The most common question we get from enterprise leaders is some variation of "Where do we start with AI?" The honest answer depends on where you are, not where you want to be. That's why RIVER Group's service model has two distinct entry points: AI Discovery for organisations still defining their AI direction, and AI Solutions for those ready to build.
What You Need to Know
- AI Discovery and AI Solutions aren't sequential stages. They're parallel entry points based on organisational readiness. Some organisations need discovery first. Others are ready to build immediately.
- Choosing the wrong entry point wastes time and money. Discovery when you're ready to build delays value. Building when you haven't validated the problem wastes investment on the wrong thing.
- Maturity signals matter more than confidence. An executive saying "we know what we want" isn't a maturity signal. Having validated use cases, assessed data readiness, and defined success criteria are maturity signals.
- Both entry points converge on the same destination: an AI foundation that compounds value over time. The difference is the path, not the outcome.
AI Discovery: When You're Finding Your Way
AI Discovery is for organisations that know AI matters but haven't yet defined where it delivers the most value. That's not a failing. It's a sensible starting position for most enterprises.
You need AI Discovery if:
- You have ideas about where AI could help, but haven't validated them against your data, processes, and people
- Multiple stakeholders have different AI priorities and there's no clear framework for deciding between them
- You've run experiments or pilots but haven't determined which ones merit full investment
- Your data readiness hasn't been assessed for AI-specific requirements
- You don't have a clear picture of the ROI case for your AI investment
What AI Discovery produces:
A validated AI strategy with prioritised use cases, data readiness assessment, architecture recommendations, and a clear scope for the first build engagement. Not a slideshow. A decision-ready recommendation backed by evidence from your actual data and processes.
The typical AI Discovery engagement runs 2-6 weeks and involves stakeholder interviews, data assessment, technical feasibility testing, and prioritisation workshops. We've written about the sprint format in detail.
The critical thing about Discovery: it's designed to give you a clear answer, even if that answer is "don't invest in AI right now" or "invest, but not where you expected." We've run discoveries that redirected investment from one use case to another entirely, saving clients months and significant budget.
AI Solutions: When You're Ready to Build
AI Solutions is for organisations that have a validated problem, understand their data landscape, and need a team to deliver a production-grade AI capability.
You're ready for AI Solutions if:
- You have a specific, validated use case with defined success criteria
- Your data has been assessed (either through a previous discovery or internal analysis) and you understand its state
- You have organisational buy-in: budget approved, stakeholders aligned, a business owner identified
- You can articulate what "done" looks like in business terms, not just technical terms
- You have (or can assign) domain experts to work alongside the delivery team
What AI Solutions delivers:
A production-grade AI capability built on shared infrastructure, integrated into your workflows, with the operational framework to sustain it. This includes the build, the testing, the deployment, the change management, and the handover to your operations team.
How to Assess Your Readiness
The maturity signals that indicate which entry point is right:
| Signal | Discovery | Solutions |
|---|---|---|
| Use case clarity | "We think AI could help with..." | "We need to reduce claims processing time by 40%" |
| Data readiness | "We have data, but haven't assessed it for AI" | "We've assessed our data sources, quality is X, gaps are Y" |
| Stakeholder alignment | "Different teams want different things" | "We've agreed on the priority and success criteria" |
| Budget framing | "We want to explore what's possible" | "We've approved budget for a specific build" |
| Success criteria | "We want to see what AI can do" | "We'll measure success by [specific metric]" |
The Readiness Test
If you can fill in these blanks with specifics, you're ready for AI Solutions: "We want to [specific outcome] for [specific users] by [specific timeline], and we'll measure success by [specific metric]. Our data lives in [specific systems] and we've assessed its quality." If any blank draws a vague answer, start with AI Discovery.
The Common Mistakes
Skipping discovery to look decisive. Executive teams sometimes push straight to building because discovery feels like delay. It isn't. Discovery costs a fraction of a misguided build. The most expensive AI projects we've seen were the ones that skipped validation.
Staying in discovery to avoid commitment. The opposite problem. Some organisations run discovery after discovery, accumulating strategy documents without ever building anything. Discovery should produce a decision. If the answer is "go," then go.
Confusing vendor demos with readiness. Seeing an impressive AI demo from a vendor doesn't mean your organisation is ready for AI Solutions. The demo used clean data in a controlled environment. Your reality is different.
Both Paths Lead Here
Whether you start with Discovery or Solutions, the destination is the same: an AI foundation that delivers compound value. Each capability builds on the last. Each investment makes the next one cheaper. The architecture grows with you.
The entry point matters less than the direction. And the direction, for every enterprise we work with, is toward AI capability that compounds over time, not features that isolate.
The organisations I'm most proud to work with aren't necessarily the most sophisticated. Honesty about readiness is the first step to getting AI right.
Isaac Rolfe
Managing Director