Meet Sophie

Sophie is a freelance English trainer. She runs 15 to 20 sessions a week across four training centers. After each session, she types up a quick email summary — when she remembers.

Think about your own post-session routine. How consistent is it?

Key Takeaway

Most trainers have a workflow. Few have a system.

The Problem

Sophie gets an email: "Audit scheduled for March 15. Please provide session records, learner follow-up history, and proof of content covered." She checks her files. Some notes are in Google Docs. Some in email drafts. Two sessions have no notes at all.

This is the moment most trainers dread. The audit isn't the problem — the missing trail is.

Which of these is hardest to reconstruct after the fact: what was covered, what the learner struggled with, or what exercises were assigned?

All three matter for audits — and all three are easy to lose without a system.

Key Takeaway

Inconsistent notes create audit risk and hurt learner follow-up.

What If There Were a System?

Sophie hears about a workflow where every session automatically produces two things: a Learner Profile (created once, after the first session) and a standardized Post-Session Message — sent to the learner, herself, and the training center. All from Notion.

This is what you'll build by the end of this training.

Key Takeaway

The goal: a repeatable workflow that produces consistent, audit-ready output every time.

The Three Databases

Sophie's system lives in three databases: Learners (contact info and learner profiles), Sessions (one entry per session, linked to a learner), and Prompts (her AI prompt library — standalone, not linked). Only Sessions and Learners are connected by a relation. Prompts are invoked through Notion AI chat.

This is the backbone. Every artifact you produce traces back to one of these three databases.

Key Takeaway

Three databases hold everything. Sessions and Learners are linked by a relation. Prompts is a standalone library — called via AI chat, not a database connection. A ready-made template with all three databases is waiting at the end of this training.

Setting Up the Learners Database

The Learners database stores one entry per learner (or group). Key properties: Name (title), Email, Sessions (relation to Sessions DB). The Learner Profile is not a property column — it lives inside the page body. We'll explore a learner profile later in this training — for now, just know it lives inside the page body, written by AI after the first session.

NameEmail(s)Sessions
Marc Dupontmarc.dupont@email.comSession 1
Léa Chenlea.chen@techcorp.frSession 3, Session 4
Group Unilever #0301@unil.com, 02@unil.com, 03@unil.comSession 1
Ana Kovačana.kovac@logistiqa.comSession 1
Group BNP #07j.martin@bnp.fr, s.lee@bnp.fr, p.noir@bnp.fr, c.varga@bnp.frSession 1

Later, you'll open a learner entry and find the profile inside the page body — not in a column. Created once, after the first session, from intake questions.

Where does the Learner Profile live?

A. In a property column of the Learners database
B. Inside the page body of the learner entry
C. In the Sessions database
D. In a separate Profiles database

Right — the profile is written inside the learner entry's page body, created from the first session's intake interview.

Not quite. The learner profile lives inside the page body of the learner entry — not in a property column.

Key Takeaway

The Learners database is the single source of truth for who you train and how they're progressing.

Setting Up the Sessions Database

The Sessions database stores one entry per session. Key properties: Session Title, Date and Time, Learner (relation to Learners), Lesson (text), Assessment (text), Exercises (text), Send Button. Inside the session page, you'll also find the AI Meeting Notes block — it's page content, not a column.

TitleDateLearnerLessonAssessmentExercisesSend
Session 2YesterdayLéa Chen✓ Sent
Session 3YesterdayLéa Chen✓ Sent
Session 1TodayMarc DupontSend
Session 1TodayGroup Unilever #03Send
Session 1TomorrowAna KovačSend
Session 4TomorrowLéa ChenSend
Session 1Day+2Group BNP #07Send
Session 2Day+2Marc DupontSend

This is where the action happens. AI Meeting Notes live inside the page. The three output columns are filled by AI. The Send button compiles them into one message and delivers it automatically.

Key Takeaway

Each session page is self-contained: notes in, message out, send with one click.

Setting Up the Prompts Database

The Prompts database is Sophie's AI prompt library. Each entry has just two properties: Prompt Title and Prompt Text (the full prompt instructions). Two key prompts live here: AI Learner Profile and AI Post-Session Notes. To use one, open the relevant session page, open AI chat, and type something like "Execute @AI Post-Session Notes for @Marc Dupont."

Prompt TitlePrompt Text (preview)
AI Learner ProfileReads the first session transcript and intake answers to produce a structured learner profile: Strengths, Areas to develop, Baseline level, Recommended focus areas.
AI Post-Session NotesReads three inputs (prompt instructions, session page context, learner profile) to produce Lesson, Assessment, Exercises.

Think of this as your recipe book. Prompts are reusable — you write them once and call them by @mention in Notion AI chat. No database links needed.

Key Takeaway

A central prompt library keeps your AI outputs consistent and editable in one place.

Sophie Opens the Template

Sophie clicks Duplicate on the shared template. A new workspace appears — three databases, dozens of properties, two pre-written prompts. For a moment, it's a lot. But then she notices: every database has a clear name, every property has a label. The prompts are already written. She just needs to start with one database and follow the trail.

This is normal. A new tool always feels overwhelming at first glance. The structure is already there — your job is to explore it one piece at a time.

When you open a new tool for the first time, what helps you feel oriented — labels, examples, or a guide?

Sophie's template has all three — and you've just seen each database in detail.

Key Takeaway

A good template doesn't just save setup time — it teaches you the system by showing you what belongs where.

Quick Check: Where Does It Live?

Let's make sure you know where everything goes.

Match each artifact to the correct database and location.

Which artifact lives inside the page body of a Learners database entry?

A. The session transcript
B. The AI-generated learner profile
C. Lesson, Assessment, and Exercises
D. The AI prompt instructions

Correct! Full map: (1) Learner profile → Learners DB / page body. (2) Transcript → Sessions DB / AI Meeting Notes block. (3) Lesson, Assessment, Exercises → Sessions DB / property columns. (4) Prompt instructions → Prompts DB / Prompt Text property.

Not quite. The learner profile lives inside the page body. Transcript → Sessions page. R/E/E → Sessions columns. Prompt text → Prompts DB.

Key Takeaway

Every artifact has a home. Knowing where things live is half the system.

Preparing for the First Session

Before Sophie records her first session with Marc, she needs to plan her intake interview. The Learner Profile is created once — during the first session only. Sophie asks targeted questions covering six areas: Background and Role, Current Skill Level, Goals and Motivation, Preferences and Learning Style, Context and Constraints, and for groups: shared goal and range of experience levels.

These six categories work for any topic — not just language training. The categories stay the same; only the specific questions change.

How often is the Learner Profile created?

A. After every session
B. Once, during the first session
C. Only when the trainer remembers

The profile is a one-time baseline. After that, only the Post-Session Notes prompt runs each session.

The profile is created once, during the first session, based on intake questions the trainer asks.

Key Takeaway

One intake interview → one profile. Every session after that focuses on the post-session message only.

Sophie's First Session with Marc

Sophie opens Marc's session page in Notion. She clicks Record to start AI Meeting Notes. For the next 60 minutes, Notion captures the full transcript. When the session ends, it generates a summary automatically.

This is the input for everything that follows. No transcript means no AI output.

Key Takeaway

AI Meeting Notes capture the raw material: transcript + summary. Everything downstream depends on this.

What AI Meeting Notes Give You

After the session, Sophie has two things inside the Meeting Notes block: a transcript (full conversation) and a summary (AI-generated key points). These stay inside the session page — attached to Marc's record.

The transcript is your evidence. The summary is your shortcut. Both feed into the prompts you'll run next.

Why is it useful to have both a transcript AND a summary?

The transcript gives detail and proof. The summary saves time when generating outputs. Together, they give the AI everything it needs.

Key Takeaway

Transcript = detail + audit trail. Summary = speed. Use both.

How to Invoke a Prompt

Here's how prompts work. Sophie is on Marc's session page — AI Meeting Notes captured. She opens Notion AI chat and types: "Execute @AI Learner Profile for @Marc Dupont". The AI reads the prompt instructions (from the Prompts database) and the current page context (meeting notes, transcript, summary). Then it acts.

In this system, the prompt knows that the learner profile lives inside the learner page — so you reference the learner in the command.

This is the keystone interaction. Everything before this was setup. Everything after depends on typing one line in AI chat.

Key Takeaway

One line in AI chat triggers the entire generation. The AI reads the prompt and the page — you just tell it which prompt to run.

Generating the Learner Profile

Sophie just ran "Execute @AI Learner Profile" in Notion AI chat. The AI reads the transcript — including the intake interview answers — and produces Marc's learner profile: strengths, areas to develop, baseline level, and recommended focus areas. This is artifact number one.

This is artifact #1. The prompt does the heavy lifting — you invoke it from AI chat and review the output.

Key Takeaway

The Learner Profile Generator turns the first session's intake data into a structured learner record.

Model Answer: Marc's Learner Profile

Here's what the AI produced for Marc:

1. Who is this learner?

Marc is a logistics coordinator at Transfrêt, a mid-size French freight company recently acquired by a British-German group (Kelner Logistics). He has been in his role for six years and never needed English beyond scanning the occasional email. Since the acquisition three months ago, he now sits on a weekly cross-site call in English, receives English-language reporting templates, and is expected to present monthly KPIs to the new UK-based operations director. He describes his level as "school English — enough to survive on holiday, not enough for a meeting."

2. Why are they learning?

  • Explicit: Participate in weekly English team calls, write clear status emails, deliver monthly KPI presentations in English.
  • Inferred: Feels embarrassed around younger colleagues more comfortable in English. Worried the acquisition could lead to restructuring — wants to prove his value. Sees English as a personal goal the acquisition forced.

3. Goals and direction

  • Near-term: Follow and contribute to the weekly call within two months. Send status emails confidently. Deliver his first KPI presentation in English.
  • Longer-term: Reach a level where English feels routine. Potentially take the Linguaskill exam. Be seen as someone who adapted to the acquisition.

4. Attitude and relationship to learning

  • Uses self-deprecating humor to manage expectations.
  • Over-prepares written work but avoids spontaneous speaking.
  • Gets discouraged by listening comprehension — UK colleagues speak fast with unfamiliar idioms.
  • Responds well to structure and templates.
  • Disengages if activities feel too far from real work.
  • Appreciates concrete proof of progress.

5. Preferences and constraints

  • Real work materials (actual emails, KPI slides) are more engaging than textbook content.
  • Role-play of specific upcoming situations works well.
  • 20 hours over approx. five months, 1-hour sessions twice a week at lunch.
  • Online via Google Meet. Limited homework time (15-20 min). Often tired at lunchtime.

6. How to work with this learner

  1. Use his real work materials as lesson content.
  2. Build a bank of reusable phrases for meetings and emails early.
  3. Address listening anxiety directly with gradual exposure.
  4. Keep sessions structured with a visible agenda.
  5. Limit homework to one short, specific task.
  6. Don't push spontaneous speaking too early — use semi-scripted role-plays first.
  7. Name his progress concretely with evidence.
  8. Respect the energy dip — switch to lighter tasks if he's fading.

Notice the structure: six sections, each grounded in what Marc actually said during the intake. The key is specificity — not length.

Key Takeaway

A good learner profile is specific, structured, and actionable.

Generating the Post-Session Message

Sophie types "Execute @AI Post-Session Notes for @Marc Dupont" in Notion AI chat. The AI reads three inputs: the prompt instructions, the session page context (meeting notes, transcript, summary), and Marc's learner profile. It fills three separate properties: Lesson, Assessment, and Exercises. This is artifact number two.

This is artifact #2. The AI fills each column individually — same format, every time. The Send button compiles them into one message.

Key Takeaway

Fixed format = consistent quality. The learner always knows what to expect.

Model Answer: Lesson

Here is the Lesson section the AI produced for Marc:

LESSON — Keys to success for early ESL

Objective: Build simple weekly habits that develop all 4 skills.

1. Listening — 5-10 min, 4-5 days/week
Use a short clip (1-3 min). Watch once for the main idea. Watch again with subtitles, note 5 useful phrases. Listen again and repeat 3 key lines (shadowing).

2. Reading — 5-10 min, 4-5 days/week
Read something easy and relevant. Highlight 5 useful expressions. Write a 1-2 sentence summary.

3. Speaking — 2-5 min daily
Record a voice note using: Update (Today, I…), Issue (We have a problem with…), Next step (Next, I'm going to…), Request (Could you confirm…?).

4. Writing — 3-5 min daily
Write 3-5 lines: what you did today, what you will do tomorrow, one new phrase in a sentence.

Rules: Consistency beats intensity. Repeat content for 2-3 days. Keep a "useful phrases" list.

Starter phrases: Quick update from my side… / We are on track. / We have a delay on… / The main issue is… / Could you repeat that, please? / Could you speak a bit more slowly?

This is what Marc receives as his lesson plan — a full lesson, structured and tailored to his profile.

Key Takeaway

Lesson isn't a summary — it's a standalone lesson the learner can follow between sessions.

Model Answer: Assessment

Here is the Assessment section:

ASSESSMENT

[x] = Acquired  ·  [ ] = In progress

[ ] Understand the main idea of a short clip (1–3 minutes).
[ ] Catch key details (names, numbers, times) with subtitles support.
[x] Understand the gist of a short, level-appropriate text.
[ ] Identify and reuse useful phrases (chunks) from a text.
[ ] Speak for 60 seconds with basic clarity (topic + 2–3 details).
[x] Improve pronunciation by repeating and shadowing 3 key lines.
[ ] Write a short message (3–5 lines) that is clear and understandable.
[ ] Self-correct basic errors (spelling, verb form, word order) using a model.

All descriptors are "In progress" after Session 1 — that's expected. Over time, the trainer checks "Acquired" as evidence accumulates. This is the audit trail for skill progression.

Key Takeaway

Assessment tracks what the learner can do — not what was taught. It's evidence-based and cumulative.

Model Answer: Exercises

Here is the Exercises section:

EXERCISES — This Week

Choose 1 task per day (rotate skills):

  1. Listening: one 2-minute clip, 3 passes + shadow 3 lines.
  2. Reading: one short text (or work email) + 2-sentence summary.
  3. Speaking: 2-minute voice note using the template (update, issue, next step, request).
  4. Writing: 5 lines + one new phrase in a sentence.

Answer key example (Writing):

  • Today: I updated the delivery report and checked Route A.
  • Today: I emailed the UK team about the delay.
  • Tomorrow: I will confirm the new ETA with the carrier.
  • New phrase: "We are slightly behind schedule."
  • Sentence: We are slightly behind schedule, but we will catch up tomorrow.

The answer key lets Marc check his own work — no need to wait for Sophie.

Key Takeaway

Exercises with an answer key = autonomy. The learner can practice and self-correct without waiting for the next session.

Sophie's Workflow Comes Together

Sophie now has everything in place for Marc's session: Transcript and summary captured. Learner Profile created. Post-session message written. Every artifact is in its database. Every property is filled.

TitleDateLearnerLessonAssessmentExercisesSend
Session 2YesterdayLéa Chen✓ Sent
Session 3YesterdayLéa Chen✓ Sent
Session 1TodayMarc DupontSend
Session 1TodayGroup Unilever #03Send
Session 1TomorrowAna KovačSend
Session 4TomorrowLéa ChenSend
Session 1Day+2Group BNP #07Send
Session 2Day+2Marc DupontSend

This is the payoff. One session — two structured outputs — zero guesswork.

Key Takeaway

When the system works, every session produces complete, consistent, traceable output.

The Send Button

Sophie opens Marc's session page. The post-session message is ready — all three sections filled. She clicks the Send button. The message goes to three recipients: Marc (the learner — delivery), Sophie (the trainer — personal archive), and the training center (audit and quality control). Total time after session: under 5 minutes.

One click, three recipients, three purposes. From session end to fully archived output: under 5 minutes.

Key Takeaway

From session end to fully archived, delivered output: under 5 minutes. One click handles delivery, archiving, and audit.

Why Three Recipients?

Learner or Group: Gets the post-session message for review and follow-up. Trainer (self): Keeps a personal archive — searchable, timestamped. Training center: Receives the same message for audit trail and quality control. Every session is documented, delivered, and archived — automatically.

Each recipient serves a different purpose. Removing any one of them breaks the chain.

What would break if the training center were removed from the recipients?

The audit trail disappears. The trainer and learner still benefit — but compliance is lost.

Key Takeaway

Three recipients = delivery + archive + audit. All three are needed for a complete workflow.

Monday Morning

It's Monday morning. Sophie opens her calendar: 15 sessions this week — individual and group sessions mixed. The workflow is only three steps (record, generate, send), but she has to commit to running it every single time. She takes a breath and starts her first session of the week.

TitleDateLearnerLessonAssessmentExercisesSend
Session 1MonMarc DupontSend
Session 1MonGroup Unilever #03Send
Session 1MonSarah NguyenSend
Session 2MonOmar BenaliSend
Session 3MonGroup TechCorp #02Send
Session 1TueAna KovačSend
Session 2TueLéa ChenSend
Session 3TueJulien MorelSend
Session 1WedGroup BNP #07Send
Session 2WedMarc DupontSend
Session 3WedCamille RossiSend
Session 1ThuLéa ChenSend
Session 2ThuNoura El IdrissiSend
Session 1FriGroup Kelner Logistics #01Send
Session 2FriThomas BergerSend

The hard part isn't learning the system. It's deciding to trust it under real conditions — 15 sessions, back to back.

What would make you hesitate to use this system for all 15 sessions this week?

The cost isn't complexity — it's commitment. By session 12, the system has already proven itself.

Key Takeaway

The leap is trusting the system before it's proven itself. The landing is easy.

What About Groups?

Sophie's next session is with a group of 4 learners from the same company. The group is already in the Learners database as a single entry with multiple email addresses. She records, runs the prompts, and clicks Send. Same three steps. Same databases. Same prompts. Nothing changes.

A group is just another entry in the Learners database. The workflow doesn't branch — it doesn't even notice.

True or false: The workflow requires different steps for a group session than for a 1:1.

A. True — groups need a modified workflow
B. False — the workflow is identical for 1:1 and groups

The process is identical. A group is just another entry in the Learners database — the system handles both by design.

Actually, the workflow is identical. A group is just another entry in the Learners database with multiple emails.

Key Takeaway

The system handles 1:1 and group sessions the same way. No branching, no special cases.

When Things Go Wrong

The AI isn't perfect. Sometimes the transcript is garbled. Sometimes the output misses a key point. Sometimes the learner profile needs updating. Sophie's fixes: garbled transcript — review summary, fill gaps, re-run. Wrong output — edit directly, then send. Outdated profile — open the learner page and edit.

AI output is a draft, not a final product. Your job is to review, edit if needed, and only then send.

Sophie finishes a session but forgets to run the Post-Session Notes prompt before clicking Send. What happens?

A. Notion warns her and blocks the send
B. The message is sent with empty content — the columns had nothing in them
C. The AI runs the prompt automatically before sending

The Send button compiles whatever is in the columns. If they're empty, the message is empty. Always run the prompt before sending.

No safety net. The Send button compiles whatever is in the columns — empty or not. Always run the prompt first.

Key Takeaway

The system handles the routine. You handle the exceptions — review, edit, re-run.

Sophie's Transformation

Before: scattered notes, inconsistent emails, audit anxiety. After: every session captured, every learner profiled, every message sent and archived — from a single workspace. Sophie doesn't have to remember the steps. The system remembers for her.

This is where you'll be at the end of this training.

Key Takeaway

A system beats willpower. Build it once, use it every session.

Recap: The Full Workflow

The complete workflow in five steps:

1. Set up 3 databases: Prompts, Learners, Sessions (with relations).
2. Record each session with AI Meeting Notes (transcript + summary).
3. Run the Learner Profile Generator prompt (first session only).
4. Run the Post-Session Notes prompt — Lesson, Assessment, Exercises.
5. Click Send — deliver to learner + archive to trainer + audit to training center.

Steps 1 and 3 happen once. Steps 2, 4, and 5 repeat every session.

Key Takeaway

Setup and profile happen once. Record, message, send repeat every session.

Transfer: Apply It Tomorrow

Want a head start? Duplicate Sophie’s ready-made Notion template — all three databases, properties, relations, and both prompts are pre-configured.

Don't wait. The best time to build the system is before your next session.

Key Takeaway

The transfer is immediate: set up before your next session and run the full workflow.

Self-Check: Question 1 of 5

What are the three databases in the Training OS?

A. Prompts, Learners, Sessions
B. Templates, Contacts, Calendar
C. Notes, Tasks, Projects
D. Prompts, Students, Classes

These three databases hold every artifact in the workflow.

The three databases are Prompts, Learners, and Sessions.

Self-Check: Question 2 of 5

Sophie records a session with Marc, but her internet drops for 5 minutes mid-session. The transcript has a gap. What's her best next step?

A. Delete the session and start over
B. Read the summary, fill in the missing details manually, and re-run the prompt
C. Send the message as-is — the AI will figure it out

The transcript is your raw evidence and the summary is your shortcut — but when either has gaps, you fill them in and re-run.

The best approach: read the summary, fill in the gaps manually, and re-run the prompt.

Self-Check: Question 3 of 5

What are the three sections of the post-session message, in order?

A. Summary, Feedback, Homework
B. Lesson, Assessment, Exercises
C. Notes, Assessment, Tasks
D. Review, Grade, Practice

Fixed format, every time. No exceptions.

The three sections are Lesson, Assessment, and Exercises — in that order.

Self-Check: Question 4 of 5

Who are the three recipients when the Send button is clicked?

A. The learner, the AI, and the database
B. The learner (or group), the trainer, and the training center
C. The learner, the school, and the government
D. The trainer, the admin, and the manager

Delivery + archive + audit. All three are needed.

The three recipients: the learner (or group), the trainer, and the training center.

Self-Check: Question 5 of 5

Sophie finishes a session and clicks Send — but she forgot to run the Post-Session Notes prompt first. What happens?

A. Notion warns her and blocks the send
B. The message is sent with empty content — the columns had nothing in them
C. The AI runs the prompt automatically before sending

The Send button compiles whatever is in the columns. If they're empty, the message is empty. Always run the prompt before sending.

No safety net — the Send button compiles whatever is in the columns, empty or not.

What Else Can Notion Do?

You've built a post-session workflow. But the three databases you just set up? They're the foundation for a lot more.

Next step: Download the 1-page PDF checklist and keep it open during your first real week. It walks through every step: Record → Generate → Send.

Here's what Sophie explored next:

notion.com/product/sites

Public resource database

Publish a curated library of exercises and reference materials.

notion.com/product/calendar

Notion Calendar

Sync sessions and create scheduling links for learners.

notion.com/help/formulas

Formulas

Track session hours, calculate invoicing, flag overdue follow-ups.

The post-session workflow was the starting point. Notion is the platform.

None of this is required to run the workflow you just learned. But once the databases are in place, these features plug straight in.

Key Takeaway

The three databases aren't just a workflow — they're a platform. Everything else builds on top of what you already have.

🎉 Congratulations!

You've completed the Post-Session Workflow training. You now have everything you need to run a consistent, audit-ready system — starting with your very next session.

What you've built

Three databases — Prompts, Learners, Sessions
Two AI prompts — Learner Profile + Post-Session Notes
One workflow — Record → Generate → Send

The system is ready. Now make it yours — start with your very next session.

Key Takeaway

You didn't just learn a tool. You built a system. Now use it.