The hardest part of the work is the part where you leave.
Charlotte Wong closes Row 47 on a Thursday afternoon. Not formally, there’s no ceremony, no farewell dinner, no plaque. She updates her spreadsheet. The column that says “Status” changes from “Active” to “Complete.” The column that says “Duration” says “4 years 7 months.” The column that says “Outcome” she leaves blank for now. She’ll fill it in later, when she has enough distance to be honest.
Row 47. Greenbox. The longest engagement she’s ever had. The one that worked.
The spreadsheet
Charlotte keeps a spreadsheet of every engagement she’s taken since she started advisory work eleven years ago. Fifty-three rows now. Each row records the company, the duration, the initial problem, the practices she introduced, the outcome, and a single-line note she writes on the last day.
Some of the notes are clinical: Good discovery culture, weak execution discipline. Some are personal: Founder couldn’t let go. Neither could I. Row 46, the meal-kit company, the one before Greenbox, the one that didn’t work, says: Built the playbook. Team didn’t run it.
She hasn’t written Row 47’s note yet. She keeps opening the cell and closing it. The words aren’t ready.
The pattern
Charlotte has done this six times now. Not all fifty-three rows are the deep engagements, most are short: a two-week assessment, a workshop series, a quarterly check-in. But six times she’s embedded with a team for more than a year. Six times she’s watched a company go from “we don’t know what we’re building” to “we know exactly what to build next.” Six times she’s made herself unnecessary and then left.
The pattern is always the same.
She arrives when the team is struggling. Not failing, struggling. There’s a difference. Failing teams have given up on understanding the problem. Struggling teams are still trying, but their tools aren’t working. They’re building without discovery. They’re planning without data. They’re making decisions based on instinct and hope, which are fine for starting something and terrible for scaling it.
Charlotte brings the practices. Business Model Canvas to surface the assumptions nobody wants to examine. Wardley Mapping to see where the industry is moving. Value Stream Mapping to find the bottlenecks that everyone feels but nobody measures. Continuous discovery to keep the team connected to the people they’re building for.
She doesn’t install the practices and leave. She works alongside the team until the practices become habits. Until the Monday standup doesn’t need her facilitation. Until the retro happens without her prompting. Until someone new joins the team and learns the cadence from their colleagues, not from Charlotte.
Then she leaves.
The leaving is the measure of success. If the team needs Charlotte to maintain the practices, she hasn’t done her job. If the practices survive her departure, if they become part of how the team thinks, not just what the team does, then the engagement worked.
Row 42: a fintech in Sydney. Charlotte left after fourteen months. The discovery cadence lasted six months without her, then atrophied. Row 42’s note says: Practices adopted, not internalised. My failure, not theirs.
Row 44: a health-tech startup in Melbourne. Charlotte left after eighteen months. They’re still running weekly customer interviews three years later. The founder sends her their JTBD framework updates every quarter. Row 44 is a good row.
Row 47 is the best row.
What Greenbox taught her
Every engagement teaches Charlotte something. The fintech taught her that practices without executive buy-in don’t survive. The health-tech taught her that founders who do the interviews themselves build stronger discovery muscle than founders who delegate them. The meal-kit company taught her that you can’t want the outcome more than the team does.
Greenbox taught her something she hadn’t expected.
She’s sitting in her home office, the week after closing Row 47, thinking about it. The wooden ducks on her bookshelf. James bought the first one as a joke, and then it became a thing, the way things become things in a marriage, watch her impassively. The afternoon light is warm. It’s quiet.
Greenbox taught her that LLMs change the speed of everything but not the fundamentals.
Tom used LLMs from the very first week. Code generation, test scaffolding, documentation drafts, sprint planning assistance. The tools got faster every quarter. By the end, the LLMA neural network trained to predict the next token in a sequence, large enough that it generalises to tasks it wasn’t explicitly trained for. was embedded in every part of the engineering workflow, ensemble sessions, ADR drafts, interview transcription, domain model generation.
And none of it mattered without shared understanding.
The LLM could generate a subscription management system in an afternoon. It couldn’t tell you whether the subscription model was correct. It could produce a decision table with four hundred rules. It couldn’t tell you which rules Mrs Patterson cared about. It could draft an architecture decision record in perfect format. It couldn’t tell you whether the decision was wise.
The speed was real. The understanding was human. The practices, the Event Storming, the Example Mapping, the JTBD interviews, the continuous discovery, were the bridge between what the team knew and what the machine could build. Without that bridge, the machine built fast and wrong. With it, the machine built fast and correctly.
Charlotte writes this in her notebook. Not the spreadsheet, the notebook. The leather-bound one she carries to every engagement, the one that’s full of sketches and half-formed ideas and the kind of thinking that doesn’t fit in a cell.
Speed amplifies whatever you feed it. Feed it confusion, you get fast confusion. Feed it understanding, you get fast understanding. The practices aren’t slower than the tools. The practices are what make the tools worth using.
The bittersweet
Charlotte is good at leaving. She’s done it before. She’ll do it again. The professional distance is real, she’s an advisor, not a team member, and the boundary matters for both sides.
But Greenbox was different.
She watched Maya grow from a founder who couldn’t say no to a CEO who knew exactly what to prioritise. She watched Tom go from “I’ll just build it” to “let’s understand it first.” She watched Priya go from a quiet graduate who asked sharp questions to the architect of a system that handles tens of thousands of substitutions a week without error. She watched Sam build an operations function from a kitchen table and twelve spreadsheet tabs.
She watched a team go from chaos to craft. Not overnight. Not smoothly. Through retros that felt pointless and workshops that went wrong and a hire who didn’t work out and a migration that went sideways and a founder who nearly burned out and a board that had to learn to govern and a thousand small decisions that nobody remembers individually but that added up, over four and a half years, to something real.
Charlotte doesn’t get sentimental about clients. She gets sentimental about craft. And Greenbox was the purest expression of craft she’s encountered: a team that chose, over and over, to understand before building, to discover before deciding, to learn before committing. Not because Charlotte told them to. Because they experienced the alternative and decided it wasn’t good enough.
She pours herself a cup of tea. The ducks observe. James will be home soon with the twins. Thursday evenings are James’s night to cook, which means pasta, because James has three recipes and rotates between them with the reliable cadence of a man who believes in systems.
Charlotte smiles. She picks up her phone and texts Maya.
Row 47 is closed. Thank you for being the best row.
Maya replies twenty minutes later. You made us better. Don’t be modest about that.
Charlotte isn’t being modest. She’s being accurate. She brought the practices. The team did the work. The distinction matters to her, even after fifty-three rows.
The note
Late that evening, after pasta, after the twins have gone to bed arguing about whether Ollie cheated at cards (he did, Max has the evidence, the tribunal will reconvene tomorrow), Charlotte sits at her desk and opens the spreadsheet.
Row 47. The “Note” column. The cursor blinks.
She thinks about Maya at the BMC session, the moment the $3 margin appeared on the projector. Not defeat, recognition. She thinks about Tom in the ensemble sessions, the way his hands moved differently when he was building with the team instead of building alone. She thinks about Priya feeding Refactor on her keyboard while the substitution engine processed ten thousand boxes. She thinks about Dave, who she met months later but whose presence she’d felt from the start — the sticky note that said “Crop Failed” was already on the wall the first time Charlotte walked into the Greenbox office.
She thinks about Lee, who was never really a client at all, but who sat in every room with the particular stillness of a person who has learned to listen before he speaks and who taught Charlotte, without either of them naming it, that the best advice is often the question you don’t answer.
She types:
The team I didn’t want to leave. They didn’t need me to stay. That’s the whole point.
She saves the spreadsheet. Closes the laptop. Goes to bed.
The new client
Monday morning. Charlotte drives to a co-working space in Victoria Park, twenty minutes east of where Greenbox started in Fremantle. The building is a converted warehouse, exposed brick, too many pot plants, a coffee machine that someone has decorated with stickers.
The company is called Lumora. Twelve people. Series A. They build software for community energy cooperatives, solar panels, battery storage, shared metering. They’ve been growing for eighteen months and the founder, a woman named Grace, sent Charlotte an email three weeks ago that said: We’ve been building for six months and I think we’re building the wrong thing.
Charlotte read the email and felt the familiar pull. The sentence that means: we’re struggling, not failing. We’re still trying. Our tools aren’t working.
She meets Grace in a glass-walled meeting room. Grace is thirty-four, sharp-eyed, talking fast. She has the energy of a founder who is running on conviction and caffeine and the slowly dawning suspicion that conviction and caffeine are not a strategy.
“We’ve got twelve people and three different ideas about what the product should be,” Grace says. “Every sprint we build something and every retro we realise it wasn’t the correct thing. I feel like we’re going in circles.”
Charlotte listens. She’s heard this before. She heard it from Maya, in a living room in Fremantle, five years ago. She heard it from the fintech founder in Sydney, and the health-tech founder in Melbourne, and the meal-kit founder who didn’t make it.
The words are always different. The pattern is always the same. And yet Charlotte still feels the pull, every time. The pull of a team that wants to get better and doesn’t know how. That’s the thing she’s addicted to, if she’s honest with herself, not the frameworks, not the practices, not the satisfaction of watching metrics improve. The moment when a team realises they’ve been solving the wrong problem, and the terror and relief that cross the founder’s face simultaneously. Charlotte lives for that moment. It’s why she does this work. It’s why she’ll keep doing it long after she should probably retire and join James in his quiet campaign to visit every national park in Western Australia.
“Can I ask you something?” Charlotte says.
“Sure.”
“When was the last time you talked to a customer?”
Grace opens her mouth. Closes it. Opens it again. “We did a survey in January.”
“Not a survey. A conversation. Sitting with someone who uses what you’re building, or who would use it, and asking them what their day looks like.”
Grace is quiet for a moment. “We haven’t done that.”
Charlotte nods. She opens her bag and takes out her notebook, the leather-bound one, the one that went to every Greenbox session, every retro, every board meeting. It’s almost full. She’s been meaning to start a new one.
She turns to a fresh page. The leather is soft under her fingers. The pen is the same one she’s used for three years, a gift from James, practical and good, like most things James gives her.
“Let’s start there,” Charlotte says.
Grace nods. She looks relieved and terrified, which is the correct response to discovering that the answer to your problem is not a better product but a better question.
Charlotte has seen that expression before. She’ll see it again. It never gets old. It’s the moment when the work begins, not the frameworks, not the metrics, not the practices, but the slow, human work of understanding something well enough to build it correctly.
She writes, in the same neat handwriting that filled four and a half years of Greenbox notes:
Day 1.
Outside, the morning traffic hums along Albany Highway. Inside, twelve people are about to discover that they’ve been building the wrong thing. Charlotte will help them find the correct one. Or the correct-enough one. Or the one that teaches them to find the next one themselves.
That’s the pattern. Arrive, teach, build, leave. The cycle doesn’t end. The notebook fills. The rows accumulate.
Row 54 begins.