Which process should we point AI at first?

95% of organizations report no measurable return on GenAI investments, as implementation bottlenecks shift to the cost per accepted change. Leaders must prioritize low-exposure, high-value domains where returns scale faster than costs.

2026-07-20 · Counsel verdict · AIssential

The question

We have budget for one serious AI initiative and several candidate processes. How should we choose which to attack first — by cost, by volume, by how much expert judgment is trapped in a few people, or by how measurably the outcome can be proven — and what should disqualify a process from being the first candidate?

Counsel's position

Prioritize a high-volume, expert-judgment-intensive process with clear, measurable outcomes and low regulatory exposure to demonstrate value.

Verdict

The verdict: Prioritize a high-volume, expert-judgment-intensive process with clear, measurable outcomes and low regulatory exposure to demonstrate value.

Initial AI initiatives succeed in low-exposure, high-value domains

Given your need to select a first candidate process, prioritize areas with lower regulatory risk where AI can augment rather than replace human judgment.

• Funding Agentic AI in HR Without Losing Control - with Carey Smith of Blue Cross and Blue Shield

AI implementation bottlenecks have shifted to cost per accepted change

When evaluating which process to attack first, prioritize those where you can clearly measure the cost per validated, accepted outcome.

• The New Bottleneck in AI-Assisted Engineering Is Token Economics

Viable AI workloads require returns that scale faster than costs

To select your first initiative, disqualify any process where usage, complexity, or poor data quality will cause costs to outpace the generated value.

• How Do We Fix This?

Enterprise GenAI pilots fail to show measurable returns

Disqualify candidate processes that lack a single accountable business owner or cannot demonstrate measurable P&L impact.

• The $30B GenAI Divide Is Coming for Soft Approval Memos

Weighted scorecards effectively filter AI use cases by time-to-value

Evaluate your candidate processes using a structured scorecard that balances projected cost savings, technical feasibility, and time to value.

• Best practice advice when building future-ready enterprise AI

Read another verdict

• Buy a tool for this process, or build around our own knowledge?
• Our documents are a mess. Clean them up before AI, or after?
• How do we measure the return on an AI workflow — and what baseline is honest?
• Our best people's know-how isn't written down — can AI even use it?
• Automate this workflow, or redesign it before we automate?
• Our AI pilot works but nobody uses it — fix the workflow or kill it?
• Rent AI from a vendor, or run your own?
• Let non-developers ship AI-generated code?

Get Counsel for your own decisions →