The first instinct of most winery owners I talk with is to point AI at the part of the business that feels strategic. The brand voice. The next-vintage launch plan. The pitch deck for the wholesaler meeting. That instinct is the thing that quietly kills the rollout in week three.
The work that turns AI into a habit is the work the employee already complains about every Tuesday. Harvest paperwork. The same six-paragraph cancellation reply to the same kind of club member. The wholesaler follow-up email she has written 84 versions of in the last twelve months. The chores that show up on her list every week, take twenty minutes each, and never get retired.
That is the hated-task rule. Pick the recurring task your employee complains about every week, and have her use AI on that. Boring beats creative. Relief beats novelty. Most rollouts stall because the task the tool got pointed at was the wrong one for habit-building.
Why strategic work fails as the first AI task
Strategic work is the part of the job your employee feels valued doing. That is exactly why she does not want a machine helping her with it.
When the GM says "let's try AI on the next-vintage launch plan," the DTC director hears something different than the GM intends. She hears that the creative judgment she has been hired for is the thing the winery wants to automate first. Even if the GM never says that, the choice of task says it. The launch plan and the brand voice carry her professional identity. Asking her to feed them through an AI tool in week one of a rollout is a quiet message that the part of her work she takes pride in is the part the leadership thinks a model could do.
She will participate. She will run a few prompts. She will not form a habit, because the work itself was not bothering her in the first place, and now the AI session is adding a layer of self-consciousness on top of work she was perfectly happy doing manually. The tool gets quietly demoted to a thing she opens once a week so the GM sees activity.
Compare that with what happens if the GM points the same tool at the four-paragraph reply she writes to every club cancellation. That reply was already on her list of things she would pay someone to take away. She does not feel ownership over the cancellation reply. She feels relief when it gets shorter. The AI session stops being a performance and starts being a Tuesday-afternoon assist. The habit forms because the underlying frustration was already there.
Relief is the motivator that lasts past week three
There is a moment around week three of every rollout where the novelty of a new tool wears off. Whatever the GM said at the kickoff meeting has faded. The employee is in her normal week, with her normal task list, and she is making a decision about whether to open the AI tool or skip it.
If the task on the table is the one she complains about every week, the math is easy. Twenty minutes of email-writing she dreads, or three minutes of editing what the model drafted. She opens the tab. That is a habit.
If the task is the creative work she takes pride in, the math runs the other way. The GM is not standing over her shoulder. Writing the launch plan herself is the part of the job she enjoys. The tab stays closed.
Novelty pulls people through week one. Relief is what carries them through week eight. Microsoft's Work Trend Index research has the habit-formation threshold at three uses a week for seven to eight weeks. The reason the threshold lands at eight weeks and not at two is that the first two are easy on a new tool, and the next six are work. The task choice determines whether those six weeks happen.
I wrote more on the eight-week math in the 3x-a-week-for-8-weeks rule. The hated-task rule is what fills the three weekly uses. Without a task that already bothers her, the three slots stay empty and the eight weeks never finish.
What "hated" means in a winery context
The word "hated" is doing some work in the rule, so it is worth being specific about what counts.
A hated task is recurring. It shows up every week, sometimes every day. A one-off project does not qualify, even if the staffer is unhappy about it, because there is no compounding return on the first AI session. The point of the rule is that the second time the task arrives, the AI session is faster than the first, and by the tenth occurrence the staffer has built her own template inside the tool.
A hated task is medium-stakes. High-stakes work — the response to a refund request from a 20-year club member, the legal language on a new compliance disclosure — needs a slower setup with human review baked in. Low-stakes work — naming a Slack channel, picking a hex code for a banner — does not justify the AI session because the time savings are zero.
A hated task is text-heavy or pattern-heavy. The model can help with email replies, list reformatting, summarizing a stack of tasting notes, drafting a wholesaler follow-up, prepping a monthly newsletter draft from the last four blog posts, writing the internal training note that summarizes a vendor demo. It cannot help with physical tasks in the cellar or with anything that requires real-time visual judgment in the tasting room.
Five tasks that show up on the hated-task list at almost every winery I look at:
- Harvest paperwork — the standard intake forms, the certification checklists, the chain-of-custody notes that need to be re-typed in three different systems
- Club cancellation responses — the same four objections, the same five empathetic-but-firm replies, the same offer to pause-instead-of-cancel
- Wholesaler follow-up emails — the third nudge to the buyer who said she would circle back in two weeks and did not
- Monthly newsletter drafts — pulling the last four blog posts into a coherent recap with a teaser for the next club shipment
- Internal training note summaries — the half-hour you sat through with the new POS vendor, written up in three paragraphs for whoever was not in the room
None of these are the work the staffer wants on her business card. All of them are the work she would gladly hand off.
A concrete example with Not Really Wines
To make the rule specific without inventing a real client, here is what it looks like at Not Really Wines — our fictional demo winery, which we run as a working showcase of the kind of vault and workflow we set up for real clients.
Priya Sandhu is NRW's DTC director. She handles the website, the email list, and The Society club. If I asked her on a Tuesday afternoon what is on her plate, she would list a dozen things. Three of them would matter to her professionally: the May member shipment notes that go out to the whole list, the relaunch copy for the rosé page, the customer-story interview she has been trying to land for two weeks. Those are creative work. Those carry her identity.
The other nine items on her list include the cancellation replies that come in through Shopify three or four times a week. Most of them are the same conversation. A member is moving and does not want the next shipment until she settles in. A member tried the rosé and prefers the chardonnay. A member's husband does not drink and she is rethinking the cadence. Priya has written some version of each reply 30 times. She does not enjoy the work. She also cannot ignore it, because every cancellation reply is a retention opportunity, and rushing them costs the club.
The first AI session for Priya is the cancellation reply folder, with the rosé relaunch copy parked for later. Sit with her for ninety minutes, capture the four most common cancellation reasons, capture the way she handles each one when she is at her best, and build a starter folder around those patterns. The next time a cancellation comes in, she pastes the member's note, gets a draft in thirty seconds, edits for two minutes, sends.
Three weeks in, she has used the folder eighteen times. The cancellation replies are faster and — because she is no longer rushing — slightly better than the ones she was writing by hand at 5 p.m. on a Friday. That is the habit forming. The rosé relaunch comes later, after the habit is already running.
When to break the rule (rarely)
There are two cases where it is fine to put AI on creative work in week one, and only two.
The first is when the staffer specifically asks for it. If your DTC director walks in on day one of the rollout and says she wants to use the tool on her brand voice file, do that. She has self-selected past the identity problem. Her enthusiasm is the asset; the task is just the vehicle.
The second is when the creative work has a hated chore embedded in it. The monthly newsletter is technically a creative artifact, but most of the work in writing it is the chore of pulling source material from the last four weeks and getting a first draft on the page. The creative judgment is in the editing pass at the end. If the tool eats the source-pulling and gives her a draft to edit, that is hated-task work dressed up as a creative task. That works.
Outside of those two cases, leave the creative work for month three. By then the habit is real, the staffer has built her own intuition for what the tool is and is not good at, and she can decide on her own whether to point it at the launch plan. That decision lands better when she makes it than when the GM makes it for her.
The same rule applies to employee number two
Once the rollout is past week eight and the first staffer is running on her own, the question is who comes next. I wrote about that decision in why the second employee matters more than the first. The hated-task rule applies just as hard to that second person, and the temptation to break it is bigger.
The temptation runs like this. The first staffer was a frontline employee — a club concierge or a tasting room manager — and the chore she handed off was a customer-facing email. The leader looks around for who is next and lands on the marketing director, the one running the campaign calendar. The instinct is to point AI at the campaign calendar itself, because that is the part of marketing the GM finds least visible and would like more help thinking through.
That is the trap. The campaign calendar is the marketing director's identity. Skip it. Find the chore she does every week that nobody on the team can see and she would happily delete — vendor-quote summaries, board-update bullet points, something on that order. Start there. The campaign calendar will come on its own once the habit is running.
The pattern holds across roles. Jordan Miyake on wholesale at NRW would start with distributor follow-ups while parking the buyer pitch deck. Riley Park on the tasting bar would start with end-of-shift visit recaps and leave the seasonal flight redesign for month three. Marcus Chen on operations would start with the vendor-quote reconciliation rather than the cap-table model. The hated task is the entry point. The creative work shows up three months later, after the habit is built.
How to find the hated task in 15 minutes
The conversation to have with the staffer is short. Three questions, fifteen minutes, in person. Notebook, not laptop.
The first question: what part of your week do you procrastinate on the most? This question is more accurate than "what do you hate doing," because most staff will not call work they have been doing for years "hated." They will, however, admit to putting it off until Friday afternoon. The procrastination pattern is the signal.
The second question: which of those is the most recurring? You want the task that arrives every week or several times a week, not the one that only shows up in March. The eight-week habit window needs roughly twenty-four touches. A monthly task does not get you there.
The third question: if I told you that you never had to do this task by hand again, would you be relieved or insulted? This is the identity test. If the answer is relieved, you have the right task. If the answer is "I would feel like I was losing part of my job," set that task aside and ask question one again about the next item on the list. The good first task always lands on the relieved side.
That is the whole conversation. You will know the right task before the fifteen minutes are up, because the staffer will say it with a kind of resignation in her voice that the other items did not have. Build the starter folder around that one. Everything else can wait until the habit is running.
A short note on the webinar
I am running a webinar on the same play this post is part of. The hour is a live demo of how to find the hated task with a frontline staffer and how to build the starter folder around it before the conversation ends, with worked examples for the wine club concierge, the DTC coordinator, and the tasting room manager.
If your last AI rollout went sideways because the staffer never formed a habit, the webinar is built for that exact situation. Check the webinar schedule for the next date and to register.
Part of a 20-post series on employee AI adoption for wineries — see the full series under AI Adoption.