Good Fences Make Good Neighbors

“Good fences make good neighbors.” The line is older than any of us, and the longer I work, the truer it gets.

Robert Frost put the phrase in a neighbor’s mouth in “Mending Wall,” and people argue about whether he meant it. I have stopped arguing. In every part of my life that has worked — partnerships, projects, and now the AI systems I build — the same quiet truth keeps showing up: a clear boundary is not coldness. It is care. A fence does not say “stay away.” It says “here is mine, here is yours, and now we can stand at the line and be friends without stepping on each other.” The clearer the line, the warmer the neighbors. That is the idea I want to walk through here — first and most importantly about people, then about business, and finally about the agents that help me work.

First and foremost: clear roles between people

Before it is anything else, this is about the people I work with. The fastest way to ruin a good partnership is not bad intentions — it is unclear ones. Almost every hard feeling I have seen between partners traces back to the same root: someone assumed someone else was handling something, and no one had ever said out loud who actually owned it. People did not fall out because anyone was dishonest. They fell out because the fence was never built, so each person quietly imagined a different map of who was responsible for what.

So the kindest, most professional thing partners can do for one another is to draw the lines early and on purpose. Who is responsible for what. Where one person’s lane ends and the next begins. What is squarely inside someone’s scope — and, just as importantly, what is outside it, so no one is left expecting a partner to do a thing that was never theirs to do. A defined role is a gift: it lets a person give their best inside the lines and rest easy that the rest is genuinely covered. Misunderstood scope is where good people end up disappointed in each other for no good reason.

Compensation agreed up front is part of being a good neighbor

The other half of the fence is money, and it belongs in daylight at the start, not in a tense conversation later. Deciding up front what each role is worth — clearly, in writing, before the work heats up — is not awkward; it is respectful. It removes the single most common source of resentment by making sure no one is left guessing whether their contribution will be recognized. Agreeing on the reward for each role before anyone has lifted a finger is simply what good neighbors and great partners do. It is how you protect the relationship from the success, which can strain a partnership as badly as failure when the rewards were never spelled out.

None of this is cold arithmetic. It is the opposite. When the roles are clear and the compensation is settled, everyone can stop watching the fence line and get back to actually building — trusting, generous, and free. That is the win-win I am always trying to create: arrangements where every person can do well precisely because no one is confused about their part or their share.

Aligned rewards: the win-win, in general terms

Speaking generally — and only generally — the structures I admire most in real-estate partnerships are the ones designed so that everyone wins together, in the right order. A well-built arrangement first returns the partners who put up capital their agreed, expected return, and only then rewards the team that did the work with a larger share of the upside. People use words like the “promote” and the “waterfall” for these mechanics, but the spirit underneath is plain neighborliness: when the people who backed the deal hit the minimum they were promised and then some, they are glad to keep doing deals with you, and the team that delivered earns more for having delivered. Clear lines turn a one-time transaction into a relationship that compounds.

That alignment is the whole point. The goal is not to win against your partners; it is to set things up so the only way anyone wins big is by everyone first being made whole. Hitting and beating the targets your capital partners expect is what earns the right to do the next deal — and the one after that. Good fences, again: each party knows exactly what they get and when, so the success is shared cleanly instead of fought over.

The same fence in everyday business

Outside of formal partnerships, the principle holds everywhere two parties rely on each other. Scope of work, deliverables, deadlines, who decides, what happens if something changes — these are just fences, and they are the reason a working relationship feels respectful instead of tense. I learned this across years of design centers, jobsites, and specifications, where “I thought you had that” is one of the most expensive sentences in the building trades. Writing the boundary down is not a sign of distrust. It is what lets trust survive contact with reality.

Good fences for the guardian agents

And here is where this old proverb meets the newest part of my work. When I build with agentic AI, I give each agent a fence too — a defined scope, a clear lane, and a hard line it does not cross without a human. One agent watches for changes in the law; another organizes documents; none of them moves money, signs anything, or speaks for me in public without a person stepping in. The fence is the human-in-the-loop boundary: the machine surfaces inside its lane, and a named human decides at the line. I keep a guardian layer whose entire job is to hold those fences — not to be creepy, not to overstep, to stay the good neighbor even when no one is watching.

It turns out the rule for software is the rule for partners: a system you can trust is one whose boundaries are explicit, agreed, and held. I wrote the twelve specific places a human must keep a hand on the work in a companion piece — but the spirit is this same one. Good fences make good neighbors, and they make good agents, too.

A companion piece: Taming the Fire — Keeper in the Loop →
← Back to My Agentic Journey