In 10 days you know what to automate. And the first processes already run.
The Process Mapping Sprint puts Luminars on the computers of up to 5 people from your team for 2 weeks. It finds the work they repeat, writes it down with the evidence, and switches on the first automations before the report meeting. You leave with a map of the recurring work, the processes already running in your tools, and a number: the hours and euros they give back every month.
The figures shown are an example, not a customer's.
You are asked to use AI. Nobody tells you where.
Every company is told to automate and to use AI. Almost none can say which 20 tasks eat the team's week, in which tools, how often. The usual answer is a discovery project: interviews, workshops, a spec document that is wrong in the details and out of date within weeks. Then a second project to build it, with an agency or IT. Months pass, the budget goes, and nobody can say what came back.
The figures show a typical project, not a quote.
10 working days. Evidence instead of interviews.
Instead of asking your team to describe their work from memory, the sprint watches the work as it happens. 5 people install Luminars on their own computers, answer a 3-minute interview about their role, then do their jobs as usual. Luminars collects the context, confirms what recurs, drafts the processes and makes them runnable. I am there for 1 short conversation with each champion and for the report meeting at the end. Week 1 asks. Week 2 confirms. Nothing to integrate, nothing for IT to do.
Everything happens inside the champions' normal working week.
Day by day. What happens, and what it needs from you.
30 minutes with whoever decides. Then 5 installs.
We pick up to 5 champions: people who do the recurring work, not people who manage it. We set the scope: which teams, which tools, what you already suspect is worth automating. Invites go out the same day. Each champion installs the app in minutes and answers Dolly's 3-minute interview: their role and the tasks they actually repeat. Context collection starts right away.
Installs in minutes. Download, sign in, 4 permissions on macOS, 1 on Windows. Done.
3 minutes with Dolly. Role, recurring tasks, apps. Luminars uses them to know what to look for.
The champion decides. Pause, excluded apps, scheduled pauses, from the first minute.
Nothing for IT. No integration, no server, no ticket.
1 conversation each. Then they show me the job.
20 minutes with each champion about how the work really goes: the exceptions, the workarounds, the parts nobody wrote down. For the tasks they already know are worth it, they press Record and do the job once, narrating as they go. Luminars writes the process while they work. By Friday the first drafts exist.
20 minutes, not a workshop. 1 conversation per champion, at a time they choose.
Record, work, stop. The floating bar over their own tools. The draft is written when they stop.
Do it, show it, or just tell it. All 3 count, and a told step is marked as told.
The why, in their words. Dolly asks about the exceptions in a chat, 1 question at a time.
The app runs in the background. Luminars confirms what recurs.
While people work as usual, Luminars looks for the tasks each champion said they do, marks what it actually saw, and finds the ones nobody mentioned. Every suggestion carries its evidence: "said weekly, seen 4 times, about 40 minutes each". Each process becomes a readable draft, and at least 2 are switched on and run on the champion's own computer before the report.
Said, and seen. What the champion said, beside what Luminars saw.
The ones nobody mentioned. The workarounds nobody wrote down show up here.
2 running before day 10. Switched on, run, hours counted.
Every run is readable. A plain-language transcript you can stop at any step.
1 meeting. Everything on one screen, with the numbers.
We go through it together: the map of what your team really does, the roadmap ranked by hours, the processes that already ran and what they returned, the inefficiencies we found, and what to do next. Not a deck about what could be automated: a record of what already was, a number for what it gave back, and a ranked plan for the rest.
The figures shown are an example, not a customer's.
Map. Every recurring task, by person and by frequency, with the evidence.
Roadmap. Ranked by hours per month and by how easily each task automates.
ROI. Hours and euros per month, per process, measured on the runs.
Next. What to switch on, what to fix, what to leave alone.
25 minutes to see it on 1 of your own processes.
A short call: Luminars live on a process you choose, and a clear answer on whether a sprint makes sense for your team.
What you have on day 11. Not a document: processes that run, and a number.
Every item traces to something a champion said or Luminars saw. Nothing is an estimate from a workshop.
A map of how your team works, and what to automate first.
The map shows your team's recurring work by person and by frequency, with the evidence beside each row. The roadmap is the same map, ranked: which processes take the most hours, which repeat most, which automate cleanly, and in what order to switch them on. Beside it, what the map shows on its own: the same work done by two people, compared side by side, the procedures that hang on one person, and the ones nobody has documented yet. You know what to automate first and what to fix by hand.
The figures shown are an example, not a customer's.
Every row has evidence.
Ranked by return.
The same work, done by two people.
What to skip.
Your processes, mapped and already running, in the tools you use.
A mapped process is not a page for a developer. It already runs. At least 2 are switched on during the sprint and run on the champion's own computer, in the apps your team already uses, with a person in the loop wherever you want one. The rest are drafted and 1 click from running. Nothing was integrated, nothing was built, and nothing needs IT to keep running.
In your tools.
A human step wherever you want one.
1 click from running.
Yours to take away.
A real number for the hours and money your processes give back.
For every process that ran, the report puts the time it took by hand next to the time it took to run, multiplied by how often it happens and by what your people cost. That gives hours per month and euros per month, per process and for the team. For the processes still on the roadmap, the same arithmetic gives a projection, marked as one. You leave the meeting knowing what the sprint returned and what the rest of the roadmap is worth.
The figures shown are an example, not a customer's.
By hand vs run.
Hours per month.
Euros per month.
The roadmap, with a value.
The console keeps counting after the sprint ends.
The number in the report is the first data point, not the last. The console shows the hours returned and the runs behind them, for as long as the processes run. Whoever decides can open it any Monday and see what the automation gave back, without asking anyone for a status.
The figures shown are an example, not a customer's.
Week after week.
Per process.
From the runs, not from estimates.
For whoever decides.
What is repetitive work costing you today?
Repetitive work, per year
€160.000
≈ 3.200 hours across 10 people
This is what the sprint maps. The report says how much of it is worth automating; the console shows how much of it came back.
Weeks, not months, and a number at the end.
A discovery project ends with a document. The sprint ends with processes running and a measured return, 10 working days after the kickoff.
The two paths as an illustration, not a benchmark.
Without the sprint
- Weeks of interviews and workshops.
- A spec document nobody trusts.
- A second project to build the automations, with an agency or IT.
- New tools to buy and learn.
- No number for what it returned.
With the sprint
- 10 working days.
- Discovery in the background from day 1.
- Processes written from evidence, not memory.
- The first ones running before the report.
- Hours and euros returned, in the report and in the console.
See where AI pays, in 10 days.
Book a call- 10 days days
- from kickoff to the report meeting
- 2++
- processes running before the report
- 1 hour hour
- of each champion's time across the 2 weeks
5 people who do the work, and decide what is collected.
A champion is the person in finance who reconciles the invoices, the person in client ops who onboards new accounts. Each gives about 1 hour across the 2 weeks: the interview, 1 conversation with me, 1 or 2 recordings. The rest is their ordinary week, with the app running in the background. Each champion controls their own collection from the first minute: pause, excluded apps, scheduled pauses, and a retention window after which screen and audio are deleted. Your company sees processes, runs and hours returned. Never screens, never audio, never keystrokes.
For the person who decides: the kickoff, the report meeting, and a message or two in between. About 2 hours in total.
How we handle dataYou control collection.
Your company sees processes and runs. In this beta, Help improve Luminars separately uploads original recordings to Luminars for evaluation; you can turn it off in Settings → Data.
Signed by Luminars, notarised by Apple.
Every macOS release is signed with a Developer ID and notarised by Apple, so macOS checks it before it runs on your computers.
Your workspace is hosted in the EU.
Your workspace is hosted in Ireland and run by an Italian company. AI requests use the provider selected in Settings.
Keep running what worked. No IT project required.
The processes that proved themselves keep running as they did: on your team's computers, in 1 click, with the hours returned visible in the console. Adding a person or a process takes minutes, not a ticket. The last page of the report proposes a plan for the team at a price we agreed before the sprint started; if you would rather run the processes in your own stack, take the exports into n8n or any agent tool. Either way the map, the roadmap and the processes are yours.
For teams of 10 to 50 people with recurring work.
Back office, client ops, e-commerce, finance and admin.
A lot of weekly work in mail, spreadsheets, a CRM and a few SaaS tools. An owner or ops lead who wants to know where AI actually pays before spending on it.
A team of 2, or work that is different every time.
Mostly creative or engineering work, or a company that already has every process documented, automated and measured. If that is you, download the app and try it on 1 process instead.
Questions before a sprint.
What do we need to prepare?
Nothing. Up to 5 people who do the recurring work, about 1 hour each across 2 weeks, and 1 kickoff call with whoever decides. No documents, no process list.
How is the return calculated?
For each process that ran: the time it took by hand, measured during the sprint, minus the active time it takes to run, multiplied by how often it happens and by the loaded cost you give us. The report shows the arithmetic per process; the console keeps counting afterwards.
Does it need IT or access to our systems?
No. Nothing is integrated for the mapping: the app watches the tools your team already uses, on their own computers. A run may ask to connect a tool the first time it needs one. No developer, no server, no ticket.
What do the champions agree to?
Each champion installs the app on their own computer and stays in control of it: pause, excluded apps, scheduled pauses, and a retention window after which screen and audio are deleted. The company sees the processes and the runs, never the recordings.
What if nothing is worth automating?
You still get the map with the evidence, the roadmap and the inefficiencies, and we say so plainly in the report meeting. No plan is proposed.
Can we stop halfway?
Yes, at any point, and any champion can pause or stop collection on their own computer at any time. Whatever was drafted, run or exported by then stays yours.
How much of my time does it take?
The kickoff, the report meeting, and a message or two in between: about 2 hours for the person who decides.
What does it cost?
We agree the sprint fee and the price of continuing before the sprint starts, on the call. There are no surprises on the last page of the report.
Start with a 20-minute call.
We look at Luminars on 1 of your own processes and decide together whether a sprint fits.