One method
for getting AI to produce repeatable, reliable results you can put your name on.
The first cohort filled in days. The next one opens soon.
Six weeks, live, for senior FP&A. You leave with one workflow you designed, built, deployed, and evaluated yourself — on your data, in your context.
Already know you’re in? Cohort 1 filled before its advertised close date, so I’m opening a few priority enrollment places ahead of the launch — you’re in before it opens to everyone, plus a 1:1 strategy call with me before the cohort starts. See the priority details.
You’ve tried it. You’re still doing the work by hand.
You have used ChatGPT. You have tried Copilot. You have saved a few prompts. You have watched a tutorial that made it look simple.
And still, you are back to the same spreadsheet. The AI experiments sit in a tab. Nothing is repeatable. Nothing is testable. Variance commentary is still on your plate, and you are still the person writing it by hand.
The problem is not AI. The problem is that nobody has shown you how to design the work around it.
Three things break the same way for almost everyone.
You have tried ChatGPT. You have tried Copilot. You have saved a few prompts. Most of it does not last more than a few weeks.
The same prompt gives a different answer.
First run is brilliant. Second run is different. Third run you stop trusting it. Finance work has to reproduce. Without a method, yours never will.
Numbers that were not in your data.
Either you have caught one, or it landed in a report with your name on it. There are real techniques to stop this happening. Most courses do not teach them because they are not glamorous.
Workflows that quietly degrade.
You finally find something that works. Three weeks in it begins drifting. You cannot tell why. Most teams have not even named this problem yet.
The work would have taken me about 2 days. From start to finish, Claude knocked it out in less than 2 hours and I was able to verify the accuracy the next morning. Prior to your course, I wouldn’t have known how to achieve those results.
Accelerator Cohort 1
Everyone tells you to write a better prompt.
So you do. You collect the good ones, you paste in more context, you add rules at the bottom. It works, right up until the moment you need it to work the same way twice.
A prompt is an instruction. It is not a process.
However well written, a prompt only exists for as long as you are sitting there typing it. Nothing around it is fixed, so nothing around it is reliable.
- You repeat yourself. Every month, same context pasted again, same corrections typed again. The knowledge lives in your head and your clipboard, not in anything that runs.
- It still varies. The same prompt on the same data returns a different answer tomorrow. A longer prompt narrows the range. It does not close it.
- It still invents numbers. Asking a model nicely not to hallucinate is not a control. The maths has to happen somewhere the model cannot reach.
- It quietly rots. The model updates, your prompt does not, and the output drifts. You will not notice until someone asks why the figure moved.
- It does not survive you. Nobody else can run it, audit it, or pick it up when you are on leave.
Prompt engineering is a skill worth having. It is simply the wrong layer to solve this at — you cannot fix a process problem with a better sentence.
Intelligent workflows. The prompt becomes one step inside a process that also decides what runs before it, what checks it, and what you review. Designed once, then run again and again.
The process, the steps we follow that make our workflows repeatable, was the missing key to using AI.
Accelerator Cohort 1
One workflow. Each force where it belongs.
Monthly variance commentary, built the way you will build yours. Notice where the model actually sits: everything feeding it is deterministic, and everything after it passes through you. Hover any step to see which force governs it and which of the three breaks that placement prevents.
The result: a board pack you can defend.
A board pack you can put your name on.
Ready to design real Finance Workflows you can actually put your name on?
Introducing the FP&A Tactic Accelerator.
A 6-week live cohort. A prep week with a live orientation kickoff, then five core sessions, side by side with me. By the end, you have one workflow you designed, built, deployed, and evaluated yourself, using your own data in your own context.
This is how the Accelerator will help you create and implement AI Intelligent Workflows.
Week one is a prep week with a live kickoff. Then we move through five building blocks together, one live session each. Click any block to see what happens in it.
Click to see the schedule of live calls +
This is the technical, hands-on part of the program, taught step by step on real FP&A examples. You will learn how to design workflows that do exactly what you need, how to build them in Claude, and how to optimize them so they run faster, use fewer tokens, and return the repeatable results you can put your name on. No coding skills required. I walk through every detail, so the practical side and the technical side both belong to you. This is not theory. You leave able to do the work.
You will want to build workflows for everything, but you never have to start from scratch.
As a bonus, you get Tactic OS — the skill library I have spent two years building to run my own FP&A work. Not a template pack. A working library that already knows how to do the job.
my own work
FP&A skills
and copy from
ends, forever
Every one of those is a skill you can call, chain into a workflow, or take apart to see how it works. Inside the cohort you get the full library, and you build your own workflow on top of it rather than from an empty file.
You keep all of it. The library does not expire with the cohort. Every skill you use, extend, or rewrite stays yours, and it keeps growing as I keep building for my own work.
Plus a Workbench for every workflow I have built myself, Monthly variance commentary
Beyond the library, you get a Workbench for each of several typical FP&A workflows. A Workbench is the whole build handed over: the workflow itself, sample data you can run it on, and a short guide to adapting it to your own context.
You are never alone running them.
A Workbench hands you the build. The Lab is where you go to chat about it.
Lifetime access to the Accelerator Lab.
A brand new community I am building for lifetime exchange between senior FP&A professionals putting AI to work on real finance problems. Peer review. Shared wins. Stuck-point threads. Ongoing discussion as new tools and models land. Access does not expire.
The people around you keep using the method on their own work. When you hit something new six months or two years from now, the same people are still in the Lab with you, working on the same kind of problems you are.
What they built, in their words.
Last Friday, my CFO requested a research project where I would have needed to go through 9+ years of SEC filings. The work would have taken me about 2 days. Claude knocked it out in less than 2 hours. Knowing how to use style guides and prior work product to create repeatable skills in Claude Code is a true accelerator.
WFP
I had used AI before the accelerator, but I couldn’t see how to bridge the technical gap from using it as a chatbot to the output others talked about. I have identified processes that could save me upwards of 10–15 hours a month across all clients. That allows me to stay competitive on pricing while taking on more clients.
MoneyPath · Fractional FP&A
As soon as I implement what I’ve learned in this course, I’m certain I’ll recover at least 30% of my time from low-value-added activities.
OUTTHINKER S.r.l. · Consulting
I never felt that prompt-based approaches fully reflected the deterministic, controls-driven nature of finance. The Accelerator gave me the missing piece: a practical methodology for designing structured workflows that combine deterministic processing with probabilistic reasoning, supported by human judgement.
Cohort № 1
I’ve just run a board deck — all formatted, charts looking good, KPIs the whole lot. Once I gave the workflow the data, I had my deck in 5 minutes. This would have taken me hours of fighting trying to get the layout to work.
Sharps Fabric Printers Ltd · Management Accountant
The next cohort price.
$1,997 one-time
Everything inside the program. Live, recorded, and yours to keep.
Yes, I’m inWhat’s included
- Live orientation + 5 live core sessions
- Practice Workflows companion course
- Tactic OS skill library + Workbenches
- Four exclusive masterclasses
- Lifetime access to the Accelerator Lab
- All recordings, yours forever
This price is for this cohort only. Future pricing will be higher.
Webinar fast-action bonus. The first 10 enrollments tonight (or until midnight ET, whichever comes first) get a 1-hour 1:1 strategy call with Carolina, plus the 30-Day Challenge: solve a real problem with what you learn, show it on your call, and get one month of your Claude subscription reimbursed (up to $100). New or existing subscription, both count.
Limited seats. Live feedback from me on your own workflow. Enrollment closes TBD.
Enrollment is closed. The next round opens soon.
Ready before the launch?
The next cohort opens publicly on September 8. I’m opening a few priority enrollment places before then — you’re in before it opens to everyone, plus a 1:1 strategy call with me before the cohort starts. Join below and I’ll send you the details.
What priority access gets you
- Early notice when next cohort dates are confirmed
- Priority enrollment before the public launch
- First read of the updated program details
Four standalone AI masterclasses, coming soon.
Keep an eye out for these. Coming soon, we will explore other cool and important topics around AI in finance. AI Masterclasses coming up:
AI Implementation Roadmap
Whether you work inside a large company, run your own consultancy, or use this for your own personal practice, putting AI to work takes best practices and a real plan. In this masterclass we lay out a clear roadmap for implementing what you have learned, in a big team, a small team, or in your own daily practice.
AI Governance, Security and Compliance
How to prevent problems. How to keep an audit trail when you are using AI. How to know when not to use AI at all.
Mini Apps for Finance with AI
How to build small, useful mini apps for your finance job using Claude Code. What to look for. How to package an app. How to deploy it. How to review it. With worked examples.
Finance Storytelling with AI
How to use AI to produce beautiful charts and beautiful presentations, and how to bring elements of finance storytelling into the skills you build.
Enrollment closes TBD.
Limited seats. This pricing is for this cohort only.
Enrollment is closed. The next round opens soon.
Take a priority place ahead of the next public launch.
Take a priority placeBuild it with me, 1:1.
A few times a year, I open a small number of spots to work directly with me on AI workflow implementation and FP&A: individuals who want a partner in the build, and teams putting AI to work on their real processes.
This is hands-on and personal. We design, build, and ship together, on your actual data and your actual problems. Because it is just me, the spots are few and I am selective about fit.
After you apply, I personally review every submission to make sure it is the right fit. If it is, I will be in touch to talk next steps.
Carolina Lago
I’m the main part of my workflows.
I’m Carolina. 25 years in FP&A across Latin America, North America, and Europe. I rebuilt my own work around AI over the last two years, and what I figured out is the method I’m teaching inside this cohort.
The same method that led me to speak on international stages like the Financial Modeling Global Leaders Council, the Association of Finance Professionals, and the Global Excel Summit, and that I’m now using to train corporate teams all over the world.
You don’t need another AI course.
You need a method, and a workflow you can put your name on.
Here is what other students have asked before joining.
How much time per week?
About 4 to 6 hours during the cohort. Two live sessions of roughly 75 minutes each, plus workflow-build time between sessions through the Practice Workflows companion course. Most students find the pace demanding but manageable alongside a regular FP&A job.
What if I cannot make a live session?
Every session is recorded. You can catch up in your own time and bring questions to the next one. The community stays active between sessions for async support.
I am not technical. Is this for me?
Yes, if you already do senior FP&A work. The method is built for finance practitioners, not engineers. If you can run a forecast and write variance commentary, you can run this.
Do I need a paid ChatGPT or Claude subscription?
The method is built and demonstrated on a Claude Max subscription, which is what I recommend you have for the cohort. Because of usage limits, I do not currently cover ChatGPT or Copilot. You can apply the method to those tools later once you have the foundation, but the live work, the demos, and the Tactic OS library all run on Claude.
What if my company blocks external AI tools?
This cohort specifically covers Claude. If your company blocks it, you will need a personal Claude Max subscription and you should plan to work on a personal computer during the cohort.
I am direct about this on purpose. You are not going to wait for your company to approve your growth.
Is this for beginners or for experienced FP&A?
The cohort is built for senior FP&A practitioners with 3 to 15 years of experience. You should already do the work. The cohort assumes you have a real workflow you would like to bring in and rebuild around AI.
Can I get a refund?
Yes. If after the first two live sessions you decide the cohort is not for you, write to me and I will refund you in full. After Session 3 we are deep in the build and refunds no longer apply.
When does the cohort start?
Live orientation kicks off TBD at 6pm, and Design begins TBD. The five sessions run through TBD — see the schedule above. Enrollment closes TBD.
What happens after the cohort?
You keep the recordings. You keep access to the Tactic OS skill library and the Workbenches. You keep access to the four exclusive masterclasses as they release. You stay in the Accelerator Lab for life. Future cohorts and standalone masterclasses are offered to alumni at alumni pricing.
Will the four bonus masterclasses be included in future cohorts?
I have not decided yet. They will be sold individually after release. Cohort members get them as part of joining now.
Will there be future cohorts?
Yes, but not at this price and not with the same bonus stack. Future cohorts will be priced higher and the bonus structure will be different.
How is this different from other AI courses?
Most AI courses are for general knowledge workers. Most finance courses are for modeling fundamentals. This is the intersection. AI specifically for FP&A, taught by a practitioner still running the work. The Three Forces lens is my own. The Tactic OS library is what I built for my own work. The cohort is where you build yours.
TBD · Enrollment closes · Limited seats
The rules of FP&A are changing. I’ll help you rewrite them.
Build one workflow you can put your name on.
Yes, I’m in Save my seat at the launch event