14%
more productive
In a study of more than 5,100 customer-support agents, access to a generative AI assistant increased productivity by 14–15% on average. Less-experienced workers benefited most.
NBER / Stanford / MIT
AI Strategy for Entrepreneurs
Cut through AI noise.
Find the workflows where AI actually helps your business.
You don't need to automate everything. You need to know where AI can save time, improve decisions, reduce friction, or create new value — and where it should stay out of the way.
Talk through an AI opportunityAI should earn its keep.
Most businesses don't need more AI tools. They need clarity.
I help owners and operators find the few places where AI can materially improve the business, redesign the work around those opportunities, and put enough measurement and guardrails in place to know whether it's actually working.
From a single workflow to an AI-enabled product, the goal is the same: measurable value without unnecessary complexity.
When AI is aimed at the right problem
The numbers can get very real.
There isn't a universal “AI ROI.” But measured implementations show what can happen when AI is attached to a real business problem and measured against an outcome that matters.
55%
lower implementation cost
Williams reported that an AI-assisted software modernization reduced implementation cost by 55% while delivering the project 60% faster.
Williams Companies, 202630%+
more productive
McKinsey reports that a redesigned legal and compliance review workflow at Blackstone is expected to lift reviewer productivity by more than 30% and save about $5 million annually by 2027.
Blackstone / McKinsey, 202690 → 30
minutes
Walmart reported that an AI-directed workflow tool reduced shift-planning time from 90 minutes to 30 minutes for overnight stocking, with a wider rollout being piloted.
Walmart, 2025Your numbers don't need to look like theirs. The meaningful win might be ten hours returned to the team each week, faster customer response, fewer errors, better inventory decisions, or more sales capacity.
The right ROI is the one that shows up somewhere in your P&L.
When AI gets ahead of the business
AI can also make the wrong process faster.
The expensive failures tend to rhyme. Technology gets chosen before the work is understood, a demo gets mistaken for a durable workflow, or usage grows faster than anyone is measuring it.
Starbucks
11,000+ stores
Automating before the real workflow was solved
Starbucks introduced automated inventory counting across its North American stores, then retired the system after recurring accuracy problems sent employees back toward manual counting.
Lesson: A benchmark can look excellent while the real-world workflow still fails.
Rippling
80% monthly growth
AI usage without cost controls
Rippling found its AI spending growing about 80% month over month and on track to consume 40% of its R&D headcount budget, including roughly $50,000 in monthly usage attributed to one engineer.
Lesson: AI can save labor while quietly creating a completely different variable cost.
Klarna
700 agents
Cost reduction isn't the only metric
Klarna initially said its AI assistant could perform work equivalent to 700 customer-service agents. Leadership later emphasized renewed investment in human service and customer experience.
Lesson: A lower-cost workflow isn't necessarily a better business outcome.
The takeaway isn't “don't use AI.” It's “know what you're asking it to improve.”
A practical way in
Start with the work. Then decide where AI belongs.
01
Clarify the outcome
Cost, speed, capacity, quality, revenue, customer experience, or better decisions. Define success before choosing a model.
02
Map the real work
Look at the people, decisions, handoffs, systems, data, exceptions, and frustrating workarounds that shape what happens today.
03
Design the smallest useful move
Decide what AI should do, what people should keep doing, what tools are necessary, and where review and guardrails belong.
04
Measure before you scale
Track the business result, not simply AI usage. Scale what earns the next investment; fix or stop what doesn't.
You don't need an “AI department” to find useful opportunities.
Customer service
Answers, triage, summaries, routing, and agent assistance.
Sales & follow-up
Research, lead preparation, meeting follow-up, and repetitive administration.
Operations
Document processing, workflow handoffs, scheduling, and back-office work.
Data & decisions
Turn scattered data into reporting, questions, trends, and usable insight.
Knowledge
Help employees find the right policy, process, document, or answer.
Products & services
Add AI where it meaningfully improves something customers already value.
Not every one of these belongs in every business. Figuring that out is part of the work.
Sources & methodology
Figures on this page come from published studies, company reports, and established reporting linked in each example. They illustrate what happened in specific workflows; they are not forecasts or promised results for every business.
Any opportunity should be evaluated against its own baseline, operating context, costs, risks, and measurable business outcome.
Start a conversation
Start with one workflow.
Bring me the part of your business that feels slow, repetitive, messy, expensive, or hard to scale. We'll figure out whether AI belongs there — and, if it does, the smallest useful next step.
No AI strategy deck required. A messy problem is enough.
1 of 10