GreenBox has 12,000 subscribers, three cities, three squads, twenty-five people, and a weekly discovery cadence that keeps each squad aligned with customers. But Charlotte has a problem: the squads are locally excellent and globally confused.
Perth is optimising the subscription experience – pause flows, delivery preferences, box customisation. Melbourne is building a B2B meal kit offering for office catering companies. Brisbane is expanding the pilot, testing new box sizes and farm partnerships. Each squad is doing good work. Each squad is talking to customers every Tuesday. Each squad is running discovery with discipline.
On a Monday morning in mid-October, the Perth squad is doing its assumptions check when Sam looks up from her phone. “Has anyone seen this?”
She reads the headline aloud: “Hartland Group acquires Freshly for undisclosed sum.”
The room goes quiet. Freshly – the VC-funded competitor from Sydney that has been GreenBox’s shadow for the past year. The company with $12M in funding, the polished app, the national delivery network, the aggressive pricing. Acquired by Hartland Group.
Maya is the first to speak. “They bought the model. They can’t buy what we built.”
Tom says nothing. He’s thinking about Dave – about the call Dave made to Maya a year ago when Freshly first offered him guaranteed volume. Dave didn’t switch. He stayed. Not because GreenBox could match the volume, but because GreenBox knew his name.
The acquisition validates the market. It also changes the competitive landscape. Freshly’s VC aggression – the below-cost pricing, the driver recruitment blitzes, the ad campaigns – will likely soften under Hartland Group’s corporate umbrella. But Hartland Group brings something else: grocery store distribution, brand recognition, and a supply chain that dwarfs anything GreenBox could build in a decade.
Charlotte, listening from the doorway, says: “This changes the quarterly planning conversation. Let’s come back to it.”
But nobody has asked whether these are the right three bets.
When the board asks Maya where GreenBox will be in twelve months, she can describe what each squad is doing this sprint. She can’t explain how those sprints add up to a strategy. She can list features. She can’t tell a story.
Lee puts it simply: “You’ve got great weekly habits. You’re missing the planning layers above them.”
The onion
Charlotte introduces the planning onion at a leadership meeting with the three squad leads. The concept isn’t hers – it’s a common agile planning model that describes how decisions nest inside each other at different time horizons.
Each layer answers a different question. Each layer uses different techniques. Each layer operates on a different cadence. The outer layers change slowly. The inner layers change fast. And each inner layer should be able to explain how it connects to the layer outside it.
“The problem,” Charlotte tells the squad leads, “is that we’ve built excellent day and sprint habits. Our weekly cadence is solid. But we don’t have a quarterly planning practice, and our yearly strategy exists in Maya’s head and her board deck. The outer layers of the onion are missing.”
Tom, the Perth squad lead, nods. “So we’re making great sprint decisions inside a quarterly plan that doesn’t exist.”
“Exactly. Each sprint is locally optimal. The question is whether the sprints add up to something strategic.”
What GreenBox already has
Charlotte maps the techniques the team already knows to the layers of the onion. The exercise takes twenty minutes on a whiteboard and it’s immediately revealing.
| Layer | Cadence | Techniques already in use | Question it answers |
|---|---|---|---|
| Day | Daily | Standups, assumptions check, Cynefin classification | What am I working on and is it the right approach? |
| Sprint | Fortnightly | Example Mapping, ensemble programming, threat modelling, retros | What are we delivering, and have we understood it well enough to build it? |
| Quarter | Quarterly | Wardley Map review (new) | Where does this capability sit strategically? |
| Year | Annual | Impact Mapping (ad hoc) | What goal are we trying to reach? |
| Vision | Rarely | Business Model Canvas (done once), JTBD (done twice) | Why do customers hire us? |
“Look at the gaps,” Charlotte says. “The daily and sprint layers are rich. Quarterly is just a Wardley Map review, which tells us about build-vs-buy but not about what each squad should focus on. The yearly layer is an Impact Map we haven’t updated in six months. The vision layer is solid – we know the job customers hire us for – but we’ve never formally connected it to the layers below.”
Ravi studies the whiteboard. “So the inner rings are full and the outer rings are empty.”
“Not empty. Just… disconnected. Maya knows the vision. The board deck has a yearly goal. But none of that is flowing down through the layers to the sprint decisions each squad makes every fortnight. There’s no connective tissue.”
Anika, the Melbourne squad lead, puts it more bluntly: “We’re three squads running three separate businesses that happen to share a codebase.”
The room goes quiet. She’s not wrong.
Filling the gaps
Charlotte doesn’t introduce new techniques. She uses the ones the team already knows, applied at the right planning horizon. The whole point of the onion is that you don’t need a different toolkit for strategy – you need the same toolkit at a longer time scale.
Yearly: the Impact Map refresh. The team’s existing Impact Map was built around reaching 300 subscribers. That was a year and a half ago. The goal, the actors, the impacts, and the deliverables have all changed. The map is a historical artefact, not a living document.
Charlotte facilitates a half-day session with Maya, the three squad leads, and Lee. They build a new Impact Map from the current strategic goal: reach 25,000 subscribers and achieve profitability by end of next year.
The map is different from the first one in almost every dimension. The actors are different – the B2B channel, the Brisbane market, existing subscribers with expansion potential. The impacts are different – subscriber retention matters more than acquisition now, because reducing churn from 4% to 2% is worth more than adding new subscribers at current acquisition cost. The deliverables are different – the team isn’t building a subscription system from scratch any more. They’re optimising an existing platform.
Maya watches the map take shape on the wall and realises something. “The last time we did this, the map was about proving the concept. This time it’s about choosing where to invest. The technique is the same. The stakes are completely different.”
The yearly Impact Map becomes the reference point. Every piece of work should trace a line back to one of these impacts. If it can’t, the team needs to ask why they’re doing it.
Charlotte pairs the Impact Map refresh with a yearly retro – not a sprint retro or a quarterly retro, but a strategic reflection. Different people in the room: Maya, Charlotte, Lee, the three squad leads. Different questions: “What did we learn about the business this year? What assumptions did we hold at the start that turned out to be wrong? What would we do differently if we were starting this year over?”
The yearly retro surfaces something nobody had articulated. Lee says it: “A year ago, we thought Brisbane would be Perth but further north. It’s not. Brisbane customers behave differently – smaller boxes, more frequent orders, delivery speed over produce quality. We spent six months assuming similarity and three months discovering difference. Next year, when we enter a new city, we lead with JTBD interviews before we build anything.”
Every layer of the onion needs its own feedback loop. Sprint retros catch process issues within a squad. Quarterly retros catch strategic drift between squads. The yearly retro catches assumptions about the business itself – the kind that compound silently until someone names them.
Quarterly: theme-based planning. Charlotte borrows from Shape Up and from the OKR tradition, but keeps it lighter than either. Each quarter, the three squads and Maya agree on a small number of themes – not detailed feature specs, but areas of focus with measurable outcomes.
For Q1, the themes are:
- Retention (Perth squad): reduce churn from 4% to 2.5% by improving the pause/resume flow and delivery preferences
- B2B launch (Melbourne squad): sign three pilot corporate customers for weekly office boxes
- Brisbane scale (Brisbane/remote squad): grow Brisbane from 800 to 2,000 subscribers and establish three new farm partnerships
Each theme connects to the yearly Impact Map. Each theme has a measurable outcome – not a feature list. The squads decide how to pursue the theme. The quarterly plan decides what to pursue.
“Notice what this does,” Charlotte says. “Perth isn’t just optimising random features. They’re optimising features that serve the retention goal. Melbourne isn’t building B2B because it’s interesting – they’re building it because the Impact Map says diversifying revenue sources is critical to profitability. Brisbane isn’t growing because growth is good – they’re growing because the Impact Map says geographic expansion is the highest-leverage path to 25,000.”
Maya adds: “And when the board asks me what we’re doing, I can say: ‘Three squads, three themes, all connected to profitability by end of next year.’ That’s a story I can tell.”
Sprint: five-minute connection. The sprint layer was already strong. The change Charlotte makes is small but important: every sprint planning session now starts with a five-minute connection to the quarterly theme.
“What’s our quarterly theme? Retention. What are we working on this sprint? The pause/resume flow redesign. How does it connect? JTBD interviews showed that subscribers who want to pause can’t do it easily, so they cancel instead. Fixing this directly reduces churn.”
That five-minute connection takes no preparation. The squad lead reads the quarterly theme aloud and asks whether the sprint’s work serves it. If it does, the team proceeds. If it doesn’t, they have a brief conversation about whether the work should be there.
Tom finds this deceptively powerful. “We used to start sprint planning by looking at the backlog and picking the top items. Now we start by asking ‘does this serve the quarterly theme?’ Twice this quarter, the answer was ‘not really’ and we reshuffled. Both times, we caught work that was locally sensible but strategically off-target. A nice-to-have feature that would have taken a week, when a retention-focused bug fix would have taken two days and moved the metric.”
Vision: a trigger, not a schedule. The vision layer – the JTBD insight that customers hire GreenBox for weeknight dinner stress relief – doesn’t have a scheduled review. It changes rarely. But Charlotte puts a trigger on it: “If churn rises above 5% or NPS drops below 40, we re-run JTBD interviews. The job might have changed.”
The trigger matters more than a schedule. Most organisations review their vision annually because it’s on the calendar. Charlotte wants the team to review it when the evidence says something has shifted. The cadence is event-driven, not time-driven. It’s the same Cynefin logic the team applies to everything else: match the approach to the situation.
The quarterly planning day
Charlotte establishes a half-day quarterly planning session. Five steps, two and a half hours, all three squads in the room (Brisbane on video).
- Review last quarter’s themes. Did we hit our outcomes? What did we learn? Where were our assumptions wrong? (30 minutes)
- Update the Impact Map. Has the strategic picture changed? Are there new actors, new impacts, new constraints? (30 minutes)
- Propose next quarter’s themes. Each squad lead proposes a theme with a measurable outcome. Maya adds strategic themes from board-level priorities. (45 minutes)
- Dependency check. Do any themes require work from multiple squads? Where will squads need to coordinate? This is where the bounded contexts from the DDD work pay off – most dependencies are data flowing between contexts via events, and those interfaces are already defined. (30 minutes)
- Commit. Agree on three to five themes. Each theme has an owner (a squad), an outcome (measurable), and a Cynefin classification (Clear themes get less discovery, Complex themes get more). (15 minutes)
The quarterly plan doesn’t specify sprints. It specifies direction. The squads break down the themes into sprint-level work using their existing practices – Example Mapping for complicated stories, experiments for complex ones, quick conversations for clear ones.
The rhythm, updated
Charlotte updates the GreenBox cadence that she introduced in Continuous Discovery:
| Layer | Cadence | What happens | Techniques |
|---|---|---|---|
| Day | Daily | Standup + assumptions check | Cynefin |
| Sprint | Fortnightly | Sprint planning, Example Mapping, build, retro | Example Mapping, ensemble, threat modelling |
| Month | Monthly | Impact Map check, customer interview synthesis | Impact Mapping |
| Quarter | Quarterly | Theme setting, Wardley Map review, dependency check | Impact Mapping, Wardley Mapping, JTBD |
| Year | Annual | Strategy refresh, Impact Map rebuild | Impact Mapping, Assumption Mapping, BMC |
Every technique the team learned over the past eighteen months has a home in this table. Nothing was added. Nothing was invented. The planning onion just made the connections visible.
What planning is NOT
Charlotte is careful to address the elephant in the room. The GreenBox team spent months learning that building on assumptions is dangerous. Planning sounds like making assumptions about the future and building on them.
“Planning isn’t predicting,” she tells the team. “The quarterly themes aren’t a commitment to ship specific features. They’re a commitment to pursue specific outcomes. How you pursue them will change as you learn. The sprint work will shift. Stories will be added and dropped. The experiments might invalidate the premise entirely.”
“If Brisbane customer interviews in week three reveal that nobody wants the weekly box size you’re piloting, you don’t keep building it because the quarterly plan said so. You adjust. The plan gave you direction. The discovery gives you truth. When they conflict, discovery wins.”
Lee, on the call, adds the context that only a coach with scars can. “The teams I’ve seen fail at planning are the ones who treat the plan as a promise. ‘We said we’d ship X by March, so we’re shipping X by March, even though the customer interviews in January told us X is wrong.’ That’s not planning. That’s obligation theatre.”
“Good planning is a hypothesis. ‘We believe that focusing on retention will have more impact than focusing on acquisition this quarter.’ That’s testable. You run the quarter, measure the results, and adjust for next quarter. It’s the same probe-sense-respond cycle from Cynefin, just at a longer time scale.”
Ravi asks the practical question: “What if our quarterly theme turns out to be wrong halfway through? Do we wait until next quarter to change it?”
Charlotte: “If you discover something that fundamentally changes the picture – a major competitor enters Brisbane, or the B2B pilot gets a huge enterprise customer – you bring it to the monthly Impact Map check. That’s what the monthly cadence is for: catching strategic shifts between quarterly planning days. Most months, nothing changes. When something does, you escalate it.”
The first quarterly planning day
The planning onion settles in. It’s not dramatic – it’s quiet. The quarterly themes give each squad a frame that makes sprint decisions easier. The yearly Impact Map gives Maya a story she can tell the board without scrambling.
The first quarterly planning day surfaces a dependency nobody had spotted. Melbourne’s B2B offering needs allergen handling that’s more sophisticated than the consumer product. The decision tables that Maya and the Perth squad built for consumer allergen substitutions don’t cover the corporate catering requirements – workplace health and safety rules are stricter than consumer preferences. Perth needs to extend the decision tables before Melbourne can ship the B2B allergen feature.
Without the quarterly planning day, this dependency would have surfaced as a blocker mid-sprint instead of a planned piece of work. The DDD bounded contexts helped – the interfaces between supply matching and customer profiles were already defined – but the business dependency was invisible until someone asked: “What do you need from another squad to hit your quarterly outcome?”
Anika calls it the most useful meeting she’s attended at GreenBox. “Two and a half hours, and we avoided what would have been three weeks of blocked work in February.”
The second quarterly planning day
The second quarterly planning day produces a surprise. The Brisbane squad’s customer interviews – Ravi’s team has been running them every Tuesday for three months – have revealed that Brisbane subscribers use GreenBox differently from Perth and Melbourne subscribers. They order more frequently, in smaller quantities, and they care more about delivery speed than box contents.
The quarterly theme shifts from “grow to 2,000 subscribers” to “understand the Brisbane usage pattern and build for it.” That’s a Complex problem, not a Complicated one – the team needs experiments, not analysis. Assumption Mapping reveals six assumptions about Brisbane customers that the team has been treating as facts. Three of them are wrong.
The shift wouldn’t have happened without the planning layer. The sprint-level work would have kept optimising for the original assumption: Brisbane customers are like Perth customers, just newer. The quarterly review forced the team to check that assumption against the evidence from three months of customer interviews. The evidence said otherwise.
Ravi is relieved. “We were about to spend the next quarter building features for a customer that doesn’t exist. The Brisbane customer isn’t a Perth customer with a different postcode. They’re a different person with a different relationship to cooking. We nearly missed that.”
The board conversation
Maya presents the Q2 board report differently from any report she’s given before. Instead of a list of features shipped, she presents three outcomes pursued, what was learned, and how the plan adapted.
The board is more engaged than she’s ever seen them. One board member leans forward: “This is the first time I’ve understood what your engineering team actually does. Before, I got a list of feature names that meant nothing to me. Now I get: ‘We bet on retention, we reduced churn by 1.2 percentage points, and here’s what we learned that changes our Q3 bet.’ That’s a conversation I can have.”
Diane, who’s been on the board since the seed round, asks the question Maya used to dread: “When will Brisbane be profitable?”
Charlotte answers before Maya can. “Each squad ships about eight stories per sprint. The Brisbane squad has roughly forty stories remaining for the core subscription flow, plus an unknown number for the smaller-box format we discovered in the customer interviews. At current throughput, the known work is about five sprints – ten weeks. The unknown work depends on what the next round of experiments reveals.”
Diane nods. “So you’re saying you don’t know exactly, but you know the range and you know what would change it.”
“Exactly. We forecast from throughput – how much we’ve been shipping – not from estimates of individual stories. Throughput is observable. Individual estimates are guesses.”
This is the estimation approach that Charlotte has evolved across the GreenBox journey. At five people, Lee used gut feel – “does this feel like a sprint’s worth?” At the roadmapping stage, Charlotte introduced t-shirt sizing for communicating rough scope to stakeholders. At two squads, t-shirt sizing helped coordinate cross-team dependencies. Now, at three squads and board-level planning, throughput-based forecasting gives Maya a credible answer to “when?” without pretending to know more than she does.
The principle is consistent: estimate at the planning layer that needs it, using the lightest technique that works. Sprint stories don’t need formal estimates. Roadmap items need t-shirt sizes. Board-level forecasts need throughput data. Each layer gets the precision it needs and no more.
Maya catches Charlotte’s eye across the room and smiles. Six months ago she dreaded board meetings because the questions were unanswerable. Now the conversation is about outcomes, learning, and honest forecasts. The board isn’t asking when features will ship. They’re asking which bets to double down on.
It’s the same shift the team went through at every other level. Stop measuring output. Start measuring understanding.
The meta-lesson
The planning onion isn’t a new technique. It’s the recognition that the techniques the team already knows operate at different time horizons, and that those horizons need to be connected.
Event Storming maps the domain. Example Mapping makes stories concrete. JTBD reveals customer needs. Impact Mapping connects work to goals. Wardley Mapping positions capabilities strategically. Cynefin matches the approach to the problem. Ensemble Programming puts all the thinking in one room. Threat Modelling asks what could go wrong. These are all layers of the same thing: understanding before building. The planning onion just makes the layers explicit and gives each one a cadence.
The trap Charlotte sees in most organisations is teams that are excellent at one layer and absent at others. Startups often have a strong vision layer and a strong daily layer, but nothing in between – the founder knows where they want to go and the team knows what to build today, but nobody has connected the two through quarterly themes or sprint goals. Enterprises often have strong yearly and quarterly planning but weak daily and sprint practices – elaborate roadmaps that decompose into sprints that nobody believes in.
GreenBox had the opposite gap. Excellent daily and sprint practices. Excellent discovery techniques. No strategic planning layer to connect them. The techniques didn’t change. The planning didn’t add process. It just made the connection between “why are we doing this?” and “what are we building today?” visible at every level.
Lee, reflecting on the GreenBox journey from the very first retro to this moment, puts it this way: “In month one, the team’s problem was that they didn’t understand the domain well enough to build anything useful. In month eighteen, the problem was that they understood the domain beautifully but didn’t have a framework for deciding which part of it to work on next. Both problems have the same root cause: invisible assumptions. Event Storming made domain assumptions visible. The planning onion makes strategic assumptions visible. It’s the same muscle, used at a different scale.”
The dashboard
It’s a Friday afternoon in late October. Maya is at her desk in the Perth office. The photo of her parents’ farm is still there, next to her monitor – the paddocks, the kitchen garden, the Colorbond roof that needed replacing. The farm that became a subdivision. The photo is faded now, the edges curling slightly. She’s never moved it. She never will.
The office is bigger than it was. The laminated Event Storm photos are still on the wall behind her – Charlotte called them “one of the smartest things a founder has ever done.” The whiteboard chart that Priya started updating in month two is digital now, displayed on a screen in the common area, but the format is the same. Subscriber count, churn rate, delivery success rate, farm accuracy. Priya still updates it every Friday.
Maya pulls up the GreenBox dashboard – the one Tom built years ago and that three squads have iterated on ever since.
14,200 subscribers. Three cities: Perth, Melbourne, Brisbane. Twenty-five people across the organisation. Thirty-one partner farms. A weekly delivery cadence that runs like clockwork – boxes packed on Wednesday, dispatched Thursday morning, on doorsteps by Thursday evening. The two-tier model – local boxes at $25, mixed-source at $20 – is generating enough margin to cover the team, the warehouses, and the courier contracts, with enough left over that the board has stopped asking about the next funding round. Three B2B pilot customers in Melbourne, one of which just doubled their order.
She scrolls through the week’s customer feedback. A woman in Fremantle writes: “I haven’t had to think about what to cook on a weeknight in six months. GreenBox just takes care of it.” A young couple in Brunswick: “The seasonal produce is amazing. We’ve started cooking things we’d never have bought ourselves.” A single dad in Brisbane, one of the pilot subscribers: “My kids now eat vegetables because they’re excited about what’s in the box. That’s not something I thought I’d ever say.” An office manager in South Melbourne: “The team fights over the GreenBox lunch boxes on Wednesdays. That’s the kind of problem I want to have.”
Maya smiles. She thinks about what’s behind those numbers.
The substitution engine that started as her brain and became a decision table and then an ensemble-built pipeline that handles three cities’ worth of seasonal produce, allergen profiles, and customer preferences. The farm relationships that started with Dave and Rachel and now include fourteen partner farms across three states. The team that started as five people in a room and now spans two time zones.
She thinks about the things that didn’t work. The Brisbane meal planner that nobody wanted. The recipe suggestion feature that the Impact Map review caught before it shipped. Tom’s first subscription model. Tom’s second subscription model. Jas’s customisation designs. The payment debugging tool that nearly logged credit card numbers to production. The quarter Perth spent optimising features that didn’t serve any strategic goal. Each failure was a lesson. Each lesson made the next iteration better.
She thinks about the things that did work. The Event Storm that changed how the team thought about the domain. The retro that changed how they thought about each other. The pause button that reduced churn by 40%. The JTBD interviews that revealed the real value proposition: not fresh vegetables, but the absence of weeknight dinner stress. The Brisbane single-person box that outsold every other size because a Tuesday interview caught what a planning meeting missed. The quarterly planning day that caught the allergen dependency before it became a three-week blocker. The board presentation where, for the first time, a board member said she understood what the engineering team actually does.
She thinks about the people. Tom, who went from rebuilding the subscription model twice to running ensemble sessions where six people think together. Priya, who went from being blocked on unanswered questions to being the team’s sharpest edge-case thinker. Jas, who went from designing the wrong product to designing a customer experience built on trust. Sam, who went from answering emails nobody could answer to catching a security gap that could have ended the business. Kai and Ravi and Anika, who joined during the scaling phase and now lead their own squads. Charlotte, who turned a set of ad hoc practices into a rhythm that twenty-five people follow – from daily standups to yearly strategy, every layer of the onion connected. Lee, who showed up one Friday afternoon and asked the question that changed everything: “When was the last time you all stopped and talked about how the work is going?”
She thinks about the planning onion on the wall of the Perth office. Charlotte drew it on a whiteboard three months ago and someone printed it on A3 paper. It’s the simplest diagram in the building and the most referenced. Vision at the outside, day at the centre, everything connected. Not because someone mandated a framework, but because the team realised that understanding before building doesn’t stop at the sprint boundary. It goes all the way up to the question of why the company exists.
She remembers the kickoff meeting – the one where she described the business as “farms list what they have, customers subscribe, we match and deliver.” It sounded so simple. It took a wall of sticky notes, twelve hotspot stickies, and a fundamental argument about whether customers choose their box contents to reveal how complex it actually was.
She remembers Tom rebuilding the subscription model twice in the first month. Jas throwing away a complete set of customisation designs. Priya blocked for two weeks on unanswered questions. The retro where Lee asked, “When was the last time you all stopped and talked about how the work is going?” and nobody had an answer.
She remembers the Event Storm that changed everything. Seven people, three hours, sixty sticky notes on a wall. The moment Tom realised his assumptions about fixed box contents were wrong. The moment Jas understood that the value proposition wasn’t choice – it was trust. The moment Dave and Rachel’s farming events appeared on the timeline and the team realised the supply side was an entire domain they hadn’t mapped.
She remembers Lee’s first piece of advice: “You’ll feel like you’re going slower. You’re not. You’re just putting the learning where it’s cheap – on sticky notes and whiteboards instead of in pull requests and production incidents.”
He was right. The learning was always going to happen. The only question was whether it happened cheaply – through workshops and conversations and sticky notes – or expensively, through wrong code, wasted sprints, and frustrated teams.
GreenBox chose cheap. Not perfectly. Not every time. They still built things that didn’t work. They still made wrong assumptions. They still had weeks where the discovery cadence fell apart and the squads drifted. But they always came back to the wall. They always came back to the conversations. They always came back to understanding before building.
And eventually, they learned that understanding before building doesn’t just happen at the story level, or the sprint level, or even the quarter level. It happens at every layer of the onion. From “what am I doing today?” all the way out to “why does this company exist?” Each layer is just discovery at a longer time horizon.
The tools changed along the way. LLMs went from a shiny novelty that helped Tom build the wrong thing faster to embedded infrastructure that transcribes interviews, generates code from decision tables, drafts ADRs, and produces first-pass threat models. The team’s relationship with LLMs matured from “give me the code” to “help me think.” That shift – from using LLMs as code generators to using them as thinking amplifiers – is the story underneath the GreenBox story.
But the LLMs never replaced the wall of sticky notes. They never replaced the Tuesday customer interview. They never replaced the five-minute assumptions check on Monday morning. They never replaced the quarterly planning day where three squad leads sit in a room and ask each other: “What do you need from us to hit your outcome?” The thinking still has to happen between humans. The LLM just makes it faster to act on what the thinking reveals.
Maya closes the dashboard and picks up her bag.
The market
Saturday morning. Late October. The Margaret River farmers’ market.
GreenBox has a stall. It’s new – Charlotte suggested it three months ago as a brand-building experiment, and Maya agreed because the market is where everything started. The stall is modest: a fold-out table, a GreenBox banner, sample boxes, and a signup tablet. Sam is running it, because of course Sam is running it. She’s talking to a woman with a toddler about the single-person box option.
Dave is at his usual spot, three stalls down. He’s selling surplus seasonal produce – the overflow that doesn’t fit into the GreenBox supply commitment. Ben handles the commercial relationship now, but Dave still comes to the market. He has since before GreenBox existed. He’ll come after GreenBox is gone, if it goes. That’s what farmers do.
Lee arrives mid-morning. He’s carrying a surfboard under one arm, wetsuit still damp. He looks slightly ridiculous and completely at ease. He buys coffee from the van at the end of the row and walks over to the GreenBox stall.
Maya is there now, standing behind the table with Sam, handing out sample produce to people who stop. She’s not the CEO at this moment. She’s the woman who grew up on a farm three hours south of here and who built a company that connects farms to families.
Lee watches for a while. Then he walks over to Dave’s stall.
“How’s the zucchini?”
“Enormous,” Dave says. “Take some.”
Lee picks up a zucchini the size of his forearm. “This is absurd.”
“That’s nature. You plant it, you water it, you stand back.” Dave glances at the GreenBox stall. “She’s done all right, hasn’t she.”
Lee follows his gaze. Maya is laughing at something Sam said. A customer is signing up on the tablet. The morning sun is coming through the market’s tin roof at an angle that makes everything look warm.
“She has.”
“That Freshly mob – the ones Hartland Group bought.” Dave pauses. “They rang me again, after the acquisition. Said Hartland Group would guarantee volume. Two hundred crates a week.”
Lee looks at him. “What did you say?”
“I said I already have a customer who knows my name.”
Lee smiles. He puts the zucchini in his bag and walks over to Maya. She’s alone for a moment – Sam has gone to get more sample boxes from the car.
“I saw the Hartland Group announcement,” Lee says.
“Everyone saw the Hartland Group announcement.”
“How do you feel about it?”
Maya considers. “Relieved. Scared. Validated. In that order.”
Lee nods. “Hartland Group bought the business model. They can’t buy what Dave and Rachel built with you.”
Maya looks at him. Then at Dave, who is patiently explaining to a tourist that no, the zucchini really is supposed to be that large. Then at Sam, who is coming back from the car with a stack of boxes, talking on her phone about a Brisbane delivery issue. Then at the GreenBox stall, with its modest banner and its fold-out table and its sample boxes full of produce from farms within fifty kilometres.
“This is what I wanted to build,” she says. “Not the tech. Not the platform. This.” She gestures at the market, the stalls, the people. “Farms and families. Connected.”
Lee doesn’t say anything. He drinks his coffee and watches the market fill up.
It’s a Saturday morning. Late October. In Margaret River, a farmer is selling zucchini next to a produce-box company’s stall. In Perth, two hundred and forty boxes are being packed for Thursday delivery. In Melbourne, an office manager is looking forward to Wednesday’s lunch boxes. In Brisbane, a woman who lives alone in New Farm is cooking dinner with exactly the right amount of produce, none of it going to waste.
GreenBox is a produce-box company connecting local farms with subscribers who want weekly produce boxes.
That’s what it was at the beginning. That’s what it is now. The same words. A different thing entirely.
Charlotte’s summary is the simplest thing anyone has said in eighteen months. It’s also the truest.
“Planning is discovery at a longer time horizon. That’s it. Everything the team learned about understanding before building – just zoom out.”