Flip the 85% failure rate: build your Compounding Engine with five simple gears

Download the deck:
Clarity Wins – Building Your Compounding Engine.

ProductCamp Cincinnati keynote recap | 1819 Innovation Hub, May 2025

Why Most AI Initiatives Fail

Eight out of ten corporate AI projects never reach production, and almost 85% never deliver positive ROI. Pilots launch, the potential is there, but customer experience stays the same, and quarterly reports show no lift. The roadblock is not data scientists or cloud credits. It is strategic clarity.

Most teams get trapped in pilot purgatory because they skip one of the five foundational gears:

1. A problem the board already loses sleep over

2. Leverage competitors cannot rent

3. An outcome so big no one can ignore it

4. An experience that hides the math and shows the win

5. A feedback loop that improves with every click

Miss any gear and the flywheel never catches. Align all five, and every user action sharpens the model, widens the moat, and turns speed into compounding growth.

Below is the framework I walked through at ProductCamp Cincinnati. Each gear builds on the last, and together they form the Compounding Engine that pulls AI out of purgatory and into profit.

Gear 1 | Strategic Problem

Begin with a mission-critical problem that is unmistakably painful for customers. Consider profit leaks that erode competitiveness, process bottlenecks that stifle growth, or experiences so flawed that they drive users to competitors. When an issue exists at that intersection and is a top priority for the board with significant customer impact, budgets are unlocked, and every team unites to solve the problem.

• John Deere See & Spray: Blanket spraying burns up to 90% of herbicide, costs growers $15–30 per acre, and breeds resistant weeds.

• JP Morgan COIN: Contract review once consumed 360,000 lawyer-hours a year. NLP now completes it in seconds, trimming millions from cost centers.

• FifthRow's Venture App: Six–month market scans now finish in less than 60 minutes, saving seven-figure consulting fees and months of runway.

A clear strategic problem gives every other gear a shared objective.

Gear 2 | Existing Leverage

Build on assets that competitors cannot rent tomorrow. Look for proprietary data, installed devices, captive workflows, or deep domain know-how that gives you a head start no open model can erase. When the solution grows out of that unique foundation, rivals struggle to copy the advantage, even if they copy the code.

• Deere already owns sprayer booms, JDLink telemetry, and the Operations Center map covering 300 million acres. Two stereo cameras convert every pass into labeled images at farm scale.

• Shopify Magic tunes GPT-4 on catalogs, clicks, and orders from 1.7 million stores, then drafts SEO-ready descriptions with one click.

• Ōura capitalizes on 20 billion hours of biometric readings; its readiness score relies on a physiology graph that no rival can rent from AWS.

If competitors can purchase the input on demand, your moat becomes fragile.

Gear 3 | 10X Outcome

Aim for an outcome so dramatic that customers and executives cannot ignore it. Think cost cuts big enough to rewrite a budget, speed gains that collapse months into minutes, or revenue lifts that reset forecasts. 10X impact changes behavior overnight and secures fast-lane approval.

• See & Spray cuts herbicide use up to 90%, so many growers pay off the upgrade in one season.

• NicheFire uncovers emerging cultural trends nine months before registering in syndicated data, cutting failed launches in half and saving brands millions in inventory costs.

• StoryTap turns three customer videos into 300 monthly and lifts conversion by 77%.

When the payoff is 10X, executives lean in and users switch habits.

Gear 4 | Frictionless Experience

Deliver the win with zero extra effort from the user. Remove new logins, training videos, and configuration steps so the algorithm disappears and the value shows up instantly, with one click at most, and none if possible. Ease converts skeptics faster than any feature list.

• Farmers drive See & Spray as usual. The monitor shows weeds in red and ticks up dollars saved.

• Shopify sellers click Generate description, skim, and publish.

• Netflix silently reshuffles rows so every title feels hand-picked.

Extra steps drain enthusiasm and shrink the 10X win.

Gear 5 | Self-Improving Loop

Design the system so every use makes the product smarter. Capture fresh signals, retrain models, and ship improvements automatically. As performance climbs, users succeed more, engagement grows, and the data flywheel spins faster, turning a single launch into a widening gap.

• Each See & Spray pass streams gigabytes of weed imagery. Overnight retraining boosts accuracy and savings, pulling more acres onto the platform.

• TikTok updates feed weights every 20 minutes on fresh engagement.

• Spotify retunes embeddings hourly from skips, likes, and replays.

No loop, no momentum, no moat. The advantage flattens fast.

Five Gears in the Wild: How John Deere Makes Them Work Together

John Deere is in the middle of a bold pivot. The 186-year-old manufacturer wants 10% of total revenue to come from software and services by 2030. See & Spray is the flagship proof point. The product bolts computer vision onto familiar green hardware and shows how each gear drives the next.

• Strategic problem: Chemical waste and rising weed resistance have drained profit.

• Existing leverage: Sprayer hardware, JDLink data, a 300 million-acre cloud map, and bolt-on cameras delivered instant labeled images.

• 10X outcome: Targeted dosing cut herbicide up to 90% and paid back investment in one season.

• Frictionless experience: Farmers drove normally while the screen highlighted weeds and displayed live dollars saved.

• Self-improving loop: Imagery retrained models nightly, updates arrived over-the-air, accuracy rose, savings grew, and subscriptions climbed.

Each weed See & Spray kills expands Deere’s data lake, accelerates the vision models that power its autonomous tillage and scouting machines, grows high-margin software revenue, and helps the company and its growers hit ESG targets. That is a compounding engine, hectare by hectare.

Ready to Escape the 85% club?

Grab the Compounding Engine canvas and stress-test your next idea:

1. Start with a strategic problem

2. Spot the unfair leverage

3. Chase 10X only

4. Show value, reduce effort, hide AI

5. Wire in the feedback

Tighten any weak gear before writing code. Then build with confidence.

[Download the Clarity Wins toolkit, black canvas on slide 25]

Speed still matters. Clarity turns speed into a flywheel.

Bring the Compounding Engine to Your Team

Ready to pull your AI pilots out of purgatory? I run fast-paced briefings and hands-on workshops that install these five gears and move projects from concept to profit. Let’s talk.

Book a Workshop →