✕

Everyone Has an AI Adviser, But When Do You Still Need an Expert?

By AI 6 min read

Recently, we worked through a software decision the way it should be done. We interviewed users, mapped their processes, gathered requirements, and built a business case with a well-researched vendor shortlist.

Then, in the final meeting, someone jumped in with: "I asked ChatGPT, and it suggested some different product."

It's a fair challenge. AI makes advice radically accessible, but a quick query and a structured analysis are not equivalent.

The AI Capability Illusion

Asking "What CRM should we buy?" seems simple, but the right answer hinges on cross-team workflows, data integration, privacy controls, and leadership metrics. AI gives you an answer, but its quality depends entirely on your ability to define the actual problem, provide the right context, and recognise when the output is wrong. This is where the danger arises and where a good professional adds value because the real risk isn't that AI fails; it is that a completely wrong answer still sounds incredibly polished and persuasive.

In a field experiment with 758 Boston Consulting Group consultants, AI drastically improved speed and quality on tasks within its capabilities. However, when given a complex business problem where the key details were buried in interview notes, AI users were actually 19 percentage points less likely to reach the correct solution.

When to Rely on AI vs. Expertise

Before committing to an AI-generated strategy, evaluate your decision against the 4 Cs to find the right balance between AI and expertise:

  • Clarity: Do you know why this problem is happening, or just that it's happening?
  • Consequence: If the advice is wrong, how much disruption could it cause?
  • Complexity: Does the decision affect one task or several teams, systems and data?
  • Confirmation: Have you checked this against someone with real experience, rather than just asking AI if it's correct?

Map your answers to this decision matrix to determine the best path forward:

The Best Path Forward Your Decision Profile (The 4 Cs) Typical Examples
Do-It-Yourself with AI
  • Clear problem
  • Low consequence
  • Single task
  • Easily confirmed internally
  • First drafts
  • Meeting prep
  • Initial option lists
  • Small trials
AI Draft + Expert Review
  • Defined goal
  • Moderate financial impact
  • Contained system updates
  • Requires a specialist to catch hidden risks
Expert-Driven (Powered by AI)
  • Ambiguous problem
  • High consequence
  • Cross-functional complexity
  • Hard to confirm internally

The Better Question Before You Commit

The choice is rarely between AI and an expert; it is about how much professional judgement your decision demands.

Use an AI chat for early exploration or easy low-risk queries, and bring in expertise when a decision is complex, costly, or difficult to reverse.

Before committing, ask yourself: is AI helping you make a better decision, or just giving you confidence in something you don't fully understand?

How Rethink Digital Operates

At Rethink Digital we combine AI with our enterprise SaaS product management and solution architecture experience to navigate the right balance between speed and strategy. This applies across the full technology lifecycle.

Discover: We start with the people and the problem, not the product. We speak with users, map processes, and define requirements, using AI to accelerate research while we work out what's really going on.

Design: With the problem understood, we design the solution that genuinely fits your business. We compare the options independently, with no ties to any vendor, and set out a clear roadmap, using AI to test ideas quickly while we make the judgement calls.

Build: Once the direction is clear, we turn it into a working solution. We manage the configuration, integration, data migration, testing, and training, using AI to streamline documentation and custom code.

Scale: The work doesn't stop at launch. We help teams adopt the solution, manage the platform, and provide ongoing advice. AI helps us work through data faster and prepare clearer recommendations, while we dictate which changes create lasting value.

Whether AI's answer is right depends on the question you asked, the details you gave it, and whether someone can catch it if it's wrong. That's exactly where we come in.

If AI has already given you a recommendation, don't act on it yet. Bring it to us. We'll stress-test it, strengthen what holds up, challenge what doesn't, and help you move forward with confidence.

Try it: AI or an expert?

Where does your decision sit?

A guide, not a substitute for judgement.