Learning path map
Detailed content
🎯 Creating your custom assistant
Build an AI assistant that understands your profession, your audience, and the way you work.
A personalized assistant is an AI configured with specific instructions about your profession, your audience, and your work habits. Unlike a "blank slate" ChatGPT, it starts out knowing who you are and what you need.
The difference between a generic assistant and a customized one is like the difference between an intern on their first day and a colleague who’s worked with you for 2 years. The customized assistant already understands your jargon, tone, and routine.
System prompt vs. conversation prompt. Persistent instructions. Professional context. Consistent tone across conversations.
The system prompt is your assistant’s “factory-set personality.” It has 5 parts: identity, target audience, behavioral rules, response format, and knowledge base.
Without a well-written system prompt, you have to repeat instructions in every conversation. With one, the assistant starts aligned and you save minutes in every interaction.
5 blocks: (1) Identity, (2) Audience, (3) Rules, (4) Format, (5) Knowledge. Each block has a clear purpose and concrete examples.
A complete system prompt for a general practitioner, showing how to fill in each of the 5 parts with real information from medical practice.
Seeing a real, concrete example dramatically reduces the time it takes to create your own. You adapt instead of inventing from scratch.
Medical terminology in the prompt. Ethical restrictions (no diagnosing). Tone for patients vs. colleagues. Report and referral templates.
A complete system prompt for a civil attorney, showing how to include legal language, document templates, and professional confidentiality rules.
Every profession has unique jargon and constraints. Comparing examples from different fields helps identify what’s universal and what’s specific to your field.
Legal language in the prompt. Confidentiality and LGPD. Formal vs. accessible tone. Petition, legal opinion, and contract templates.
A universal template that works for any independent professional. You copy it, fill in the fields with your information, and paste it into ChatGPT, Claude, or Gemini.
The template eliminates the blank page. Instead of wondering “where do I start?”, you just fill in the fields and have a working assistant.
Template with [FILL IN] placeholders. 5 required fields. Optional fields for refinement. How to paste into ChatGPT/Claude/Gemini.
A checklist of 5 test questions to validate whether the assistant is responding with the tone, depth, and accuracy you expect.
An assistant that seems good in the first test may fail in real situations. The checklist reveals weak spots before you rely on it with clients.
5 test questions. Generic vs. customized comparison. Iterate on the prompt in 3 rounds. Know when “it’s good enough.”
📚 Teaching the Context of Your Profession
Help AI understand your field, your style, and your references so it can produce work as if it were you.
Professional context is the set of information that helps AI produce responses relevant to YOUR practice: your specialty, your audience, your market, your references, and your standards.
Without context, AI responds generically. With context, it responds like someone who knows your field. The difference is between “useful” and “indispensable.”
Context = information the AI doesn’t have but needs. The more specific it is, the better the output. Context accumulates across conversations (in saved assistants).
An organized list of the information you should provide to the assistant: firm/office details, specialties, target audience, differentiators, competitors, and typical cases.
Most people provide too little information or provide it in a disorganized way. A structured brief helps the assistant deliver from the very first conversation.
6 briefing blocks: Identity, Specialty, Audience, Differentiators, Restrictions, Common cases. Fill-in template.
Techniques for calibrating the assistant's voice: from formal to casual, from technical to accessible. Includes "before and after" examples of tone adjustments.
If the assistant writes in a tone that doesn't sound like you, you'll have to rewrite everything. Getting the output tone right saves more time than any other optimization.
Formal-to-informal spectrum. Reference examples ("write like this text"). Words and expressions to use/avoid. The "does this sound like me?" test.
How to give the assistant reference documents: contract templates, technical standards, clinical protocols, price lists, and other materials it should consult.
When the assistant has access to your documents, it can draft materials using your own references instead of making them up from scratch. The quality jumps to a new level.
File uploads in ChatGPT/Claude. GPTs with knowledge files. Context limits. How to organize documents so AI can find them quickly.
Techniques for identifying and correcting unwanted behaviors: wordy responses, the wrong tone, made-up information, or unsuitable formatting.
No assistant is perfect on the first try. Knowing how to diagnose and fix problems is what separates those who give up ("AI doesn't work") from those who keep optimizing until they become productive.
Diagnosis: which part of the prompt is causing the problem? Iteration: adjust, test, repeat. Negative rules ("never do X"). Examples of corrections.
A final checklist with all the items that make a professional context complete: from basic information to nuances of tone and ethical constraints.
The checklist works like an audit. Even if your assistant is already good, going through each item ensures you haven’t forgotten anything important.
12 checklist items. Priority (3 required, 5 recommended, 4 optional). How to revisit the checklist every 30 days.
🛠️ Daily AI workflow
Integrate the assistant into your actual routine: from starting the day to wrapping up, including communication, production, and service.
How to use the assistant during the first 10 minutes of your day to organize priorities, prepare for client sessions, and anticipate pending tasks.
Starting the day with clarity changes the pace of everything. AI can cross-reference your calendar with previous pending tasks and suggest an intelligent task order.
“Morning briefing” prompt. Calendar integration. Prioritization by urgency and importance. 10 minutes that save 1 hour.
Workflow for using the assistant in your daily communications: drafting emails, replying to WhatsApp messages, and preparing announcements, all in the right tone.
Communication takes up 30-40% of a self-employed professional’s day. An assistant that already knows your tone cuts that in half without sacrificing quality.
Email template by type (reply, follow-up, proposal). Formal tone vs. WhatsApp. Reply in 3 clicks: paste, adjust, send.
How to use the assistant to draft technical documents: reports, opinions, assessments, and professional documents, starting from notes and raw data.
Technical documents are where independent professionals spend the most creative energy. AI turns notes into structured drafts, and you just refine them.
Workflow: raw data → structured prompt → draft → review. Document templates by profession. Post-AI review checklist.
How to use the assistant to prepare before a consultation, meeting, or client session: case summary, history, key questions, and possible next steps.
Arriving prepared changes the client’s perception. AI creates in 2 minutes the summary that would take you 15 minutes of reading medical records and old notes.
“Pre-appointment briefing” prompt. Case data as input. Questions suggested by AI. Automated post-appointment follow-up.
AI end-of-day routine: summary of what was done, items pending for tomorrow, follow-ups to send, and note organization.
Ending the day with clarity keeps pending tasks from being forgotten. AI turns your scattered notes into an actionable summary for the next day.
“Daily wrap-up” prompt. Automated follow-up list. Context transfer between days. 5 minutes of structured wrap-up.
A 4-week plan to adopt AI gradually, starting with the areas offering the greatest gains and expanding as you build confidence.
Trying to change everything at once leads to frustration. Gradual adoption lets you validate each step before moving on and ensures AI really helps instead of getting in the way.
Week 1: Communication. Week 2: Production. Week 3: Customer service. Week 4: Complete routine. Indicators that it’s working.