out. Again: no speculative numbers here.
6. What we'd do differently
[to be completed with real engagement] — Indicative shape: we expect the retrospective to focus on getting legal at the table from day one rather than day thirty, and on the value of a short "agent-in-the-loop vs. human-in-the-loop" workshop before go-live.
Case study 3 — Qualifying out
Engagement story pending CoE Lead review — contribution guide below.
1. Customer context
[to be completed with real engagement] — Indicative shape: a customer with a strong AI appetite but a messy underlying process and data foundation. Executive keen; operating reality not ready.
2. Trigger
[to be completed with real engagement] — Indicative shape: inbound request for a custom agent build against a specific business need. On the surface a good-fit Build conversation.
3. Approach
[to be completed with real engagement] — Indicative shape: Kainos AI Navigator discovery surfaced that the underlying process and data wouldn't support a reliable agent. Right answer: do not build now. Fix the foundation first, revisit in six months.
4. Agents deployed
[to be completed with real engagement] — Indicative shape: none. Deliberate outcome. We walked away from near-term Build revenue to protect the customer's long-term outcome and our credibility.
5. Outcomes
[to be completed with real engagement] — Indicative shape: to be evidenced — relationship trust maintained; phase 0 remediation engagement scoped instead; customer returned for the agent conversation once ready.
6. What we'd do differently
[to be completed with real engagement] — Indicative shape: we expect the retrospective to note the importance of naming the "not yet" answer in writing, early, so the stakeholder who championed the ask isn't left exposed.
"Use, Adapt, or Build" includes a fourth, unspoken option — don't. This case study exists to make that option visible. Practice credibility is built on the deals we declined as much as the ones we delivered.
Pattern library
Across the engagements we've run and shaped, four recurring shapes show up. If your opportunity maps to one of these, you already have a playbook — reach into the relevant case study and start from there, don't start from scratch.
Name the pattern in the customer conversation. "We see this shape a lot — here's how it tends to go." It shortens the discovery cycle and signals experience without leaking any specific customer's story.
How to contribute a new case study
If you've delivered — or qualified out of — an engagement worth sharing, it belongs in this library. The process is light, but the controls are firm. Sanitisation is not optional.
1. Draft in the template
Six headings, same order. Retrospective section is mandatory. No placeholders in the submitted draft — if something's not yet evidenced, say so.
2. Sanitise before review
Strip customer name, geography specifics, sector tell-tales, named people, and anything that could identify the account. If a reader could guess the customer, sanitise harder.
3. Sanitisation approval
Account Director for the engagement signs off that the sanitised version is safe to share Practice-wide. No Account Director sign-off, no submission.
4. Publication by the CoE AI Lead
The CoE AI Lead owns the library, does a final read for voice and consistency, and publishes. The module is the single source of truth — do not fork it into local decks.
Where the library lives and how it's versioned
The case study library lives alongside this module in the Practice enablement space. Each case study is versioned; material changes trigger a version bump and a note in the CoE AI update. If you reference a case in a customer proposal, cite the version you used so we can trace what was shared.
What NOT to put in a case study
A strong case study is confident and specific. A dangerous one is specific about the wrong things. Three hard lines.
Customer-identifying detail without consent
No names, no logos, no sector-plus-geography combinations that narrow the field to one. If the customer has explicitly agreed to be named, that's a different artefact — a referenceable case — and follows a separate approval path.
Speculative claims about outcomes
If it isn't evidenced, don't put a number on it. "Reduced cycle time materially" beats "cut cycle time by 47%" when the 47% is a modelled estimate. Credibility compounds; hype erodes.
Competitive disparagement
We don't name competitors negatively in case studies. If a customer came from another platform or another SI, describe the situation, not the party. The story is our work, not their failure.
Anything you wouldn't say in front of the customer
The safest test. If the named customer read the sanitised case study and felt misrepresented, we've failed the sanitisation, not the writing. Write as if they will read it — because eventually someone will.
Confident, direct, pragmatic — same voice as every other module. Case studies are Practice IP, not marketing copy. Leave the superlatives and the emojis out.
Check your understanding
Three questions. Each explains why every answer is right or wrong — the reasoning matters more than the score.
Next steps
- Everyone: loop back to Module 4 — Positioning with customers — and try telling one of these patterns aloud in under two minutes.
- Customer-facing: Module 3 — Kainos AI Navigator: our approach in depth. Every case study here is an application of that approach; the two modules reinforce each other.
- Delivery leads: when you close an engagement, book 30 minutes with the CoE AI Lead within two weeks of go-live to capture the retrospective while it's fresh.
- Sales & Customer Success: before the next customer conversation, pick the pattern closest to the opportunity and rehearse the one-line framing.