TRAINING

Become an AI power user — not a spectator

Applied AI training for those who need to use it, not just understand it. Built by someone who implemented AI in his own operation, not by someone who read about it. The goal is not to learn what AI is — it's to leave knowing how to automate your own work.

Who it's for

Each profile arrives with a different pain. The training speaks each one's language.

Lawyers and independent professionals

The billable hour is your most expensive asset — and it's draining away on repetitive tasks.

Document review, research, first drafts of petitions and opinions. Without compromising confidentiality and without outsourcing legal reasoning. AI that gives billable hours back to your day, not that replaces your judgment.

See track →

Business owners and C-level

You need to decide about AI in the company without knowing, in practice, what it can handle.

Understand AI at the depth of someone who will hold it accountable for results. Leave with the criteria to separate vendors who deliver from vendors who demo, and with the first automations running in your own management routine.

See track →

Board members

The AI agenda reaches the table and you need to ask the right questions.

Vocabulary and criteria to play the board member's role facing the AI agenda: distinguish architecture from standalone tools, hype from implementation, and demand a decision with an owner and a deadline instead of one more presentation.

See track →

In-company teams

Your team uses AI in an uncoordinated way — each with their own prompt, no one with a method.

Closed training for the team, with cases from your own operation. Instead of thirty people improvising with ChatGPT, a team with a shared method, common standards and a productivity gain that shows up in the process.

See track →

TAILORED TRACK

Don't see your group here?

The four above are examples. The same course adapts to any group's reality — doctors, electrical or mechanical engineers, investors, sales or marketing teams. We rewrite the cases, the language and the exercises around your group's real pains, because that's how the content turns into practice the next day. This is the very method we used to build the tracks above. We build tailored closed cohorts — ask us about the terms.

Talk about my group's track

Lawyers were the first profile we tested in depth, because the equation is clear: the firm's most expensive asset is the time of those who bill by the hour, and much of it is consumed by tasks AI does well — as long as it's run with method and confidentiality. The focus is not to replace the lawyer. It's to give hours back.

What you walk away able to deliver — and the time you get back

Contract review with clause and page cited

You load the contract and get the risk points flagged with the exact clause and page — traceable, not a vague summary.

Weeks → days

Contract opinion in a table

Risks, legal grounds and litigation probability organized in a table, ready for your review.

1–5h/week back

Verified case-law research

Precedents mapped with the source flagged for you to check, without mixing facts across cases.

~70% less time

Due diligence with your own checklist

A reusable workspace that applies your own review checklist to every new transaction.

12h → 3.5h per deal

A single command for your most repeated task

Contract review, hearing prep or precedent research triggered by one phrase, always to the same standard.

80% less effort

Opinions in your firm's voice

In your tone and format, without the “machine” feel — ready for you to review and sign.

Approved on first draft

Automated morning briefing

Your emails and the day's calendar summarized first thing, with deadlines highlighted.

30–60 min → 3–5 min

Full pre-hearing briefing

Emails, calendar and case documents cross-referenced into a summary — and even the slides — before the session.

2–3h → 20 min

Contingency and case-tracking spreadsheet

With formulas that recalculate on their own and the risk map of your case portfolio.

4–6h → ~20 min

Daily monitoring across multiple courts

Your cases' updates consolidated into a spreadsheet, with deadlines flagged.

45 min/day → 5 min

Each of these deliverables is built with you across the Dominando IA course — from concept to automation, with method and professional confidentiality. The rule is always the same: AI drafts and researches; you review and decide.

You don't need to become technical, you need to decide well. This track puts AI on YOUR management desk: you learn to tell a vendor who delivers from one who only runs demos, bring a well-grounded recommendation to the board, analyze contracts and proposals before the meeting, and automate parts of your own decision routine — always under the principle that AI drafts and researches while you review and decide. The result: less guesswork, more free time for what only the owner/executive can do.

What you'll start doing and producing in running the company

Board-ready AI adoption recommendation

A 3-minute executive script (up to 450 words) with 1 market data point with source, 1 real-world case with results, 1 material risk, estimated cost, 12-month ROI and 3 milestones — exactly what the committee needs to approve (or reject) an AI investment.

Decision material ready in ~25 min, replacing guesswork with well-grounded reasoning at your next board meeting.

Criteria to tell a vendor who delivers from one who only runs demos

Practical criteria for selecting model and plan by task (Haiku for triage / Sonnet for 80% of the day / Opus for deep analysis; Free to test, Pro for daily use, Team ~R$1,500-1,800/month for 10 people, Enterprise only with compliance) — so you can challenge the cost and scope of any AI proposal that lands on your desk.

Avoids paying for Enterprise/Opus where Sonnet does the job; Opus pricing dropped 67% in 2 years — you choose by the task, not by the salesperson's reputation.

Contract and proposal analysis before the meeting

Comparative analysis of real documents (proposal + addenda, contracts, M&A) with a table of risk clauses, inconsistencies and points of attention citing clause and page — using up to 1M tokens (~1,500 pages) in a single conversation.

Cross-analysis of multiple documents in ~20 min before you sit down with the other party; you walk into the meeting knowing where to push.

Clear delegation to AI with the R-C-E-S framework

3 real requests from your management week (communication with partners/board, proposal analysis, decision synthesis) rewritten with R-C-E-S, tested before/after and saved as reusable templates.

Less rework: the R-C-E-S version comes out closer to the final format and needs less rewriting than starting from scratch.

Library of recurring executive deliverables

6 ready-to-use templates to drop into your management routine: a difficult email (partner/client/board), a board meeting summary, a proposal analysis in XML, a 1-page executive report, a critical review of the AI's response, and the Prompt Improver.

Standardizes recurring deliverables; reports become reads of up to 5 min and emails fit defined limits — less of your time spent writing.

A rubric so you never bring the board an answer that only 'looks good'

A 0-10 rubric with 5 criteria (adherence, use of context, format, hallucination risk, next step) and a minimum score of 8 for what matters, applied in a v1-to-v2 loop before any decision leaves your desk.

Swaps gut-feel assessment for evidence-based assessment — catches what fails a criterion before it goes to the committee or the client.

Management context curated in 3 layers

A context map of a real decision classified into Project / System Prompt / Inline, with fixed rules (don't invent data, flag gaps, deliver ready for review) and an antidote to context rot for long analysis conversations.

Less noise and more precision in long analyses; you stop re-pasting briefing and rules with every new decision conversation.

A 'Company/Unit' Project configured for recurring management

A Project in Claude Desktop with Project Instructions (role, executive tone, format, confidentiality, mandatory human review) and a curated Knowledge Base with 3 documents that change the answer — business numbers, policy/strategy and a decision template — with sensitive data masked.

Eliminates rebuilding context every week; the starting point of your management analysis is already in place and consistent.

A secure Drive/calendar connector with C-level governance

Connecting an official tool via Connectors/OAuth (no JSON editing) with a pre-connection governance checklist (what will be accessed, for how long, who reviews) and the read document, summarize, create task flow — with human review before any sensitive action.

Proof of value with minimal scope and least privilege; you free up reading time without giving up control or the ability to revoke access.

Automation of your start-of-week triage (no code)

A 7-block Skill (name, context, rules, output, human approval gate, log and rollback) applied to a management routine that hurts every week — e.g., triaging emails/requests into urgent, decide today, decide later and needs review — with the final send always dependent on your approval.

Eliminates manual triage at the start of the week: AI prepares the classification, you just decide — decision time freed up safely and reversibly.

Anchored in the "Dominando IA" course: AI drafts and researches to speed up your decisions; you review and decide what goes to the board, into the contract, and to the client.

Today, each person on your team uses AI their own way: different prompts, different criteria, no shared standard. The knowledge of those who use it well never reaches the others, onboarding is slow, and the productivity gain never shows up in the process. This track trains the entire department as a unit to operate with a common method: the team walks away with standardized, versioned, and auditable prompts, Projects, Skills, and workflows that any new hire inherits on day one. The manager starts measuring productivity in the process, not in individual effort.

What the department gains together by the end of the training

Shared vocabulary and fundamentals for the entire department

A leveling session that brings the whole team to the same starting point: distinguishing generative AI from ML and RPA, understanding Context Window, Tokens, Temperature, and RAG, and speaking the same language in meetings. It ends the divide where some have mastered it and others haven't even started.

Leveling the team in ~1 session instead of months of scattered, uneven learning across people

Team-standard anti-hallucination checklist (3 steps + 6 techniques)

A single verification protocol that everyone follows before using an AI response: classify the risk, check against a primary source, review with an expert. It becomes a department rule, not an individual choice — no one delivers unverified data because they didn't know the method.

Verification costs minutes; standardizing the protocol prevents one member's mistake from becoming rework or a reputational risk for the team

Model and plan policy for the department (Haiku/Sonnet/Opus + Team)

Shared criteria for which model to use per task type (Haiku for triage, Sonnet for 80% of day-to-day, Opus for deep analysis) and which plan to contract for the team — Team around R$1,500-1,800/month for 10 people, Enterprise only when compliance requires it.

Stops the team from spending on Opus/Enterprise for simple tasks and standardizes the plan decision for the whole department

Shared R-C-E-S prompt library for the department

The department's recurring tasks (difficult email, meeting summary, proposal analysis, 1-page report, critical review) rewritten with the R-C-E-S framework and saved as templates everyone pastes and reuses. One person's good prompt becomes everyone's standard.

Reduces repetitive rework by standardizing the team's recurring deliverables; recurring deliverables become a read in up to 5 minutes

A single 0-10 quality rubric for the team

A 5-criteria rubric (adherence, use of context, format, hallucination risk, next step) with a minimum score of 8 for important deliverables, used across the entire department. It replaces 'it looks good in my opinion' with an objective criterion that is the same for everyone before anything reaches the client.

Replaces evaluation by individual gut feeling with standardized evidence, with consistent quality across different people

Three-layer context engineering standard for the team

A shared model of how the team organizes AI context — what belongs in a Project (persistent), what belongs in the System Prompt (fixed rule), and what is Inline (just the conversation) — using the 3-question rule. Everyone structures context the same way, with the same fixed working rules.

Reduces noise and improves response accuracy consistently across the whole team, not just those who already know how

Reusable, versioned client/topic Projects for the department

Projects in Claude Desktop configured by client or topic with Project Instructions (role, tone, format, confidentiality, human review) and a curated Knowledge Base — owned by the department, not by one person. Anyone who joins the team inherits the ready-made Project and produces to standard on day one.

New-member onboarding drops from weeks to hours; the starting point is already built and standardized for everyone

Department standard for Knowledge Base and Artifacts

A shared rule for which 3 documents go into each Project's base (briefing, reference, template/policy) with sensitive data masked, and which Artifact formats the department delivers (executive summary, plan, comparison, checklist). It standardizes what goes in and what comes out.

On-target, tone-consistent responses across the whole team, with less correction and lower hallucination risk from uncontrolled input

Connector and governance standard for the department (Drive, tasks)

A single connection procedure via Connectors/OAuth with a pre-connection governance checklist (what is accessed, for how long, who reviews) and a standard reference workflow: read a Drive document, summarize it, create a task in the team's system — always with minimum scope and human review.

Safe repeatability across all members and standardization of the read-summarize-create-task workflow, with least privilege applied by everyone

Team Skills for the routines that hurt every week

The department's repetitive tasks (e.g., triaging client emails on Mondays) turned into 7-block Skills — name, context, rules, output, human gate, log, and rollback — versioned and shared. Automation becomes a department asset, with mandatory human approval before any send.

Eliminates the team's repetitive manual triage at the start of the week and standardizes automation with controlled tests and rollback

Based on the anchor course Dominando IA, adapted to train the department as a unit: AI drafts and researches, the team reviews and decides — and now everyone follows the same method.

You sit on the board to judge, not to produce. But the board package lands on Sunday, the AI proposal arrives full of hype, and management mistakes a standalone tool for architecture. This track equips the board member to use AI in their own preparation — reading the board package, sizing risk, doing due diligence on proposals — and to arrive at the table with questions that separate real implementation from a pretty slide. This is not governance theory or an advisory seat: it is what you start doing better in Thursday's meeting. AI speeds up reading and research. The decision — with an owner and a deadline — still belongs to the board.

What you walk away delivering at the table — and the time you get back for your board preparation

Critical reading of the board package via RAG

Drag in the meeting package (CEO report, financial statements, investment proposal) and extract into a table the risk points, omissions and numbers that don't reconcile across the documents — with page citations to raise at the table.

80-page package read critically in ~15 min (analysis via API for under US$3 vs. an analyst hour)

List of tough questions for the meeting, before walking into the room

From the package, generate the 7 questions management doesn't want to hear: where the scope is poorly defined, which assumption doesn't hold, what material risk was left out. You arrive asking, not reacting.

From 'found out on the spot' to prepared before the agenda

Due diligence on an AI proposal: architecture vs. standalone tool

Submit the AI proposal the team brought and separate what is real architecture (proprietary data, validation pipeline, human review) from what is just 'we bought an AI' hype. Demand a real case, cost, 12-month ROI and milestones.

Hype dismantled in ~25 min, with 1 data point, 1 case and 1 material risk in hand

Cross-referencing the main proposal + addenda in a single analysis

Proposal, contract and two addenda on the same table — AI flags where an addendum contradicts the main body and which dependency was left implicit. No reading one and forgetting the other.

Cross-review of proposal + 2 addenda in ~30 min

Anti-hallucination checklist for every factual data point before voting

Before approving based on a market figure, case law or standard cited in the package, classify the risk and verify it against the primary source. A board member who approves on fabricated data is held accountable too.

Verifying costs minutes; not verifying costs liability (RAG reduces hallucination 40-70%)

Hallucination hunt within the board's own subject matter

Ask the AI for 5 regulatory data points or studies from the last 12 months (CVM, BACEN, ANS, McKinsey data) and check each one against the official source. Trains the instinct to never repeat in the minutes a number that doesn't exist.

Avoids carrying into the minutes a repealed law, a nonexistent study or a fabricated standard with perfect formatting

Board Project with the base you always come back to

Create a Project in Claude Desktop with the bylaws, shareholders' agreement, last 3 sets of minutes and the risk policy. Every analysis starts from the same criteria, without rebuilding context at each meeting.

No more starting over each agenda — the starting point is ready

Fixed board instructions (role, confidentiality, review rule)

Record in the Project the rules that always apply: board member tone, never invent a number, flag a gap before recommending, mark what is confidential, and that the final decision belongs to the body. Confidentiality and secrecy built in.

Secrecy and criteria applied every time without rewriting the instruction

Executive meeting summary with decision, owner and deadline

Turn the discussion into a single page: facts, decisions made, open items — each with an owner and a date. You leave the meeting with an actionable next step, not with 'let's schedule another presentation.'

Action-item minutes in ~1 page, each open item with owner and deadline

Board reading connector + follow-up task

Connect the board's Drive through the panel (no JSON editing), let the AI read the document, summarize it and open the decision's follow-up task — with minimal access scope and human review before any action.

From document to next step without leaving Claude, with minimal scope and a human gate

Everything anchored in the "Dominando IA" course and the board member's golden rule: AI drafts and researches the package, the proposal and the risk; you review and decide — with an owner and a deadline.

ANCHOR COURSE

Mastering AI — from concept to automation with Claude Desktop

The complete course: it starts from the concept and ends with you automating real work on your machine. It's not theory about AI — it's practice guided from scratch to building your first automations, with Claude Desktop as the work environment. You leave using it, not just knowing it exists.

2 modules · 6 courses · ~16.5 hours of content

I want the anchor course

Listen to the episode about the course

Dominando IA podcast audio, about 26 minutes.

Three formats

Live, in a cohort

Cohort with a set date, run live, with room to ask questions about your real case. The next cohort doesn't have a date yet — join the waitlist and you'll be the first to know.

Join the waitlist

In-company

Closed training for your team, with examples and processes from your own operation. Scope and hours adjusted to the team. By quote.

Request an in-company quote

Recorded, lifetime access

The complete course recorded, at your own pace, with lifetime access. For those who prefer to start now without waiting for a cohort. Direct purchase.

Buy lifetime access

Why Claude Desktop — and why that matters less than it seems

We use Claude Desktop because today it's the best environment to go from concept to real automation on a work machine. But the tool is the vehicle, not the destination. What you take away from the training is method: how to think about problems in terms of automation, build workflows and become an AI power user. If the tool changes tomorrow, the method still holds — that's where the gain is.

Join the waitlist

The next live cohort doesn't have a date yet. Leave your email and you'll be the first to know when it opens — no spam.

Become an AI power user — not a spectator

Who it's for

Lawyers and independent professionals

Business owners and C-level

Board members

In-company teams

Don't see your group here?

LAWYERS & LIBERAL PROFESSIONALSFor lawyers: the billable hour back in your day

BUSINESS OWNERS & C-LEVELFor business owners and C-level: decide on AI based on what it actually delivers — with the first automations already running in your management routine

IN-COMPANY TEAMSFor in-company teams: a single AI method that the entire department uses the same way, with reusable standards and fast onboarding

BOARD MEMBERSFor board members: arrive at the table with the right question — and demand a decision, not one more presentation

Mastering AI — from concept to automation with Claude Desktop

Three formats

Live, in a cohort

In-company

Recorded, lifetime access

Why Claude Desktop — and why that matters less than it seems

Join the waitlist