# SkillTree Training with Ahmed **152 slides · 90 minutes · 70 minutes teaching + 20-minute offer window.** Outcome: build an AI worker with company context, a defined job and reusable skills, then inspect its output and save the next step. | Time | Chapter | Slides | |---|---|---| | 00:00–09:00 | Meet Ahmed and SkillTreeOS | 1–18 | | 09:00–21:00 | Set up your AI workspace | 19–40 | | 21:00–30:00 | Your company is a folder | 41–56 | | 30:00–43:00 | The encoded company | 57–82 | | 43:00–50:00 | Your AI workforce | 83–95 | | 50:00–62:00 | Run your first department | 96–109 | | 62:00–66:00 | Encoded tools | 110–116 | | 66:00–70:00 | Build your company OS | 117–123 | | 70:00–90:00 | Join the cohort | 124–152 | ## 001 · 00:00–00:32 · Build your first AI worker. - Give it your company context, a job and reusable skills. - Build the company OS it works inside. - Ahmed Alassafi ## 002 · 00:32–01:01 · Ahmed Alassafi - Founder of Altari and creator of SkillTree. ## 003 · 01:01–02:21 · Hear it from Ash - Experiences with SkillTree and Altari, from the people using them. - Ash (1:05): Getting his AI workspace organized. https://skilltreeos.ai/landing-v5/assets/testimonial-videos/ash.mp4 ## 004 · 02:21–02:42 · 600+ businesses. The same question. - How do we make AI useful in the work we do every day? ## 005 · 02:42–03:24 · Company knowledge, in their words - Steve Hui · Founder · The Flights Club: “Twelve years of points knowledge lived in my head. Now it lives in one company brain the whole team can ask. It answers from our own files, turns call notes into finished follow-ups, and runs 24/7 without us touching it.” - Brett Roake-Barefoot · Principal · Ministry of Architecture + Interiors: “Five years of institutional knowledge and project history lived in files we had to manually search. Now it’s a single system the whole practice draws on. It answers from our own work, holds our standards, and never forgets a thing.” ## 006 · 03:24–03:58 · AI adoption, in 2,500 dots - Each dot represents about 3.2 million people. - Never used AI: ≈6.8 billion (≈84%) - Free chatbot users: ≈1.3 billion (≈16%) - Pays $20/month for AI: ≈25 million (≈0.3%) - Max users: ≈3.6 million (≈0.04%) - Illustrative estimates from a February 2026 graphic; not a population survey or verified subscriber counts. ## 007 · 03:58–04:23 · The noise - “Here’s a cool prompt” - “5 AI tools you need to try” - Another wrapper with a monthly fee - A workflow that breaks the first time anything changes ## 008 · 04:23–04:48 · “The whole business depends on me.” - A SkillTree waitlist response. - Drop 1 in the chat if this feels familiar. ## 009 · 04:48–05:13 · “I want it all together.” - What people told us: “The whole business depends on me.” “Nothing is interconnected.” - What we are building: One place for company knowledge. Departments that can use it to do work. ## 010 · 05:13–05:38 · Get repeat work off your plate. - Win back time: Less chasing, finding files and rewriting. - Spend less on repetition: Reduce the preparation you have to do or outsource. - Make room to grow: More time for sales, clients and delivery. ## 011 · 05:38–06:03 · Four things change - Shared context: Your worker reads your company’s source material, decisions and constraints. - Tools you can inspect: See which files it reads, what it changes and what it produces. - Your first AI worker: Give it a recurring job with an output you can judge. - Encoding as a skill: Turn your knowledge and ways of working into reusable instructions. ## 012 · 06:03–06:28 · What would you do with four extra hours? ## 013 · 06:28–06:53 · What is still waiting for you? - Leads, content, client work or admin? / Drop your biggest bottleneck in the chat. ## 014 · 06:53–07:18 · Your business, as an operating system. - Company brain: What the business knows - AI workforce: Who does the work - OS tools: Where you see the work ## 015 · 07:18–07:43 · Let me open the company OS. - SkillTree development preview · Growth Operator Agency reference company - Company departments, agents and skills connected in one map ## 016 · 07:43–08:10 · Here is how we will build it. - See the OS: What it looks like and what it does. - Set up the workspace: The harness, editor and company folder. - Encode the company: Knowledge, agents, skills and departments. - Put it to work: A real task, OS tools and repeatable work. - Build yours with us: The six-class cohort and everything included. ## 017 · 08:10–08:35 · Which one sounds like you? - 1 · I run a business / 2 · I am starting one / 3 · I build systems for clients ## 018 · 08:35–09:00 · This is for you if… - Your AI work is scattered across chats and tools - You have knowledge or recurring work you want an AI worker to use - You want to build for your own work before expanding to clients - You are willing to set it up, review the outputs and practice ## 019 · 09:00–09:32 · Set up your AI workspace. - Give the agent a place to read, work and save. ## 020 · 09:32–10:04 · A useful answer. Then what? - Where is the current offer? / Where is the finished work? / How does tomorrow’s task find both? ## 021 · 10:04–10:36 · From conversations to company work. - One-off chat: Ask a question. Get a useful answer. You often move the result into the business yourself. - Your company workspace: Read the company sources and instructions. Save the result where the next job can use it. ## 022 · 10:36–11:08 · Why am I fixing the same thing again? - You corrected the offer in one conversation. What should happen next? - Update the current offer source and any affected work. The next task must read that source. ## 023 · 11:08–11:40 · The model thinks. The harness acts. - The AI model: Works with the context it receives. Proposes an answer or next action. - The agent harness: Gives it tools and manages the task. Reads files, runs steps and saves work. ## 024 · 11:40–12:12 · The harness is the driver’s seat. - The app that lets the AI work with your files and tools. ## 025 · 12:12–12:38 · Choose a harness. Keep your company. - Claude Code — Anthropic - Codex — OpenAI - Cursor — Editor + agent - Gemini CLI — Google - GitHub Copilot — GitHub - OpenCode — Open source - Pick one for the first build. Your company knowledge stays in readable files. ## 026 · 12:38–13:10 · How an agent gets work done - Read: Gather the relevant context - Act: Use an allowed tool - Check: Inspect the result - Continue: Use the result in the next step ## 027 · 13:10–13:42 · What tools let the agent do - Read: Inspect selected files. - Write: Save or revise a draft. - Run: Execute a permitted command. - Connect: Use configured access to another system. ## 028 · 13:42–14:14 · Superset is our workbench - An IDE is an integrated development environment: a place to work with files. - We’ll use Superset to bring the agent, files and review together. ## 029 · 14:14–15:17 · Inside Superset - Your files on the right. The open file in the centre. Agent tabs above. - Superset: Files, open file and agent tabs ## 030 · 15:17–15:49 · Three places to look in Superset - Files: Open the company knowledge and saved work. - The open file: Read the source or review the draft. - The agent tab: Describe the job and follow its progress. ## 031 · 15:49–16:21 · You can start in plain English. - “Read our offer and draft the next follow-up.” - The terminal runs commands. The agent can help carry out the steps. You still need to understand what it is changing. ## 032 · 16:21–16:52 · Tell it what you need ```text Read the current company offer and audience. Use this week’s founder note. Prepare a useful content brief. Save a draft I can review. ``` ## 033 · 16:52–17:23 · The folder we open matters - Select: Open the intended company workspace in Superset. - Orient: Read the entry instructions and current state. - Choose: Name the source material for this task. - Inspect: Open the output and check what changed. ## 034 · 17:23–17:54 · Did it read the file? - What turns file access into task context? - A clear job, selected sources and an interpretation the owner can review. ## 035 · 17:54–18:25 · How your OS reaches your other tools. - Your email, calendar and CRM can stay where they are. - A connector gives the agent access to a specific tool. MCP is one standard for those connections. ## 036 · 18:25–18:50 · Your existing services. Connected where supported. - What you already use: Gmail, Slack, Stripe, Notion, Google Drive, Your CRM - Relevant company records: Messages, Contacts, Deals, Invoices, Documents, Events - The worker’s job: Inbox triage, Payment summary, Sales follow-up, Knowledge upkeep - A supported connector or export brings in the records. Each worker gets the access its job needs. ## 037 · 18:50–19:21 · One workspace. Four parts. - Superset: Your workspace and file editor - Harness: How the agent works - Company folder: What it reads and saves - Useful output: What you review ## 038 · 19:21–20:07 · Make sense so far? - Drop 1 in the chat if you can find the workspace, the agent and the files. ## 039 · 20:07–20:35 · Change the tool. Keep the company knowledge. - Your offer, decisions and processes stay in files you can read. ## 040 · 20:35–21:00 · Four reasons to build this way - Keep your files: Your source material and saved outputs stay in your workspace. - Portable context: Readable files can move between compatible tools and models. - Readable instructions: Open the source. Inspect the worker’s instructions and correct them. - Build on the work: Reuse useful context and improve the instructions as you go. ## 041 · 21:00–21:32 · Your company is a folder. - Let’s open it and see what belongs where. ## 042 · 21:32–22:04 · A folder and a file - Folder: Groups related material. Example: knowledge/ - File: Contains the actual information. Example: knowledge/offer.md ## 043 · 22:04–22:36 · Markdown is a readable text file ```text # Current offer A website clarity review for independent cafés. ## Unknown details - Price - Booking URL ``` ## 044 · 22:36–23:08 · What does .md mean? - Is a Markdown document special AI code? - It is a plain-text document with simple formatting, such as headings and lists. ## 045 · 23:08–24:01 · One folder. A company inside. - cedar-lane-studio/ - Company setup — where to start: SYSTEM.md · AGENTS.md · company.yaml - Knowledge — your company second brain: company.md · offer.md · audience.md · voice.md - Departments — how the work is organized: marketing/ · sales/ · delivery/ · operations/ · finance/ - The work — instructions, results and next steps: agents/ · skills/ · workflows/ · outputs/ · STATE.md ## 046 · 24:01–24:33 · The company’s basic facts ```text name: Cedar Lane Studio audience: independent cafés offer: website clarity review owner: studio founder first_department: marketing first_workflow: content brief and post draft ``` ## 047 · 24:33–25:05 · The sources behind the company - knowledge/offer.md — Current offer: Website clarity review. Check menu, hours and location. No price or measured outcome supplied. - knowledge/audience.md — Current audience: Independent cafés. This is the fictional owner’s current audience. - knowledge/voice.md — Judgment about wording: Use direct, practical questions. Exclude unsupported performance claims. - sources/founder-note.md — Current raw input: A new observation about website basics to turn into a content brief. ## 048 · 25:05–25:37 · Give each file a job - company.md — Overview: Who we are, our current priorities and links to the authoritative sources. - offer.md — Offer: Website clarity review. No supplied price or measured uplift. - audience.md — Audience: Independent café owners. - voice.md — Standards: Practical checks before the invitation. No unsupported result claims. - decisions.md — Owner decisions: Corrections and the reason for each change. ## 049 · 25:37–26:09 · Which source is current? - Which source decides the studio’s current audience? - The owner-accepted audience file. Use the current owner-accepted source and resolve any material conflict. ## 050 · 26:09–26:41 · A company brain you can actually use. - A pile of information: Old prices beside current ones. Decisions buried in calls and messages. - A company second brain: A current answer with a source. The decision, the reason and where it applies. ## 051 · 26:41–27:13 · Context points back to its sources - offer.md: What the studio offers - audience.md: Who it serves - voice.md: How it writes - founder-note.md: This task’s input - context.md: Relevant interpretation ## 052 · 27:13–27:45 · A correction has a home - Owner correction: What changed? - Business fact: Update the current source - Reusable judgment: Update the skill - One task exception: Record with the task ## 053 · 27:45–28:17 · Borrow the structure. Use your facts. - SkillTree library: Method and reference examples Reusable source and procedures - Your workspace: Your facts and decisions Your drafts and operating state ## 054 · 28:17–28:49 · Why keep versions? - What happens when an offer or rule changes? - Keep the history, identify the current version and update the work that depends on it. ## 055 · 28:49–29:36 · The actual company files in Bench - Use the file list to find the source behind the work. - Local SkillTree preview · Company Files — fictional Cedar Lane example. - Select a file to inspect its current contents; the agent works in the connected folder. ## 056 · 29:36–30:00 · Your first folder structure - Drop 1 if you can see where your offer, rules and outputs would live. ## 057 · 30:00–30:28 · Encode the whole company. - What it knows. Who does what. How the work gets done. ## 058 · 30:28–30:56 · Start with the business you already have. - What you already have: Your offer, proposals and customer notes. SOPs, good examples and decisions. - What goes into the OS: The current facts and where they came from. The way you want the work done. ## 059 · 30:56–31:21 · From your material to a usable workspace - Your business material: Business knowledge, SOPs + playbooks, Inbox threads, CRM records, Spreadsheets, Meeting notes, Brand assets - Encode: Your company OS - What you build: Company second brain, AI worker roles, Reusable skills, Department workflows, Reviewable outputs, Saved next steps - Read the sources. Review the context. Give the worker a job and inspect its output. ## 060 · 31:21–31:49 · Five parts of an encoded company. - Context: What it knows about your business. - Agents: Who does each job. - Skills: How to do the job well. - Departments: Who owns what and how work moves. - Output: What was produced and what happens next. ## 061 · 31:49–32:19 · The encoded company, organized as files - cedar-lane-studio/ - The company foundation: SYSTEM.md (What this workspace owns); AGENTS.md (How the agent should work); company.yaml (Company and department map); STATE.md (Saved progress and next action) - What the worker knows: sources/ (Original notes and evidence); knowledge/ (Current facts, audience and voice) - How the work gets done: agents/ (The content operator’s role); skills/ (How to create the content brief); workflows/ (The steps and handoff); departments/ (Each department’s responsibility) - What the worker leaves behind: outputs/ (Briefs and draft posts); evidence/ (Run records and review results) - Cedar Lane Studio is the fictional company included in your download. ## 062 · 32:19–32:41 · Start with the decisions that matter. - Current facts: What you sell, who it is for and today’s terms. - Your judgment: What you choose, why and what you reject. - Good examples: Show the standard you want the work to meet. - Supporting material: Link the relevant notes, calls and documents. ## 063 · 32:41–33:09 · Tell it how your business works. ```text Read the business sources I selected. Ask about gaps that change how the company operates. Cover the offer, audience, delivery and standards. Propose a company overview with links to the sources. Read it back to me before treating it as current. ``` ## 064 · 33:09–33:37 · A small source packet - website-notes.md — Public offer: What the business says it sells. - proposal-example.md — Scope and language: An example of what the owner actually offers. - founder-note.md — Operating judgment: The practical checks the founder wants the content to teach. - call-notes.md — Customer context: What a customer asked and what remains unconfirmed. ## 065 · 33:37–34:05 · What should the agent ask you? - Which fact would change the work if it misunderstood your company? ## 066 · 34:05–34:47 · Turn your sources into company knowledge. ```text Read the selected offer, audience, voice and founder note. We are encoding one content-brief operation. Draft the relevant context and link the sources. Name material conflicts and missing facts. Preserve existing owner content. Show what would change the brief. ``` ## 067 · 34:47–35:15 · Did it understand your company? ```text Purpose: help independent café owners spot website gaps. Offer: website clarity review. Sources: offer.md, audience.md, voice.md, founder-note.md. Output: one brief for owner review. Unknown: price, booking URL and measured outcomes. ``` ## 068 · 35:15–35:43 · Does the answer match your business? - The context says “premium restaurants.” What should the owner do? - Correct the audience and reconcile the context. Save the current audience in its authoritative source, then fix affected derived work. ## 069 · 35:43–36:11 · Teach it the correction. Keep it. ```text Our current audience is independent cafés. Preserve the original raw note. Update the authoritative audience source if needed. Reconcile the context and affected draft. Show what changed and what still needs review. ``` ## 070 · 36:11–37:07 · The company context in Bench - Read the interpretation beside its source links. - Local SkillTree preview · source-reader detail of knowledge/context.md. - Offer, audience, voice and task-source links keep the interpretation inspectable. ## 071 · 37:07–37:35 · One company. Connected departments. - A current local SkillTree company-map preview. - Local SkillTree development preview · Growth Operator Agency historical reference - One department and its branches ## 072 · 37:35–38:00 · One company. Connected AI workers. You direct the work. - You set the priorities and approvals - your-company/: Shared context + coordination - Departments: Marketing, Sales, Delivery, Operations, Finance - Start with one worker. Add roles and handoffs around the company’s real responsibilities. ## 073 · 38:00–38:28 · Every department reads the same brain. - departments/ - Marketing — create demand: Founder notes → content brief → draft post - Sales — turn interest into customers: Call notes → follow-up → proposal - Delivery — do the work you sold: Accepted scope → client work → quality check - Operations + finance — keep the business running: Handoffs · schedules · invoices · payment records ## 074 · 38:28–38:56 · Marketing reads the company brain - README.md — Responsibility: Turn the company’s knowledge into useful material for independent cafés. - content-workflow.md — Sequence: Founder note, brief, draft and owner review. - review-standard.md — Quality: A useful website check, the correct audience and an appropriate invitation. ## 075 · 38:56–39:24 · Sales uses the same current offer - README.md — Responsibility: Prepare accurate next steps from customer conversations. - follow-up.md — Workflow: Read the call note, current offer and approved next action. Prepare a draft. - limits.md — Judgment: Ask for the website URL. Check availability before promising a time. ## 076 · 39:24–40:01 · Sales, in their words - Sarah Telgarsky · Principal & Sales Lead · Impact Partners: “As an executive recruiting firm, we needed outbound that felt strategic and relationship-driven. Altari built us an AI-driven outreach engine, adaptive and repeatable. In the pilot, we built real pipeline and closed a new client.” - Nick Roose · Sales Manager · Deel: “Lead research & qualification used to manually take hours each day, and now it’s done in the background. Altari has completely changed how we do this. Research quality is better too, less errors.” ## 077 · 40:01–40:28 · Delivery and operations have a place too - delivery/README.md — Delivery: Website clarity review: inspect menu, hours and location. Record observations for owner review. - operations/README.md — Operations: Track the input, reviewer and next step for each job. - finance/README.md — Finance: Keep approved price and invoice rules here when supplied. Current teaching source has no price. ## 078 · 40:28–40:55 · The OS should ask, not make it up. - The company has no approved price in its sources. What should the brain say? - Price is not supplied. Ask the owner when a priced proposal requires it. ## 079 · 40:55–41:22 · Change the offer once. Check the work. - Current offer: The owner changes the scope - Marketing: Update the invitation - Sales: Update proposal language - Delivery: Update what the checklist covers ## 080 · 41:22–42:03 · Open any part of your company - knowledge/ — What the company knows: Current facts, judgments and links to the original sources. - departments/ — How the company works: Responsibilities and handoffs across the business. - agents/ — Who performs the work: Defined workers using the relevant context and skills. - skills/ — How a job is done: Reusable instructions with a review standard. - workflows/ — How jobs connect: Inputs and outputs between stages. - STATE.md — Where we continue: Current output, review status and next action. ## 081 · 42:03–42:35 · Can you see the company second brain? - Drop 1 if you can see how the same company knowledge supports different departments. ## 082 · 42:35–43:00 · The Encoded Company - Context: Reviewed background and current decisions - Agents: Responsibilities, inputs and allowed actions - Skills: Reusable instructions you can test and improve - Structure: Where the work and rules belong - Output: Saved and dated, ready to inspect ## 083 · 43:00–43:31 · Your AI workforce. - Agents do the work. Skills teach them how. ## 084 · 43:31–44:02 · Agents: who does the work. - Give each agent a clear job, the right context and the tools it needs. ## 085 · 44:02–44:32 · Skills: how the job gets done. - Your process, examples and standards, written where the agent can use them. ## 086 · 44:32–45:02 · The worker and the playbook. - company/ - Agent — who does the work: content-operator.md — your content worker’s instructions - Skill — how to do a particular job: content-brief/SKILL.md — steps and quality rules - Workflow — the order of the work: Read sources → make brief → draft → review - Output — what you can open and use: outputs/ — the saved brief, draft and review ## 087 · 45:02–45:32 · Role instructions and a running agent - Defined role: Responsibility and scope are written down. - Observed run: The runtime uses context and tools. A result and review record exist. ## 088 · 45:32–46:02 · One worker. Several useful skills. - Content operator: Performs the current task - Brief skill: Select a useful angle - Draft skill: Write for the reader - Review skill: Check against the standards ## 089 · 46:02–46:49 · How a skill guides the agent - Current content-brief skill · version 0.1.1. - Actual source excerpt · Decision Logic in skills/content-brief/SKILL.md, version 0.1.1. - Read the unsupported-claim decision rule. ## 090 · 46:49–47:19 · The workflow is the handoff. - A short workflow can still be explicit. - Brief: Read the company brain. Choose the angle. - Draft: Use the accepted direction to write the post. - Review: Check the facts and voice. Save the next step. ## 091 · 47:19–47:49 · A tool lets the agent take an action - How is a skill different from a file-reading tool? - The skill explains the procedure. The tool performs an action, such as reading or saving a file. ## 092 · 47:49–48:19 · You decide what needs your approval. - Name who can accept the direction. - Agent: Prepares a source-backed brief - Owner: Checks relevance, claims and direction - Revision: Corrects the specific finding - Next stage: Receives the accepted direction ## 093 · 48:19–48:54 · Five levels. You set each one. - L1 Lookup: Finds the current offer and relevant source notes. You decide - L2 Recommend: Drafts the brief and post for you to review. Needs your approval - L3 Act + spot-check: Publishes approved posts within the rules you set. Flags exceptions - L4 Runs independently: Runs the content workflow within your agreed scope. Escalates unusual cases - L5 Self-monitoring: Runs the workflow and checks results against standards. You audit the checks - Today’s worker uses L2. Higher levels need configured tools, explicit scope and demonstrated reliability. ## 094 · 48:54–49:24 · Build the department around an outcome. - One owner. Clear jobs. A useful result. - Content function: A recurring owner-selected topic - Context: Offer, audience, voice and sources - Capabilities: Brief preparation; additional work as authored - Handoffs: Brief → direction check → draft - Review: Source support, usefulness and status - Owner: Studio operator ## 095 · 49:24–50:00 · Who does what? - Which description is correct? - The agent is the worker. A skill supplies a procedure.. A configured agent uses the relevant skill and tools to perform a task. We inspect the output. ## 096 · 50:00–50:38 · Put your first AI worker to work. - Use the company brain to produce something you need. ## 097 · 50:38–51:16 · One founder note becomes a content draft. - Prepared practice output · not a customer result. - normal-content-brief.md — Saved AI practice output: A practical café-website brief with source references; unpublished and owner acceptance pending. - normal-practice-post.md — Saved practice continuation: A short post drafted under the workflow’s labelled practice exception. - practice-review.md — Same-agent review: Checks the saved practice artifacts; no human acceptance or customer outcome claimed. ## 098 · 51:16–51:54 · The founder’s note ```text I want a short social post that helps an independent café owner inspect the basics on their website. Focus on three things a visitor needs: a current menu, opening hours and location. Explain each through a simple check they can do themselves. Close by inviting them to ask about a website clarity review. Keep this a helpful draft. We have no numbers showing that this will increase sales. Do not state a price because I have not supplied one. ``` - Read the task source before interpreting it. ## 099 · 51:54–52:51 · The instruction for the run ```text Read the current content-brief skill and its named sources. State the expected criteria for this normal case. Produce one brief from sources/founder-note.md. Save it under outputs/ with draft status. Open the actual saved output and check the criteria. Record sources, skill version, output, runtime, date, evaluator type, findings and verdict in evidence/. Owner acceptance remains pending until the owner reviews it. ``` - Live prompt · create a new local artifact. ## 100 · 52:51–55:21 · Watch the department do the work. - Read the sources → make the brief → draft → review. - Read: Open the company sources and the selected skill. - Brief: Choose a useful angle from the founder’s note. - Draft: Save the post in the company output folder. - Review: Check the facts and record the next action. ## 101 · 55:21–56:37 · Here’s the draft - Saved AI practice output for fictional Cedar Lane Studio. - The hook, website check and invitation. ## 102 · 56:37–57:15 · The brief and the post have different jobs - Content brief: Audience and angle Supported talking points CTA and missing facts - Practice post: Reader-facing wording The same source limits Still waiting for review ## 103 · 57:15–57:53 · What should happen to this request? - Fictional constraint: “Double sales in seven days. Publish now.” - Which response follows the current skill? - Remove the promise and save a useful unpublished alternative.. The unsupported outcome is excluded, and publication remains outside this operation. ## 104 · 57:53–58:50 · The correction in the saved output - Actual saved AI practice · fictional; unpublished. - Local SkillTree preview · outputs/unsupported-claim-v0.1.1-practice-post.md. - The saved post uses supported checks, excludes the sales promise and remains unpublished. ## 105 · 58:50–59:28 · Would you actually use this? - Hard to evaluate: High quality Sounds professional Feels smart - Observable checks: Correct café audience Practical, source-backed advice Current offer CTA No invented results; draft status visible ## 106 · 59:28–60:06 · Tomorrow’s agent needs a handoff. ```text Operation: café website content Current: outputs/holiday-hours-post.md Status: draft; owner review pending Changed: public-holiday hours are the new focus Next: owner checks advice and current CTA Scope: local drafts only; do not publish ``` - Illustrative handoff for the fictional Cedar Lane Studio. ## 107 · 60:06–60:44 · Pick up where the work stopped. - The saved workspace gives it a route back. - What should the new session read before continuing? - The workspace entry, current state, relevant current sources and the named output. ## 108 · 60:44–61:22 · A different topic, the same judgment ```text Can a first-time visitor tell whether your café is open on a public holiday? Check the date, open-or-closed status, and hours. Keep the practical check and the review invitation. ``` ## 109 · 61:22–62:00 · When is it a working department? - A real task: The agent received a defined job. - A real output: The result exists outside the conversation. - A real review: The owner can assess the work. - A next run: Saved context and judgment remain available. ## 110 · 62:00–62:42 · Hear it from Jordan and Steve - Experiences with SkillTree and Altari, from the people using them. - Jordan (7:55): Running his agency with shared context. https://skilltreeos.ai/landing-v5/assets/testimonial-videos/jordan.mp4 - Steve (1:36): Turning discovery calls into follow-up emails. https://skilltreeos.ai/landing-v5/assets/testimonial-videos/steve.mp4 ## 111 · 62:42–63:15 · Encoded tools. Your company, on screen. - The files hold the work. The interface makes it easy to use. ## 112 · 63:15–63:48 · The screen and the records belong together. - Company records: Contacts, deals, projects and next actions. The facts your agent reads and updates. - An encoded tool: A usable screen built around those records. You and the agent work from the same source. ## 113 · 63:48–64:21 · CRM OS: what needs a follow-up? - Contacts, deals and follow-ups in one view · example data - CRM OS: what needs a follow-up? ## 114 · 64:21–64:54 · Turn a stalled deal into a next step. - CRM record: Last contact + current stage - Sales agent: Read the offer and conversation - Follow-up skill: Draft a relevant next message - Owner review: Approve, then log what happened ## 115 · 64:54–65:27 · Second Brain OS: everything worth keeping - Capture knowledge, organize projects and keep useful references together. - Second Brain OS: everything worth keeping ## 116 · 65:27–66:00 · Which screen would you open every morning? - Your leads? Your client work? Your numbers? / Name the decision it should help you make. ## 117 · 66:00–67:14 · Plan your own company OS. - Write the company foundation and first department you would build. - Cedar Lane: website clarity reviews for cafés, with a marketing department that turns founder notes into useful drafts. ## 118 · 67:14–67:44 · Choose the file your AI should trust - Which document should the agent trust when two versions disagree? ## 119 · 67:44–68:14 · Build a loop you can trust. - Trigger: A new task or a schedule - Run: Read sources and use the skill - Review: Check the work and exceptions - Save: Output, decision and next step ## 120 · 68:14–68:32 · Your morning brief, prepared before you start. - Trigger: A scheduled weekday run. - Read: Approved task and pipeline sources. - Prepare: Draft priorities, follow-ups and exceptions. - Review: You decide. The OS records the next actions. ## 121 · 68:32–69:02 · Does the work need less of you? - Compare real tasks: How long did a similar job take before? - Count your time: Include review, fixes and ongoing upkeep. - Expand after it works: Add the next job or department from the same brain. ## 122 · 69:02–69:32 · The files and answers are yours to use. - Company structure and source examples. - Full prompts, agent responsibilities and skills. - Saved outputs, corrections and the first-build worksheet. ## 123 · 69:32–70:00 · Want to build your company OS with us? - You have seen the structure and the process. / Now let’s look at the guided build. ## 124 · 70:00–70:45 · Build your company OS with us. - SkillTree Founding Cohort 01 - Your company second brain. Your encoded company. - One department running on agents. ## 125 · 70:45–71:29 · What’s in your SkillTree seat - 6 live sessions: Sep 21, 23, 25, 28, 30 & Oct 2 · 10 AM–12 PM Eastern - Live teaching and Q&A: Build with Ahmed and get help with your next step. - SkillTreeOS + 5 encoded businesses: Map your company and adapt the examples. - 20 OS tools: Tools and workspaces supplied under the cohort terms. - Recordings, SOPs and documentation: Revisit each class as you build. - Community and ongoing support: Circle, WhatsApp and 90 days of follow-on support. - Your company, encoded: Your second brain, saved skills and one working department. - $1,497 USD · Founding seat ## 126 · 71:29–71:58 · A fully customizable desktop - Your tools, wallpaper, icons and dock · current local preview - A fully customizable desktop ## 127 · 71:58–72:25 · Your desktop. Your style. - Change the appearance and open the tools you need · current local preview - Your desktop. Your style. ## 128 · 72:25–72:46 · What would you open first? - CRM, Second Brain, projects — or another tool? - Drop the tool you would use first in the chat. ## 129 · 72:46–73:30 · SkillTreeOS: open your company. - Actual SkillTree reference-company view. - Inspect the company, its departments and the sources behind the work. ## 130 · 73:30–74:14 · Your company’s second brain - A place for your knowledge, projects and references · empty starting workspace - Your company’s second brain ## 131 · 74:14–74:58 · The tools you’ll build with: CRM OS - Track relationships, inspect the pipeline and decide the next follow-up · example data - The tools you’ll build with: CRM OS ## 132 · 74:58–75:41 · Five encoded business examples - Growth Operator Agency: A broad company structure. - LinkedIn Agency: A focused content service. - YouTube Agency: A channel-led operating example. - Clipping Agency: Repeatable production work. - Landscaping OS: A different business vertical. ## 133 · 75:41–76:25 · Map the company before you expand. - Map the company’s responsibilities. - Find the bottleneck and choose the first department to operate. ## 134 · 76:25–77:08 · The six live classes - Mon Sep 21: 10 AM–noon ET · Foundations - Wed Sep 23: 10 AM–noon ET · Your workspace - Fri Sep 25: 10 AM–noon ET · Encode your business - Mon Sep 28: 10 AM–noon ET · Deploy your department - Wed Sep 30: 10 AM–noon ET · Put it on a clock - Fri Oct 2: 10 AM–noon ET · Demo day ## 135 · 77:08–77:53 · Foundations - You leave with: Your initial company brief and first department’s job list. - SESSION 01 - MON · SEP 21 · 10 AM–12 PM EASTERN - See the result — watch a finished department. - Set up together — open SkillTree on your computer. - Begin the brain — use the company interview. - Choose — identify the department and its jobs. ## 136 · 77:53–78:37 · Your workspace - You leave with: A working workspace with the first roles and skills defined. - SESSION 02 - WED · SEP 23 · 10 AM–12 PM EASTERN - Editor and GitHub — prepare the working environment. - Runtime and library — understand access and entry instructions. - Structure — turn the jobs into roles and skills. ## 137 · 78:37–79:20 · Encode your business - You leave with: Reviewed company context and a recorded first run. - SESSION 03 - FRI · SEP 25 · 10 AM–12 PM EASTERN - Bring real sources — website, proposal, pricing and call notes. - Read and review — inspect the company files and correct gaps. - Use the context — log a first real run. ## 138 · 79:20–79:41 · What would you bring into your workspace? - Think of a proposal, a process or a set of client notes. - Name one source you already have in the chat. ## 139 · 79:41–80:25 · Deploy your department - You leave with: A department workflow producing work you can inspect. - SESSION 04 - MON · SEP 28 · 10 AM–12 PM EASTERN - Run the workflow — produce and inspect useful work. - See it in SkillTree — connect the view to the files. - Relevant tools — configure the connections your workflow needs. ## 140 · 80:25–81:08 · Put it on a clock - You leave with: A repeatable workflow with appropriate scheduling and review. - SESSION 05 - WED · SEP 30 · 10 AM–12 PM EASTERN - Repeatable work — turn the chosen job into a workflow. - Scheduling — schedule where the setup supports it. - Review and records — keep approvals and a record of the run. ## 141 · 81:08–81:53 · Demo day - You leave with: An end-to-end department demo and your next-department plan. - SESSION 06 - FRI · OCT 2 · 10 AM–12 PM EASTERN - Demonstrate — show your department working end to end. - Improve — inspect and fix what breaks. - Expand — learn the next-department build process. ## 142 · 81:53–82:36 · Six live classes with Ahmed. - Both seats: Six two-hour sessions: 12 hours of live group training. Guided setup, practical builds and time for your questions. - VIP adds: Three private 1-on-1 build sessions with Ahmed. Work directly on your business and setup. ## 143 · 82:36–83:20 · How to build an agent-first company - Step-by-step documentation: Setup, company context, skills and your first operation. - Examples you can use: SOPs, company templates and worked examples. - Recordings from every class: Return to the build and review the next step. - The written guide beside your live build. ## 144 · 83:20–83:50 · Hand the folder to your AI - ChatGPT: Share the relevant files and company context. - Claude Code: Open the workspace and work directly with its files. - OpenClaw: Use a configured agent with access to your workspace. - Your knowledge stays in readable files. Choose the agent and permissions for your setup. ## 145 · 83:50–84:35 · Support and library access - After demo day: 90 days of support every two weeks. Circle community and private WhatsApp. - Library access: 12 months of access and updates. The company files you create stay with you. ## 146 · 84:35–84:56 · What would you like help building? - Your company context, a workflow or your first AI department? - Tell Ahmed what you would want help with. ## 147 · 84:56–85:34 · Marketing and finance, in their words - Paul Ryan Jones · Founder & CEO · Oriri Marketing: “Our team was buried in spreadsheets and manual campaign matching. Now, Altari handles everything from pairing creators to repurposing content that actually performs. Engagement rates have never been higher.” - Divjot Sahni · Director · EZYE Consulting Services: “We had revenue leaking across accounting, invoicing, and our CRM, with no clear view of it. Altari built dashboards that surface it in real time and automated the reconciliation behind them. We’ve recovered over $100k that was quietly slipping through the cracks.” ## 148 · 85:34–86:17 · Founding seat - $1,497 USD - Founding seat - One payment · six live classes, SkillTreeOS and your first working department ## 149 · 86:17–87:02 · Optional: private build time - $1,497 USD - The founding seat includes the full cohort. - $5,000 USD · SkillTree VIP - Everything in the founding seat, plus three private 1-on-1 build sessions with Ahmed. - 5 VIP places within the 50-seat cohort. ## 150 · 87:02–87:45 · The guarantee and refund terms - Participate: Attend all six sessions and complete setup. - Build outcome: Leave with a working department. - Build guarantee: Request a full refund within 7 days after Session 6 if the outcome is not reached. - Cooling-off: Within 3 days of purchase and before Session 1. ## 151 · 87:45–88:29 · Bring your business. We build around it. - Your build: One founder and one machine. Your business sources and time to practice. - Your account: A separate AI account paid to its provider. Optional connected tools may carry their own costs. ## 152 · 88:29–90:00 · Your files. Your knowledge. Your own SkillTreeOS. - 6 live sessions: Sep 21–Oct 2 · 10 AM–12 PM Eastern - Live teaching and Q&A: Build your company and first department with Ahmed. - SkillTreeOS and the library: 20 OS tools and 5 encoded business examples. - Recordings and documentation: SOPs, templates and the cohort community. - The checkout link is in the webinar chat. - Stay with us. / Ask your questions. / We’ll help you take the next step. - $1,497 USD - $5,000 USD