The Orchard and the Algorithm

A parable to begin. A landowner plants an orchard on the edge of town. He spares no expense on infrastructure—miles of drip line, rust‑proof valves, a smart timer in a locked metal box. He hires pickers by the season and foremen by the month. When drought comes, the pipes hold. When frost threatens, the timer switches on the heaters like a faithful metronome. But one winter, an early cold snap hits on a holiday weekend. No one answers the phone. The timer sticks. The valves crack. The orchard survives, barely—but the next harvest is lean because the knowledge that lives in people wasn’t there when it mattered. The landowner has invested for years in things that will not leave him. He has not invested for years in people who will not fail him.

That’s the modern enterprise in miniature: multi‑year bets on applications, infrastructure, and platforms—paired with at‑will, here‑today contracts for the very people who breathe life into the system.

The asymmetry we rarely name

Most companies will happily sign a three‑ to five‑year agreement for an application suite. They’ll fund roadmaps, allocate depreciation, and pencil out ROI across fiscal calendars. The same companies hesitate to give an employee a guaranteed five‑year contract.

Why? Some of it’s math, some of it’s law, much of it is habit.

• Accounting favors code over craft. Software can be capitalized, amortized, and showcased on a slide with tidy arrows. Training, mentorship, and human continuity hit the P&L as expense. What we can itemize, we tend to prioritize. “Where your treasure is, there your heart will be also” (Matthew 6:21). Our spreadsheets reveal our loves.

• Risk is framed as exit, not formation. A multi‑year app contract has clauses, SLAs, and remedies. A multi‑year people covenant feels like risk without control. We like options on people and certainty from tools.

• Predictability sells. Applications don’t ask for parental leave, change managers, or outgrow their roles. Humans do—because humans grow. Growth is good; it’s also lumpy. Tools perform. People transform.

• Culture chases speed. In a market that feels like perpetual Santa Ana winds—hot, fast, unforgiving—we bolt down what looks stable and keep everything else on casters. We call it agility. Sometimes it’s just fear wearing athletic gear.

None of this makes leaders cynical; it makes them tired. The structure pushes them toward the purchase order, away from the promise. And then along comes AI.

AI and the rise of the “virtual employee”

AI systems—models, copilots, agents—arrive with the seduction of software and the promise of labor. They look like apps on a contract but work like teammates on a schedule. You can scale them at midnight, spin them down at noon, and negotiate price with procurement instead of with a person. For a CFO, that’s the HOV lane compared to surface streets at 5 p.m.—faster flow, fewer surprises, fewer intersections.

So what happens when we prefer multi‑year apps to multi‑year people and then discover an app that feels like a person?

Three near‑term shifts:

1. Budgets migrate from headcount to “heads‑of‑compute.” Teams will justify output through capacity units—tokens, jobs, agents—rather than FTEs. Some work will move from “who do we hire?” to “what do we wire?”

2. Supervision replaces management. We’ll assign owners not just to projects but to model behaviors, prompts, data diets, and drift. “Unit tests as seatbelts” becomes an HR policy as much as an engineering one. Cheap protection; always buckle up.

3. Apprenticeship gaps widen. AI will devour repetitive tasks where juniors learn the craft. If we don’t design new rungs on the ladder, we’ll wake up with no rungs at all—and wonder why nobody can reach the top.

Three long‑term risks to face with courage:

• Operational brittleness. Models drift like shifting dunes. They’re brilliant until the wind changes. Guardrails, evals, and red‑team routines must become as normal as coffee and code reviews.

• Moral muscle atrophy. Virtual employees execute. They do not care. Compassion, judgment, and courage grow in people through long obedience in the same direction. If we outsource the small wrestlings, we weaken the large ones.

• Cultural hollowing. Work is more than output; it’s how a people becomes a people. A company that treats humans as variable cost and bots as durable asset will get what it pays for: throughput without testimony.

What would a wiser pattern look like?

If we’re going to fund applications for years, let’s learn to fund humans with at least equal imagination. Not sentimentally—practically.

1) Five‑year mission charters for teams, not just tools

If an application earns a five‑year roadmap, a team should earn a five‑year purpose. Name the long game (customer promise, capability frontier, compliance journey). Fund the formation (apprenticeships, rotations, sabbaticals). Tie renewal to outcomes the way you tie license renewal to usage.

Practice: Draft a one‑page charter with three columns—Promise, People, Platform—and write five‑year commitments in each. Review annually. Amend with care.

2) A “skills depreciation” schedule that funds renewal

We reserve budget to refresh servers. Do the same for humans. Not because people wear out, but because the world does. Set a minimum annual investment per person in skill renewal—tracked and protected the way hardware refresh cycles are.

Practice: Commit a fixed percentage of automation savings to reskilling—say, 10% of AI‑generated productivity gains returned to human capability, automatically.

3) Apprenticeship, reimagined for AI

Let AI do the repetition while humans do the reflection. Design new rungs: shadowing AI workflows, annotating edge cases, writing evals, curating datasets, leading post‑incident learning. Juniors can become model stewards, not just task runners.

Practice: Pair each “virtual employee” with a human apprentice responsible for improving it. The buddy system—classic scouting wisdom—applies here too.

4) Governance that treats AI like a supply chain, not a mascot

Virtual employees need owners, SLOs, audits, and incident playbooks. Model risk is business risk. Write it down. Test before you trust. Knot check as validation.

Practice: Create an AI change log and a weekly “drift stand‑up.” Small, boring rituals prevent big, exciting failures.

5) Covenant language, not just contract terms

Contracts define scope; covenants define spirit. We can’t guarantee every job for five years. But we can promise five years of care—a commitment to transparency, reskilling, and humane transitions if roles change. That’s more than a policy; it’s a posture.

Practice: Publish a short “Human Promise” alongside your AI policy:

• We will invest in your growth at least as much as we invest in our tools.
• We will tell you early when change is coming.
• We will help you land, inside or outside, with dignity.

6) Portfolio thinking: perennials and annuals

In the orchard, you plant trees for decades and cover crops for seasons. In organizations, make clear which roles are perennials (core, compounding value) and which are annuals (project‑specific, time‑bound). Don’t blur the categories. Name them, fund them accordingly, and honor both.

Practice: Create one slide for the board: Perennial Roles (5‑year investment thesis) vs. Annual Roles (1‑year thesis). If a role moves categories, say why.

The counter‑argument—and the reply

“But the market is too volatile to promise five years to anyone.”

True: the weather is changing. That’s why shelter matters. A five‑year covenant isn’t a handcuff; it’s a canopy. It gives leaders the permission to form people, not just deploy them, and it gives people the confidence to grow into the work rather than guarding the exits. Think of June Gloom along the coast: the marine layer lingers, gray and cool. It still burns off by noon because something warmer is steadily working above it. Covenant is that steady warmth.

“But AI is simply cheaper.”

Sometimes. Until it isn’t—when a subtle model shift poisons a pipeline, when a regulator calls, when customers notice the compassion dialed down to zero. Humans are not cheap. They are priceless. That’s not a slogan; it’s an operating truth. Price buys output. Value builds a future.

“But guaranteed contracts are inflexible.”

So is a brittle system with no one who knows how it works. Flexibility comes from formed people who can adapt, interpret, and lead—people who have been given time to grow roots.

A quiet recalibration

Here’s a simple reframing you can start this quarter:

• For every multi‑year application you fund, name the multi‑year human formation it requires. If there isn’t one, ask whether the app is a crutch or a catalyst.

• For every “virtual employee” you hire, appoint a human owner and a human apprentice. Tools don’t make culture; people do.

• For every unit of speed you gain, purchase one unit of patience. Put it on the calendar: retros, ethics checks, listening sessions. Street‑sweeping day comes whether you remember it or not; calendars prevent tickets.

• Measure what matters. Track not only uptime and throughput, but also mentorship hours, internal mobility, and the ratio of automation gains reinvested into people. You become what you measure.

If you do this, AI becomes an instrument in a human orchestra, not a backing track that drowns out the strings. Virtual employees become amplifiers of human intent, not substitutes for human covenant. And your orchard—pipes and people—survives the winter.

A parable to end. A lighthouse keeper on a rocky coast receives a gift: a new automatic lamp with a fuel line and a clockwork that never seems to tire. Grateful, he lets the apprentices drift to other posts. For a year the light is perfect. Then a storm like no one remembers slams the bay. The clockwork chokes on salt. The fuel sputters. The keeper, now alone, scrambles through the spray, hands numb, mind racing to remember the old steps. He gets the lamp lit—late, flickering, but lit—because an old apprentice, passing by on leave, sees the dark tower and runs up the stairs two at a time to help. The town survives the night. In the morning, the keeper orders new parts for the lamp and posts a notice: “Apprentices wanted. Five‑year term. We will teach you the craft, and we will keep the light.”

That’s the future worth building—applications that hum for years, and people we promise to form for just as long.