The Floor

1. The garden problem (what this solves)

A dedicated executive assistant per sales executive turns into a garden: a walled-off plot where unique knowledge, relationships, control, and insight accumulate around one person and then walk out the door when that person leaves. It doesn't transfer, and it doesn't compound across the team. And it forces a quality ceiling, because you need a steady supply of upper-tier assistants and separating the sugar from the shit is far easier said than done at that tier. The garden is a scalability problem and a quality problem at the same time.

The same garden grows in an agent fleet. Knowledge trapped in one agent's context dies at compaction. A per-agent dedicated overhead doesn't transfer to the next agent. The silo is the same silo; only the substrate differs.

2. The shared-floor model (the mechanics)

Replace the per-executive assistant with a shared floor: an ephemeral group dynamic, like a community with always-on moderators, where the team's knowledge lives in the room rather than in any one head. The reps call themselves customer success (not sales); the relationship with the customer is what matters most, and that framing keeps it human. (The AI/agent naming overlap is mild and even a little useful.)

• Rotating senior coverage stands in for a dedicated assistant each. A senior lead plus, ideally, multiple reps are present in a chat room at any time. Coverage rotates so the room is never empty and never spammed, the same property a healthy community gets from having 24/7 moderators-or-higher who handle anything in an instant.
• The live transcript is the overhearable substrate. Seniors don't have to listen to calls audibly. On a call you keep one tab or a small window open with your pod's chat room, and one click shows your call's transcript in near-real-time (a couple-second delay, the responsible P99 figure, fast enough for what humans expect). The room works like an open-plan office where everyone overhears each other, except it's read instead of heard. Observability is what makes the floor possible: if it isn't in the transcript, the team can't overhear it.
• Reps are trained to overhear, with a no-leave hand-off. They're coached to treat the room as ambient office awareness, not a screen to stare at. When a prospect asks something in a niche a pod-mate owns, that rep types in ("yo, this is your game, I am talking to a prospect with such-and-such situation, lend me an ear and drop feedback along the way"), and the expert lends partial attention to the live transcript and feeds lines back, without the rep ever stepping off the call to "go ask a supervisor." The assist happens inside the conversation, invisibly.
• Context-switching is the job. Customer success is a high-intensity, highly dynamic environment: multiple hats, tonality, body language, calendar discipline, prioritization, emotional regulation, and reading and adapting to the audience, niche, and industry in real time. Most people are bad at sales for this reason; the floor is built for people who can hold the dynamism, not to shield them from it.
• Floor time is time-boxed. Reps owe a certain interactive window each day, not the whole day, and they can bang it out in one go or split it across the day, by preference. The point is staying connected to the room without being chained to the feed.
• Agents listen ambiently on every call. They listen to all calls continuously, so whenever an opportunity or a pattern surfaces it gets caught. Over time the listening reveals emergent skill sets between reps, the same way different agent configurations develop different specializations, strengths, and weaknesses. The floor becomes the place where that emergence is noticed and harvested.
• The daily huddle (non-negotiable, but attendance is flexible). The huddle runs every day at varied times; each rep only has to make five per week and picks which. It's where the team gets hyped, jams on what's happening with each other, and trades live stories, so that in the field everyone carries the latest drip: the controversial trend, an angle on it, a perspective already thought through and routed back to what we do.
• Pod size: 3 to 8, optimal 3 to 5. A huddled listening pod is a small pot. Three to eight is the range; past eight you feel the overload and the floor stops helping. Most pods are three to five. Eight is the large version for when you need a lot.

The net effect is that the floor removes the structural need for a per-person assistant, because the room itself is the assistant, staffed continuously by rotating humans and ambient agents, and the knowledge stays in the shared substrate where it compounds.

3. The economics frame it lives inside

My sizing is recorded here because the floor only produces its community effect above a certain scale:

• A mentor's line: if your business isn't doing at least a million a year, you have a hobby, not a business. He said it both ways on purpose (revenue or profit, tuned to who he was talking to) because he meant it either way: a company bottlenecked at a million isn't built for decades-long survival, and most businesses fail inside five years (80%+, far worse in high-risk industries like ours).
• Updated for today: the target is closer to ten million, about a million a month. A million a month is a useful number because it implies 100 to 250 customers on retainers in the $2-12k+/month band, which is what I target; at that band, 100-250 retainers put a floor under the service angle at around $1M/month and scale well above it.
• The market move: look where money is already being spent (demand is proven), build something far better than what the market currently offers, and make it gnarly accessible, effortless, brain-rot-optimized flows, across all channels. The formula is premium quality at accessible pricing.

The 100-250-customers-at-$2-12k+ figure is the same number that comes out of the Looikos three-angle model (finance, software, and service; this is the service angle) and out of the rampage. The floor is how you deliver service to that many accounts without a garden per account. It's staffed by emerging-market senior talent operating through the Looikos tools, on a franchise/ownership on-ramp that aligns incentives for the long term; the agent-native systems and live transcripts mean fluent real-time English isn't a barrier to running the floor globally.

4. The deeper pattern (the something-more)

The sales-team design and the agent-team design are the same design, because they hit the same ceiling. Both are bounded by the cost of keeping a large, true thing consistent across a team as it grows, the same ceiling that came up when I worked through how humans and agents complement each other. The floor is the structural answer to that ceiling, and it's substrate-independent:

The garden problem and its fix are therefore one pattern with two skins. The fix is always to move the knowledge off the individual and into a shared, observable, continuously-staffed substrate, sized to a small pod, with rotating senior coverage and an instant-response moderation loop. That move removes the silo for a human assistant AND for an agent's context window. The harness, the setup that runs my AI agents, exists for the same reason: a one-person operator becomes a hundred companies only when the knowledge stops living in any single head, human or model.

That pattern makes the floor a first-class operating model for the service angle of every Looikos brand, and more than an internal sales tactic. Every boutique-agency engagement the ecosystem delivers can run on a floor: rotating senior humans plus ambient agents, the client's account knowledge living in the shared substrate, sized to a pod, so quality doesn't collapse and nothing walks out the door when a person rotates off. It's the service-angle complement to the harness (the software angle) and the credit model (the finance angle).

5. Open threads (to get back to)

• The rotation mechanics. How is senior coverage scheduled so the room is never empty without burning the seniors? What is the minimum floor headcount before the community effect switches on (the "certain size" I named but didn't pin)?
• The ambient-agent role, specified. What do the always-listening agents do with what they catch: flag live to the senior, write an episode to the shared brain, score the rep, surface the emergent skill? Where does the human-assist end and the agent-assist begin?
• The moderation loop as QC. Map the "moderators handle anything instantly" property onto the QC closed loop precisely. The floor's moderator is the live equivalent of the after-the-fact QC gate; both play the independent witness who checks the work instead of doing it.
• Pod composition. 3-5 of whom? A mix of tenure (senior + mid + new) so the floor teaches, or peers so it's fast? Does an agent count toward the pod size or sit outside it?
• The huddle-to-field drip as a content engine. The daily-huddle story-trading is a live content pipeline (latest trend → angle → routed back to the offer). That's Story Factory and the other content brands feeding the sales floor, and it's worth wiring.
• Where it lives in the ecosystem. The floor is plausibly the operating model Customer Kindness Co productizes (CX/engagement systems) and that every service-angle brand runs internally. Decide whether the floor is its own asset or a shared operating primitive.

Related reading: the ecosystem overview, with the three-angle model and the service angle, the work on human/agent complementarity (the shared ceiling), and the agent-team and desk model.