Kainos AI Navigator - Our Approach in Depth
Going deeper on the methodology that helps customers navigate the Workday agent landscape - from readiness through adoption and, where needed, custom build.
Welcome
The Kainos AI Navigator is our approach, not a product. It's the methodology we use to help Workday customers make sense of a fast-moving agent portfolio and turn it into a sequenced plan that delivers value.
Module 1 introduced the three legs - Guide, Adopt, Build. Module 2 set out the agent landscape those legs operate on. This module goes deeper: what actually happens in each phase, how engagements are shaped, and how we commercialise the work.
By the end you should be able to explain the methodology to a customer, describe a realistic first engagement, and know which parts of the CoE AI to pull in at each step.
Learning objectives
The methodology - Guide → Adopt → Build
The Kainos AI Navigator is three phases, run in that order. Most customers spend longest in Guide and Adopt. Build is the exception, not the rule - and always with Sana Agent Builder or a Kainos-delivered custom capability.
Guide
Readiness across people, data and process. Map the Workday agent landscape to the customer's priorities. Prioritise use cases by value and feasibility.
Adopt
Deploy GA agents with the right guardrails. Pair with change management and measurement so activation turns into real adoption.
Build
Only when no GA or roadmap agent fits. Custom capability via Sana Agent Builder or a Kainos-delivered build, governed like the rest of the estate.
The sequence matters. Jumping to Build without a Guide phase produces custom agents nobody uses. Skipping Adopt produces activation metrics with no business outcome. The methodology exists to stop both failure modes.
We say Use, Adapt, or Build - never "Use, Extend, or Build". Workday has a product called Extend; using that word muddies the positioning. Use a GA agent, adapt around it, or build a custom one with Sana.
How engagements start
Customers rarely arrive asking for "the methodology". They arrive with a question, a curiosity, or a specific agent in mind. We meet them at one of four entry points - each lands them in the Guide phase, just from a different doorway.
Executive briefing
Half-day session for a CIO, CHRO or CFO and their leadership team. Sets a shared view of the landscape and what "good" looks like.
Readiness assessment
Structured review across people, data and process. Produces a readiness report and a prioritised backlog.
Use-case workshop
Two to three day working session with business and IT to qualify candidate use cases against value and feasibility.
Specific-agent pilot
The customer has already picked an agent - often Recruiting, Self-Service, or a contract agent - and wants help deploying it well. We take them through a scoped Adopt phase while quietly doing a lightweight Guide in parallel.
Whichever door a customer comes through, the output of the first few weeks is the same: a shared view of the landscape, an honest read on readiness, and a prioritised list of agents to deploy next.
Guide phase in detail
Guide answers three questions: are you ready?, where does this portfolio fit you?, and what should you do first? It is deliberately short - weeks, not months - because the point is to get to Adopt with confidence, not to produce a strategy deck.
Readiness assessment
We assess readiness across three dimensions. Weakness in any one will sink the next phase.
People
Does the customer have the sponsorship, the change capacity, and the end-user appetite to adopt new ways of working? AI fails quietly when the people layer is weak.
Data
Is the tenant clean enough for agents to do useful work? Job architecture, security model, master data - all affect what an agent will actually do when switched on.
Process
Are the underlying business processes clear enough to automate or augment? Agents expose process ambiguity faster than any consultant will.
Governance
Is there a forum that can make AI decisions? If not, we help stand one up as part of Guide - otherwise every later decision gets re-litigated.
Landscape mapping
We take the full agent landscape from Module 2 and overlay it on the customer's priorities - the strategic themes their leadership already cares about. The output isn't a catalogue; it's a short list of agents that map to outcomes they've already committed to.
Use-case prioritisation
We prioritise candidate use cases on two axes: value (the business outcome, quantified where we can) and feasibility (readiness, data quality, dependencies). High value and high feasibility goes first. High value and low feasibility becomes a remediation workstream. Low value is parked, honestly.
A readiness report, a prioritised backlog, and a deployment plan for the next three to six agents. That's the handover into Adopt.
Quick check
Adopt phase in detail
Adopt is where GA agents actually get deployed - with the right guardrails, paired with change management, and measured against the outcomes we committed to in Guide.
Deployment
We deploy GA agents in the tenant following the same discipline we use for any Workday change: security reviewed, tested, staged into sandbox first, released with a clear back-out. Agents inherit tenant security - the guardrails from Module 2 apply - but configuration, enablement and comms still need to be done well.
Change management
Technology that nobody uses is just cost. Every Adopt workstream is paired with a change plan: personas, messaging, training, champions, feedback loops. The detail lives in Module 9 - Change management for AI; in Adopt we apply it.
Measurement
We separate activation (the agent is switched on) from adoption (people use it and it moves the outcome). Baselines are set in Guide; we measure against them from day one of Adopt. The measurement framework - what to track, how to report it, how to course-correct - lives in Module 10 - Adoption, not activation.
Build phase in detail
Build is the smallest phase for a reason. Most customers don't need custom agents in the first year - they need to use the GA portfolio well. Build is reserved for needs that are genuinely unique and large enough to justify the lifecycle of a custom capability.
When Build is the right answer
- No GA or near-term roadmap agent covers the need.
- The value case is sized and sponsored - this isn't AI for AI's sake.
- The underlying data and process are clean enough for a custom agent to succeed.
- There's an owner in the business who will live with the agent long after go-live.
How we build
Always with Sana Agent Builder or a Kainos-delivered custom capability. We don't bolt third-party AI onto the side of Workday - that undoes the guardrails we spent Guide and Adopt protecting. Sana keeps custom agents inside the same trust, security and audit posture as the rest of the tenant.
What changes in the methodology
Build adds an extra loop: design, build, test, deploy, then back into Adopt for change and measurement. The discipline is the same - the scope is just wider. Module 11 goes deeper on patterns, reference architectures, and Kainos-delivered Sana builds.
If a customer asks about Build before we've done Guide, slow the conversation down. Nine times out of ten the need they've described is already covered by a GA or roadmap agent - and starting with Build would be the more expensive, slower, riskier path.
Typical engagement shape
A first engagement is 6 to 12 weeks. Short enough to keep momentum, long enough to land a readiness report, a prioritised backlog, a deployment plan and the first visible wins.
Team composition
CoE AI Lead
Owns the methodology on the engagement. Sets the agenda, runs the prioritisation, handles exec-level conversations.
Architect
Owns the technical view - tenant, data, security, integration. Sense-checks feasibility and rules out anti-patterns early.
Change lead
Owns people and adoption. Builds the change plan alongside deployment, not after it.
Delivery
Owns the actual deployments. Brings the functional and configuration depth from the wider Workday practice.
Core deliverables
Commercial model
Our commercials match the methodology. Three things to know.
Outcome-led framing
We contract against outcomes, not activities. Not "40 days of consulting" - "readiness signed off, three agents deployed, adoption baselined". That framing forces both sides to agree on what good looks like before work starts.
Fixed-price phased SOWs
Each phase - Guide, Adopt, Build - is a discrete, fixed-price statement of work. The customer decides whether to proceed to the next phase; we don't pre-sell work we haven't earned. This keeps commercials honest and makes it easy for customers to start.
Optional managed adoption service
For customers who want ongoing support - new agents as they go GA, adoption tracking, quarterly re-prioritisation - we offer a managed adoption service. It's optional. Most customers take it after the first engagement, not before.
We don't sell T&M strategy engagements that stop short of deployment. We don't sell Build before Guide. We don't sell licences the customer hasn't sized to their outcomes. The methodology is the product.
Check your understanding
Three questions. Each explains why every answer is right or wrong - the reasoning matters more than the score.
Next steps
- Customer-facing roles: Module 4 - Positioning with customers - how to tell this story in sales conversations.
- Architects: Module 5 - Flex credits & AI architecture - the commercial and technical substrate beneath the methodology.
- Delivery & Customer Success: Module 8 - Adoption playbook - the operating detail behind the Adopt phase.
- All learners: revisit Module 2 if the landscape doesn't feel second-nature yet. The methodology only works when the portfolio is in your head.