Consulting

By day, they call me a consultant. By night, when the city’s dashboards blink like constellations and warehouses hum like beehives, I answer to a different name: the person companies call when their data starts telling ghost stories and their plans need a cape. I don’t build websites. I build momentum—powered by AI that minds its manners, logistics that behave like choreography, and strategies that survive contact with Monday morning.

The first thing you should know: every transformation starts with a map. Not the kind drawn in marker on a whiteboard that’s already crying, but a living atlas of how value moves—where signals are born, where decisions get stuck, and where a single hour hides a week’s worth of waste. I wear a utility belt filled with questions that pry at bottlenecks until they squeak: What’s the customer trying to do? What’s the fastest honest way to help them do it? Where does our machine learning sharpen the pencil instead of chewing erasers?

I travel with a sidekick—an unassuming model named Sprocket, trained to take notes like a court reporter and forget secrets like a vault. Sprocket does not predict the future; Sprocket politely narrows the universe of maybe, so humans can be brilliant on purpose. We never begin with the fanciest architecture; we begin with the smallest useful truth, measured in hours saved, miles not driven, and headaches that forgot to exist.

On one adventure, a fleet of delivery vans had the tragic habit of arriving just after customers left for work, leaving packages to sunbathe on porches or boomerang back to the depot. The map said routes were ‘optimized.’ The drivers said otherwise. I listened to the drivers, because heroes listen. We fed Sprocket three simple ingredients: historic stops, micro‑weather, and building quirks whispered by people who know where the sneaky elevators live. The result? Not magic. Better mornings. ETAs tightened. Routes flexed around school zones and surprise parades. The drivers took back ten minutes per route and used them to be human: a wave here, a quick cart rescue there. Ten minutes multiplied across a fleet is a portal to a better quarter.

Sometimes the villain is a well‑meaning spreadsheet that grew up to be a labyrinth. We met one masquerading as demand planning. It was bold, colorful, and catastrophically wrong on holidays. We didn’t replace the planner with an algorithm; we drafted a truce. The planner taught the model what the data forgot: that the first warm weekend in April flips a switch in people’s hearts, and that lemonade mix outsells logic whenever the baseball team wins two in a row. In return, the model watched a thousand stores at once without blinking. Together they became a duo: human hunch and machine memory keeping promises to shelves.

Strategy, if it wants to wear a cape, must consent to chores. We anchor big talk to small calendars. Quarterly OKRs become weekly experiments: a pilot here, a guardrail there, telemetry everywhere. I bring a portable command center—two dashboards and a ruthless checklist. Metrics are the sidekicks that never get top billing but always save the day: time‑to‑decision, percent of manual handoffs removed, on‑time‑in‑full, pick accuracy, model drift. We don’t celebrate launches; we celebrate Tuesdays, when the change keeps working.

There are villains I can’t fight: physics, for one, and heroic shortcuts for another. But I can introduce friction in the right places. We implement consent by design, privacy by default, and auditability as if the next curious intern were a friendly detective. Governance is my quiet superpower. A RACI that fits on one slide? That’s a utility belt. A model registry that remembers who did what when? That’s the grappling hook. Guardrails turn ‘move fast’ into ‘move fast without stepping on the cat.’

The warehouse chapter always begins with a whisper: ‘We lose a day a week to looking.’ Looking for pallets, for SKU labels, for the last person who knew why the bin says 24 when it clearly contains 23 and a mystery bolt. Sprocket shines here. We teach it to read barcodes, hear forklifts, and notice the dance pattern of a shift. Then we let it tell the shift lead, gently, that a new slotting pattern would save fourteen thousand steps and three colorful metaphors per night. We do not automate people. We automate regret.

Here’s what a typical mission looks like. Day 1: meet the skeptics. They’ve heard the pitches. They’ve survived the reorgs. They’re tired of being ‘the last mile’ blamed for everyone else’s first mistake. I do not wear a cape to these meetings. I bring coffee and the question I was trained to ask by a forklift driver named Lena: ‘What could we stop doing this week and no customer would notice?’ That becomes our first target. That becomes our first win.

Day 7: an evidence‑backed plan. Not a prophecy—a shopping list with receipts. What data we’ll need. Which APIs are real. Where the gnarly state lives. Who owns the Friday 4 p.m. escalations. We draw the path from pilot to platform. We estimate ROI in calendar time, not press releases: how many hours will return to the people, and how soon will the boring tasks apologize and leave?

Day 30: the first pilot wobbles into daylight. We budget the wobble. We expect the wobble. We hold a retro before the confetti, asking: did we simplify the right thing? Are we teaching the model with the best examples or just the most convenient? Who owns the pager when the algorithm gets the hiccups? We write runbooks that assume new teammates will join mid‑chase and still catch up. The second pilot goes faster. The third stops being a pilot and becomes Tuesday.

Not every mission ends with applause. A few end with mercy. We’ve retired pipelines that were carbonating data into froth. We’ve unshipped features that looked heroic in a demo and looked haunted at 2 a.m. We’ve told the truth about cost curves, especially the invisible ones—annotation debt, governance drift, the tax of debugging cleverness. Honesty is the most aerodynamic thing you can bring to a strategy.

People ask for my stack. I answer with verbs: collect, clean, label, learn, deploy, observe, adapt. The nouns come later, chosen for fit and patience. We choose tools that survive handoffs because the hero of every story is the person on call. If they can roll back with a single command and understand yesterday’s failure from today’s breadcrumbs, we picked right. If the system can explain itself to a reasonable human on a bad day, we can sleep.

A favorite case: the food distributor whose trucks were playing Tetris with reality. Pallets wanted to ride in the wrong order; drivers wanted to retire. The old heuristic was: ‘First in, last out, unless Bill says otherwise.’ Bill retired. Chaos did not. We stitched together a humble model with the brains to pack by delivery rhythm instead of raw volume, then taught it to respect a quirk called ‘Karen’s Corner,’ a convenience store that treats Wednesday mornings like holidays. On day one we shaved fifteen minutes off every stop on Route 12. On day seven the crew named the algorithm “Stackula.” On day thirty the complaints channel went quiet. The CEO thought that was silence. I recognized it as gratitude.

In the city’s hospital network, inventory once lied like a teenager about curfew. Supplies were everywhere and nowhere. Nurses kept private stash drawers defended with Post‑it diplomacy. We resisted the urge to ‘AI’ everything. We started with an agreed‑upon truth: the master list of critical items and the five events that ruin a shift when they go missing. From there we built a little sentry that watched consumption patterns and whistled when tomorrow looked pickier than today. The result wasn’t sexy. It was kind. Fewer scavenger hunts. More time for care. No one put that in a keynote. Everyone felt it.

Transformation is a team sport that looks suspiciously like hospitality. We set the table for skeptics: visibility into decisions, the right to veto unsafe ideas, and a change log that speaks human. We keep crumbs off the table by retiring jargon. We don’t ‘synergize.’ We ‘stop double‑entering addresses.’ We don’t ‘accelerate time‑to‑value.’ We ‘arrive when we said we would.’ The cape only works when the language is honest.

Some nights, the city summons me downtown to a boardroom with a view of a river that pretends it’s not plotting to flood in spring. The executives want guarantees. I hand them ingredients instead: a portfolio of small bets, staged like stepping stones, so each one teaches the next what to do with friction. We talk about risk the way sailors talk about weather. We don’t eliminate it; we learn when to leave harbor and when to reef the sails. We commit to alignment rituals—weekly ‘three truths’ where legal, ops, and data science each bring a fact the others must respect.

Rollouts are where capes catch on door handles. So we put handles on the doors. Training is modular, five minutes at a time. Every new tool arrives with a skipping rope—two ways to bail out gracefully without losing work, and a big button that says ‘I think it broke, please help’ that routes to humans who answer like neighbors. We give names to the scary parts, then we make them boring: drift checks, rollback drills, privacy audits. The team learns to treat alarms like smoke, not drama.

Ethics shows up like a mentor in a good comic book—early, calm, and inconvenient in the right ways. We do red‑team exercises where we pretend to be clever, lazy, or malicious and then design against ourselves. We don’t ask whether we can; we ask what happens when we shouldn’t. Sprocket has a rule: if the explanation won’t fit on a notecard, the model isn’t ready to drive. We respect the right to be left alone. We keep the receipts for consent. We make the default ‘no,’ and the opt‑in ‘yes’ feel like a choice, not a trick.

Another night: a national retailer wanted AI to staff stores like chess grandmasters—impressive, intimidating, and terrible with feelings. We taught the system manners. It learned that a schedule is not just math; it’s birthdays, bus routes, and the chemistry of crews who like working together. We encoded constraints that read like kindness: no closing then opening unless you begged and were bribed, no splitting weekends like a wishbone. The result was not a perfect schedule. It was a schedule that apologized less. Turnover slowed. Sales rose. The capes stayed in the closet; the people stayed on the team.

Resistance is not a villain; it’s an immune system. When a team says ‘we tried that in 2019,’ I ask for the autopsy. We harvest what failed for parts. Maybe the model was right but the org chart was wrong. Maybe the data was fine but the incentive plan considered reality optional. Strategy isn’t a slogan; it’s gravity. If you want new behavior, you have to tilt the floor. We adjust metrics so the right thing feels like less work. We retire a meeting to fund an experiment. We treat attention as finite and design for mercy.

Sometimes the ask is absurd in the way that makes you grin: ‘Can you deploy a forecasting model in a week?’ No. Can we run a shadow model in a day that tells us whether a week would be worth it? Absolutely. The superhero move is not acceleration; it’s triage. We prioritize problems that, when fixed, stop five other problems from needing to exist. The cape is leverage. The secret identity is sequencing.

The dicey bits are where we earn the emblem on the chest. A vendor promises a black box with sparkles; we smile, dissect, and often lob it gently into a museum for cautionary artifacts. A rogue spreadsheet fights back; we de‑fang it by reproducing its only useful feature in a system that doesn’t forget what month it is. A model begins to drift toward nonsense; we tighten feedback loops, reward the reporting of weirdness, and treat recalibration like brushing teeth—twice daily would be obsessive, never is a horror story.

Logistics will always offer plot twists. Weather turns childish. Bridges renovate themselves without warning. Radio dead zones expand like shy galaxies. We layer resilience into the system like secret passages: alternate carriers, graceful degradation modes, and last‑mile partners who love the smell of a challenge. When everyone else sees a detour, we see a chance to be the reliable one.

Across missions, the pattern holds: small pilots, crispy metrics, human‑in‑the‑loop by design, and a sunset plan for anything that doesn’t age well. We write the exit before the entrance. We decide how to measure success and how to declare victory without moving the goalposts. When success arrives, it looks ordinary: the line that used to kink, straightens. The count that used to lie, tells the truth. The Slack channel where complaints gathered goes oddly quiet. We check to make sure quiet isn’t neglect; it’s relief.

I keep souvenirs: a retired barcode scanner with a sticker that says ‘be kind, rewind’; a paper napkin map of a cross‑dock that ran on vibes until it learned math; a Post‑it that reads ‘we are not behind; we are between.’ They remind me that the work is not to dazzle; the work is to deliver—via people who will still be here when the consultants leave and the capes go to the dry cleaner.

People want to know what it feels like when AI implementation finally lands. It feels like this: the meeting that used to spiral now ends early. The forecast that used to squint now wears glasses. The supervisor’s radio is quiet enough for music. A customer who used to say ‘where is it?’ now says ‘thanks for the heads‑up.’ The change does not post selfies. It shows up as fewer apologies.

Every so often I’m asked for my origin story. It isn’t glamorous. I learned to say, ‘I don’t know, but I can find out,’ and people kept letting me into rooms. I learned to carry two pencils and to return borrowed time with interest. I discovered that logistics is the love language of civilization and that strategy is logistics whispered across a year. Mostly, I learned that technology is a promise kept or it is a prank, and pranks do not scale.

There’s a rooftop where I like to stand when a mission ends. The city looks less like a tangle and more like a circuit that just clicked into place. Freight trains gossip with river barges. Semis trade nods with bicycles. Drones pretend they are bats and remember to be polite. Somewhere, Sprocket hums through a healthy batch job, noticing nothing dramatic and generating nothing scandalous. That is the dream: calm competence wearing the cape like a lined jacket on a windy day.

If you’re reading this because your operation is a page‑turner with too many plot holes, I offer this: we can make it a better story. Not with fireworks—though I’ve seen a forecasting residual graph sparkle like New Year’s—but with craft. We will catalog the hard parts. We will choose a villain we can defeat in six weeks. We will design a secret identity for the change so it doesn’t show up to work wearing a conference badge. We will build dashboards that answer the question your CFO actually asks, not the one the vendor hoped they would.

And when the first result arrives, we won’t declare victory. We’ll take better notes. We’ll teach the model to say ‘I’m not sure’ in four different dialects of humble. We’ll give the team something smaller to do with the time they just won, and then something bigger the next month. We’ll tie the incentives to livable goals and celebrate status reports that tell the truth even when the truth has frizzy hair.

The city doesn’t become Gotham or Metropolis after a good quarter. It becomes itself—just a little kinder, a little sharper, and a little better at keeping promises. The same is true for companies. The best AI implementations don’t make you unrecognizable; they make you unmistakably you, at your best, on purpose. Logistics stops being a rumor and becomes a song. Strategy sheds its mask and becomes a schedule.

I cannot promise you capes or comic‑book epilogues. I can promise a craft: a way of moving from idea to first working thing to reliable habit. I can promise that your people will learn to trust the system because they helped teach it manners. I can promise that we will measure the right stuff, retire the wrong stuff, and politely ignore the shiny stuff until it earns its place.

If you want that, invite me in. Bring your messiest processes, your stickiest SKUs, your forecasts with stage fright. Bring your best skeptics. Bring the last three retros you’re proud of and the one you’re not. I’ll bring Sprocket, the checklists, the maps, and the stubborn belief that the ordinary heroics of logistics and the quiet power of well‑behaved AI can make your work feel like it’s wearing the right suit of armor at last.

Then we’ll get to it. No fireworks. No ceremonial ribbon that later becomes a trip hazard. Just the work: aligning the plan to reality, answering the radios, and turning the city’s blinking lights into a language that says, over and over, we’ve got this. Because we do.