Choose done-for-you when the system is complex, customer-facing or rarely changes, and you're happy to pay someone to maintain it. Choose done-with-you when the workflow will keep changing, someone on your staff can own it, and you'd rather not pay for every tweak. The deciding question is who fixes it at four o'clock on a Friday.
Done-for-you means an expert designs, builds and usually maintains the system while your team simply uses it. Done-with-you means the expert builds it alongside a named person on your staff, who does some of the work, learns how it fits together and takes it over. Both can use the same tools and cost similar amounts up front. The difference shows up between month three and month twelve, when things need changing and something eventually breaks.
What each model leaves you holding
| Done-for-you | Done-with-you | |
|---|---|---|
| Time to go live | Faster: the expert works at full speed | Slower: explaining and pairing takes time |
| Your staff time during the build | Low: briefing, examples, testing | Higher: a few hours a week from the person taking over |
| Cost shape | Build fee plus ongoing support | Smaller expert fee plus your staff's hours, lower ongoing cost |
| Who can change it afterwards | The expert, usually | Your named owner, for everyday changes |
| Documentation | Written for the expert's colleagues unless you ask otherwise | Written for your owner, because they helped create it |
| If the expert disappears | You need a new expert to learn it | Your owner keeps it running; harder changes wait |
| Best for | Complex, customer-facing, stable systems | Internal workflows, prompts and templates that change often |
Neither is the premium option. Done-for-you buys speed and expertise on the hard parts; done-with-you buys independence. Most small businesses end up needing some of each, and the useful skill is telling which parts of a project belong where.
Systems that suit done-for-you
Hand the whole job to an expert when getting it wrong is expensive and changing it is rare. The signs: several systems joined together, money or customer data moving through it, customers seeing the output directly, and a design that, once right, can stay the same for months.
An illustrative family butcher with four shops wants online orders routed automatically: each order goes to the right shop's pick list based on the collection point, delivery orders go onto the driver's route sheet, and payments are matched to orders in the accounts each evening. That's four systems, real money and customers waiting at a counter. The staff member who might take it over has never built an automation, and a mistake on a busy Saturday means a queue of people whose orders aren't ready. This is done-for-you territory: build it properly, test it on a month of past orders, document it, and pay for support.
Systems that suit done-with-you
Keep the job in-house, with expert help to set it up, when the content changes often and the person who knows the content is on your staff. Prompts, reply templates, weekly price lists, product descriptions and internal email sorting all fit, because the rules behind them change with the season, the suppliers and the specials board.
The same butcher's second job is exactly this kind. Every Thursday the shop manager writes the weekend specials for the website, the shop windows and social posts, and sorts a pile of supplier emails about price changes and delivery slots. An expert can set up the AI prompts, a template for the specials and a flow that labels and summarises supplier emails. But the specials change every week and the supplier list changes every month, so the person who has to be able to edit the prompts and rules is the shop manager, not an outside expert on an hourly rate. DIY or done-with-you for AI marketing covers the marketing side of this in more depth.
A four-shop butcher splits the work
Here's how the butcher's two jobs might be costed over the first year. Expert time is priced at goLance's published band for senior AI consultants, $120 to $200 an hour; the shop manager's time at $24 an hour and the owner's at $50. All figures are illustrative.
| Line | Order routing (done-for-you) | Specials and supplier emails (done-with-you) |
|---|---|---|
| Expert time to build | 40 hours: $4,800 to $8,000 | 20 hours: $2,400 to $4,000 |
| Staff time during the build | Owner, 8 hours: $400 | Shop manager, 3 hours a week for 8 weeks: $576 |
| Upkeep for the rest of the year | Support allowance of 2 hours a month for 9 months: $2,160 to $3,600 | Manager, 1 hour a week for 44 weeks: $1,056; expert check-in each quarter: $480 to $800 |
| First-year total | $7,360 to $12,000 | $4,512 to $6,432 |
| Who changes it | The expert, within the support allowance | The shop manager, weekly |
Two things stand out. The done-with-you job isn't free just because the butcher's own staff do much of it: the manager's hours are real, and the quarterly check-in is worth keeping. And the done-for-you job's upkeep is a real line, not an afterthought. If the butcher had tried to make order routing done-with-you, the manager would have needed far more than 3 hours a week to learn four systems, and a Saturday failure would have landed on someone still learning. If it had made the specials done-for-you, every weekly change would have been an email to the expert and a line on the next invoice.
What done-with-you asks of your side
Done-with-you only works if the person taking over has the time and the brief. Write a short role card before the work starts. The butcher's, filled in:
- Owner: shop manager, main shop. Backup: assistant manager.
- Time: 3 hours a week during the eight-week build, protected on the rota; 1 hour a week afterwards, Monday mornings.
- During the build: supply past specials and supplier emails as examples, test each step, write the "how we do it" notes in their own words.
- Afterwards: edit prompts and templates when the specials or suppliers change, check the flow's log each Monday, raise anything odd with the expert at the quarterly check-in.
- Not expected to: build new automations from scratch, connect new systems, or fix failures in the order-routing system, which stays with the expert.
- Skills to have by week eight: change a prompt and test it on three examples; pause and restart the flow; read an error message and decide whether to fix it or escalate.
Here's how the first four weeks of the butcher's done-with-you build might run, with the shop manager's part in each:
| Week | Expert | Shop manager |
|---|---|---|
| 1 | Walks through how specials and supplier emails are handled today; sets up the shared prompt document | Brings the last eight weeks of specials and 30 supplier emails; writes down the three rules they always follow |
| 2 | Drafts the specials prompt and template; shows how each part works | Runs the prompt on last week's meat list; marks what's wrong; edits the prompt with the expert watching |
| 3 | Builds the supplier email flow: labels, summaries, a daily digest | Tests it on the 30 saved emails; decides what counts as urgent |
| 4 | Adds failure alerts to the manager's inbox; starts the handover notes | Pauses and restarts the flow alone; writes the "what to do if" section of the notes in their own words |
By week four the manager has touched every part of the system at least once. That, more than any document, is what makes the handover real.
The weekly job the manager ends up owning looks like this. The prompt lives in the shared document, and the manager pastes in the week's list:
Write this weekend's specials for our website and shop windows.
Use only the cuts, prices and notes in the list below. Do not
invent prices, origins or awards. Keep each item to one line,
plus a two-sentence intro in our usual friendly tone.
LIST:
[paste this week's specials list from the pricing sheet]An illustrative output might open: "This weekend at the counter: slow-roast pork shoulder at $9.80 a kilo, perfect for Sunday, and award-winning dry-aged sirloin at $38 a kilo." The manager's read-through before posting catches the problem: nothing on the list mentions an award. The fix takes a minute, because the manager wrote half the prompt: add "never describe anything as award-winning unless the list says so" and rerun. Under a done-for-you arrangement, that same fix would have been an email to the expert and a wait.
The "not expected to" line matters as much as the rest. Done-with-you fails when the owner is quietly expected to become an automation specialist on one hour a week. For a business without anyone technical at all, implementing AI with no tech team covers how to choose that owner and support them.
Upkeep: the line that usually decides it
Automations fail quietly, and the platforms' own rules show why someone must be watching. Zapier automatically pauses a Zap once 95% of its runs have errored over seven days, and it doesn't send error emails when an error handler has run. Power Automate switches off a flow after 14 days of continuous failure. Make renamed its error handlers in September 2026 (Ignore became Skip and Break became Retry), so a handover note written in the summer may already use the old names.
Here's how that plays out with done-for-you. In this illustrative case, a coffee roaster pays an agency to build a subscription reminder flow. Months later the online shop changes the name of a field, the flow starts failing on every run, and after a week the platform pauses it. The failure alerts go to the agency's inbox, where they sit behind other clients' work. The roaster finds out when subscribers start asking why their reminder emails stopped. Nothing about done-for-you caused this; the gap was that nobody at the roaster could see the alerts. Whichever model you choose, make sure failure alerts reach someone inside your business, and that they know whether to fix the problem or call for help.
That's also the strongest argument for done-with-you where it fits. A shop manager who checks the log every Monday will spot a paused flow within days. Auditing the Zaps and scenarios nobody owns is a useful exercise whichever model you pick, and AI maintenance costs after go-live shows how to budget for the upkeep line honestly.
Telling a genuine done-with-you offer from a relabelled one
"Done-with-you" has become a popular label, and not everything sold under it matches the description. Three common mismatches:
- A course with a new name. Group sessions on how to use AI tools, with no build on your own data and no one-to-one work with your owner. Useful, but it's learning, not implementation.
- Done-for-you with a video at the end. The expert builds everything alone, then records a walkthrough. Your owner hasn't touched it and won't be able to change it.
- A template install. A generic workflow dropped into your accounts with your logo on it, and little adaptation to how you actually work.
Four questions sort them out. Will my named person build parts of it themselves, with you watching? Will we use our own real examples from the first session? What will my person be able to change on their own by the end, listed specifically? And what's in writing at handover, and who writes it? A genuine done-with-you offer answers all four without hesitation, and the answer to the third is a list, not "anything they like".
You can also tell a lot from a sample of the documentation. Here's the same step described two ways:
Written for another expert: "Step 4: text parser on body_plain, regex for price lines; output mapped to Sheets row, col F. On error: 3 retries, then alert to #ops-agency channel."
Written for your owner: "Step 4 pulls the new price out of each supplier email and puts it in column F of the Prices sheet. If a supplier changes how their emails look, this is the step that stops working. You'll get an email titled 'Price flow failed'. Check the latest supplier email; if the layout has changed, forward it to the expert. Don't edit this step yourself."
Both are accurate. Only the second helps the shop manager on a Friday afternoon, and notice that the first sends its alerts to the agency's own channel. Ask to see a page of real handover notes from a past project, with client details removed, before you sign either kind of engagement.
For done-for-you, the equivalent questions are about ownership and exit: whose accounts, what documentation, what support costs, and what happens if you part ways. The AI consultant handover checklist covers those in detail.
Switching models part-way through
You don't have to choose once and for all. Two switches work well for small businesses:
- Done-for-you now, handed over later. Useful when you need the system fast but want independence eventually. Agree up front that the documentation will be written for your staff, not for the expert's colleagues, and price a handover phase in the original quote so it isn't a surprise later. Six months in, when the system is stable, your owner pairs with the expert for a few sessions and takes over the everyday changes.
- Done-with-you, with the hard part done for you. Your owner builds most of a workflow with guidance, and the expert builds the one fiddly connection alone, then explains it. This keeps the expert's hours low while making sure the part your owner couldn't manage is still solid.
The switch that tends to go badly is the unplanned one: a done-for-you expert leaves, and a staff member is asked to take over a system they've never seen, from notes written for someone else. If there's any chance you'll want to bring a system in-house, say so in the first conversation. It changes how the expert builds and documents it, and costs far less to plan for than to fix.
The practical test for your own project is short. List each part of the system, and next to each write who you want to be changing it in a year's time. Where the answer is an outside expert, that part is done-for-you. Where it's someone on your staff, it's done-with-you, and that person should be in the room from the first week.
Further reads
- AI Consultant vs Agency vs Freelancer: Which Should You Hire? — Which kind of supplier fits each model.
- Who Owns the AI Workflows a Consultant Builds for You? — Ownership terms that matter most for done-for-you.
- How to Get Staff Buy-In When You Introduce AI — Helping the person taking over actually want the job.
- No-Code AI Automation for Small Businesses: Where to Start — The tools a done-with-you owner will be working in.
- Is an AI Consulting Retainer Worth It After Your First Project? — Whether ongoing support after a build pays its way.
- How to Monitor AI That Talks to Customers: Hand-Offs and Errors — Watching customer-facing AI once it's live.
- What an AI Consultant Can't Do for You, and What You Must Own — Seven responsibilities that stay with the business when you hire AI help, an ownership card for every automation, and promises no consultant should make.
- What AI Implementation Actually Involves for a Small Business — The six pieces of work behind a small-business AI implementation, with rough hours for each and one workflow followed from baseline to handover.
- Fixed-Scope vs Hourly AI Consulting: Which Protects Your Budget? — How fixed-price and hourly AI consulting quotes shift risk, with a worked catering example, a cap clause and questions to ask before you sign.
- AI Strategy vs AI Implementation Consultant: Which Do You Need? — What an AI strategy consultant and an implementation consultant each hand you, a decision table for eight common situations, and a worked garden centre example.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Zapier help pages (auto-pause at 95% errors over 7 days; error handler notifications), Microsoft Power Automate documentation (flows turned off after 14 days of failure), Make help pages (error handler names, updated 8 September 2026); goLance AI consultant rate guide (2026) for the rate band used in the sums.