Move Beyond Prompts With AI Agent Workflows

Welcome

Your role.
Your task.
Your goal.

Introduce yourself in the chat

In the chat

What AI tools are
you using today?

Type the tool names, or “none yet.”

In the chat

Who are you looking
for AI help for?

A: Myself
B: My team
C: Both

Reply A, B, or C.

Who we help

We help leaders and teams
turn one-off AI experiments
into repeatable workflows.

Today’s roadmap

One task.
One repeatable workflow.

Choose · Build · Reuse

Working together

Bring the work.
Keep your judgment.

Pick one task.
Share approved examples.
Ask questions. Check outputs.

Why this matters now

AI can now automate
multi-step knowledge work.

Research · Analysis · Writing · Coding

We need workflows we can repeat and review.

Template slide 8 of 42. WHY. AI can perform multi-step knowledge tasks, including research, analysis, writing, and coding, using tools and connected workflows. This is not a claim that every knowledge task can be fully or reliably automated. Capabilities and limitations are supported by https://openai.com/index/introducing-chatgpt-agent/ and https://www.anthropic.com/research/anthropic-economic-index-january-2026-report, reviewed October 12, 2026. Explain why the workshop matters: leaders set direction and quality standards; teams need a shared way to provide context, perform the work, and review the result instead of rebuilding each experiment. We begin with one recurring task and leave with a workflow brief, checked draft, and quality checklist. Expanded presenter copy from the previous version: Why this matters now AI can now automate multi-step knowledge work. Research · Analysis · Writing · Coding One-off experiments leave each person starting over. Teams need shared workflows with clear inputs. Leaders set the standard and review the results. Today, we start with one recurring task.

The problem

Everyone wants AI.
Most are stuck experimenting.

Tool hype

High expectations.
Disappointing results.

News fear

Alarming headlines.
Hesitation.

Template slide 9 of 42. THE PROBLEM. This problem framing comes from the workshop organizer: leaders and teams want to use AI, but their one-off experiments are not becoming repeatable workflows. Tool marketing can create unrealistic expectations about instant results; disappointing real-world attempts reduce confidence. A doom-and-gloom news cycle can create fear and hesitation before people have a practical way to try the work. Invite the room to notice which pressure they recognize. Connect back to the workshop's shared process: one recurring task, clear inputs, a checked draft, and human review. Expanded presenter copy from the previous version: The problem Everyone wants to use AI. Teams are stuck experimenting. One-off experiments aren’t becoming repeatable workflows. Tool marketing hype Tool promises set unrealistic expectations. Real work falls short. News doom and gloom A constant cycle of alarming headlines fuels fear. Teams hesitate to act.

Frustration 1

“I’m not sure
what to say to AI.”

The blank prompt

How much context?

How specific?

What format?

Frustration 2

“I’m not sure
which AI tool to use.”

Template slide 11 of 42. Frustration #2. The organizer's second frustration is using the wrong AI tool for the job. Tools and models differ in capability, and the configuration matters: the tool must support the kind of work, the model must be able to handle its complexity, and the setup must provide the relevant context, permitted access, and instructions. A disappointing result may reflect a mismatch in any of these, rather than the wording of the prompt alone. Keep this slide on recognizing that mismatch; it is not a vendor ranking or a claim that one model is best for every task. Expanded presenter copy from the previous version: Frustration 2 “Is this the right AI for the job?” AI tools and models aren’t equally capable. How you set them up matters. The fit and setup matter Tool Does it support the work? Model Can it handle the task’s complexity? Setup Does it have the context, access, and instructions?

Frustration 3

“Why can’t I get
consistent
results?”

The fear

“Why bother?
But what if
we fall behind?”

AI feels wasteful.
Falling behind feels worse.

The opportunity

Automate work with AI agents.
Save time. Make money.

Without complicated prompt engineering,
leaking sensitive information, or wasting tokens.

Template slide 14 of 42. THE OPPORTUNITY. The organizer's opportunity is to automate recurring work with AI agents to save time and make money. Connect saved time to work that creates value, such as serving customers, improving delivery, or growing revenue. These are goals to work toward, not guaranteed savings or earnings. The practical foundation remains the right tools, clear context, and quality checks, so we can trust the results and reuse the process instead of starting from scratch each time. Match the tool and model to the task, preserve the context the work needs, and check outputs against a human-defined quality standard. Today's workshop begins with one recurring task, a workflow brief, a checked draft, and a quality checklist. Expanded presenter copy from the previous version: The opportunity Turn recurring manual work into repeatable AI workflows. Right tools · Clear context · Quality checks Get results we trust, without starting from scratch every time.

What we want · 1

Know what
to say to AI

Without complicated
prompt engineering.

What we want · 2

Know the right AI tool
for the job

Without leaking sensitive information
or wasting tokens.

Template slide 16 of 42. Answer the second frustration: knowing the right AI tool for the job without leaking sensitive information or wasting tokens. Tool choice includes model capability, approved input access, permissions, and review controls. Connect the supporting line to choosing an approved setup, sharing only permitted inputs, and keeping context and agent runs scoped to the task. These are operating goals, not a guarantee that a tool prevents every disclosure or eliminates all token costs. The four gears describe levels of usage; move only as far as the task needs. Expanded presenter copy from the previous version: What we want · 2 Choose the right tool for the work A capable model, access to the right inputs, and a setup that fits the task. Set up the handoff workflow Capability Can it work with our notes? Access Are these inputs approved for it? Control Can we review the draft before sharing?

What we want · 3

Get consistent results
we can trust

Repeatable workflows
that improve with feedback.
Without starting from scratch every time.

Template slide 17 of 42. Answer the third frustration: getting consistent results we can trust through repeatable, self-improving AI workflows without starting from scratch every time. Save the brief, context, output format, and quality checklist so the next run begins with an established process. Self-improving means reviewed corrections and feedback feed into future runs; the owner checks changes against the quality standard. Trust comes from checking the output against explicit criteria. Repeatable describes a process with stable inputs, instructions, and review; it does not mean identical or error-free model outputs. Expanded presenter copy from the previous version: What we want · 3 Get results we can trust A shared quality standard makes each draft easier to check, improve, and reuse. Define a useful handoff Accurate Every action traces to the notes. Complete Missing details stay visible. Reusable The same format works next week. Match the live AI 30 offer’s practical feedback loop: review the result, capture corrections, save useful lessons in the instructions or checklist, and check the next run. The participant controls saved changes. The slide’s concise feedback wording introduces this idea before practice and the invitation.

What we’re working toward

AI agents do
the recurring work

So our team can save time
and make money.

Template slide 18 of 42. Connect the desired outcomes to AI agents doing recurring work so our team can save time and make money. Give the team more capacity for serving customers, improving delivery, and growing revenue. Start with one recurring task, a saved process, and a clear owner. Today we build a first reviewed workflow draft; adoption needs a repeatable trigger, review, and evidence from later runs. Track time spent, corrections, and review effort to evaluate the workflow’s usefulness. Expanded presenter copy from the previous version: What we’re working toward A workflow the team actually uses One recurring task. A saved process. A clear owner. A useful result each time. Make reuse part of the work Trigger The weekly operations meeting ends. Owner One person runs and reviews the handoff. Measure Track corrections and review effort.

AI for
recurring work

Healthcare case study

Template slide 19 of 42. Introduce the published Technical Leaders healthcare case study: https://www.technical-leaders.com/healthcare-ai-case-study, verified October 13, 2026. Connect the aspiration to the case’s foundation: shared prompts, tool guides, quality checks, and department workflows. The published evidence supports the workflow and adoption story, not deployed-agent results, task-level outputs, measured savings, or measured revenue gains. The optional fictional handoff in the exercise speaker notes is separate teaching material and is not a reconstruction of this client’s work. Use the case-study divider to hand off from the aspiration into the published workflow evidence. The time-and-money goal was stated on the preceding slide; keep this divider focused on introducing the example.

Published healthcare case study · Team training

The foundation:
repeatable workflows

30 leaders · 3 weeks

Operations automation · Lead scoring · Content generation
Shared prompts, tool guides, and quality checks.

Healthcare case study

Template slide 20 of 42. Source: https://www.technical-leaders.com/healthcare-ai-case-study, verified October 13, 2026. Published first-party facts: 30 participants in a private three-week program; department workflows including operations automation, documentation, quality assurance, lead scoring, and content generation; practice reviewing and validating outputs; and a shared company AI playbook with prompts, tool guides, and quality checks. Explain the connection to agents: saved instructions, approved inputs, consistent output formats, and review criteria give recurring work a reusable foundation. Those workflows can become agent tasks with a trigger, an owner, and feedback from reviewed results. Operations workflows target time spent on recurring work; sales and marketing workflows support revenue-generating work. Saving time and making money is the goal of extending this foundation into agents. The published page does not report deployed-agent results, measured time savings, or measured revenue gains. Do not attribute the illustrative handoff walkthrough to this client. Expanded presenter copy from the previous version: Published healthcare case study From scattered experiments to a shared playbook Before Different AI understanding. No common framework. The work 30 leaders. Three weeks. Department workflows. Human review Practice prompting, reviewing outputs, and validating results. What changed A shared playbook with prompts, tool guides, and validation checklists. technical-leaders.com/healthcare-ai-case-study This published case describes team training and workflow foundations. It is not a case study of the newly updated individual AI 30 offer or evidence of its 30-day outcomes.

What recurring work could
AI agents do for your team?

To save time and make money.

Three foundations
for AI agents

Right tools
Clear context
Quality checks

Outline adapted from The+Million+Dollar+Workshop.pdf, physical page 22: The Client Machine Workshop slide overview. Source overview slides 026–029, 058–061, and 104–109 use a short set of big ideas before the how. Introduce the same three foundations for AI agents doing recurring work so teams can save time and make money. Right tools: match capability, approved access, and setup to the job, with efficient use of tokens. Clear context: save the task, inputs, instructions, and expected output so the agent can reuse the workflow. Quality checks: define good work, review results, and feed reviewed corrections into future runs. These three connect directly to the frustrations, desired outcomes, and shared playbook case. The next slide shows how to choose the smallest AI gear for the job; then apply the five-step method to one recurring task. The later implementation steps add ownership, usage mode, and measurement. Keep all five existing teaching slides available. Expanded presenter copy from the previous version: Three big ideas The foundations of a repeatable workflow Right tools Match the tool and model to the work. Clear context Save the inputs, instructions, and expected output. Quality checks Define good work and review against that standard. The three foundations remain the complete visible takeaway. Connect their value to the recurring task participants identified, then move into selecting the appropriate gear.

The 4 Gears of AI Usage

RIGHT AI. RECURRING WORK. SAVE TIME. MAKE MONEY.

1 · ChatAnswer a question
2 · AssistantFinish a deliverable
3 · AgentsComplete an objective
4 · LoopsSustain an outcome

Foundation Human Intelligence

Sets direction · defines good · reviews the work · owns the outcome

Your building process

Five steps to repeatable
AI agent workflows

1. Install Capabilities
2. Setup
3. Assign
4. Equip
5. How Might We AI Formula

Step 1 of 5

Install
Capabilities

Download the desktop app
and mobile apps.

Animated official Claude download screens showing the desktop download and mobile app entry. — Claude app download ↗

Template slide 23 of 42. Step 1: Install Capabilities. Download and sign in to the desktop and mobile apps for the AI tools participants will use. Give them access to their AI workspace where recurring work happens. Keep app-specific demonstrations appropriate to the tools available in the room. The following step connects the workspace to business tools and data. Visual aid: Current official download-page captures from October 13, 2026, cropped to desktop and mobile entry points. No download or installation was started. Use the equivalent official app page for the participant’s approved tool. The GIF is an animated sequence of authentic screenshots, not a continuous recording. The participant’s selected task remains the main exercise. Pause holds the current screenshot; Play resumes the sequence. Reduced-motion viewing starts with the static poster. Open the image for the original full-size GIF. All source frames and crop bounds are recorded in docs/sources/ai-workshop-visual-aids.json.

Step 2 of 5

Setup

Connectors for business tools,
internal data, and third-party systems.

Animated public Claude connector directory screenshots, showing business-tool options and a Google search. — Claude connector directory ↗

Template slide 24 of 42. Step 2: Setup. Use connectors to make the relevant business tools, internal data, and third-party systems available to the AI workspace. Identify which sources the recurring work needs and use approved access and permissions. For the weekly handoff example, start with the approved meeting notes and team context. These sources become the SOURCES part of the AI Formula. Visual aid: Current public connector-directory captures from October 13, 2026. This is browsing and searching, not an authenticated setup recording. No connector was added or given permissions. Choose approved sources and inspect the provider’s access requirements before connecting. The GIF is an animated sequence of authentic screenshots, not a continuous recording. The participant’s selected task remains the main exercise. Pause holds the current screenshot; Play resumes the sequence. Reduced-motion viewing starts with the static poster. Open the image for the original full-size GIF. All source frames and crop bounds are recorded in docs/sources/ai-workshop-visual-aids.json.

Step 3 of 5

Assign

Pick the right recurring job
for AI to do.

Step 4 of 5

Equip

Pick the right AI tool
for the job.

Step 5 of 5

How Might We
AI Formula

“How might we VERB OUTCOME
without CONSTRAINTS
based on SOURCES”

Animated real Longhand formula explorer selecting VERB, OUTCOME, CONSTRAINTS, and SOURCES and explaining each part. — Longhand AI Formula ↗

Template slide 27 of 42. Step 5: How Might We AI Formula. Exact wording from /longhand-ai, public/longhand-ai/index.html: “How might we VERB OUTCOME without CONSTRAINTS based on SOURCES”. VERB names the work, OUTCOME defines done, CONSTRAINTS state the boundaries, and SOURCES identify the material to use. Translate the assigned job into this plain-language brief. Handoff example: How might we summarize the weekly operations handoff into actions, owners, due dates, and open questions without inventing missing details based on the approved meeting notes and team roles? Check every action and date against the source; leave missing details as “Not stated.” Save the brief, expected format, and review checklist for reuse. The later exercise and implementation slides cover checking, ownership, fresh inputs, scheduling, and continuous improvement. Visual aid: Real clicks in the existing Longhand AI Formula explorer, captured October 13, 2026 from the local standalone page public/longhand-ai/index.html. The complete formula remains visible on the slide. Each screenshot shows the selected part and its explanation; these are source UI states, not an invented interface. The GIF is an animated sequence of authentic screenshots, not a continuous recording. The participant’s selected task remains the main exercise. Pause holds the current screenshot; Play resumes the sequence. Reduced-motion viewing starts with the static poster. Open the image for the original full-size GIF. All source frames and crop bounds are recorded in docs/sources/ai-workshop-visual-aids.json.

Which step do you need
to complete next?

Let’s build one

Your turn

Pick a workflow
to agentify

Bring your own.
Or choose a library example.

Fortune 100 Prompt Library

Animated live prompt-library screens showing Start Here, a weekly-project search, and the Weekly Project Status workflow. — Prompt library walkthrough ↗

Template slide 32 of 42. Ask each participant to choose one workflow to agentify: a recurring task they know well, or a useful example from https://www.technical-leaders.com/fortune-100-prompt-library. The live library was verified October 13, 2026; its Start Here section offers meeting notes to next steps, rough notes to an email, weekly project updates, and explaining a task. Let participants browse by their task instead of choosing the same example for everyone. Pick work with available approved inputs and a result they can check today. Open the selected workflow and use its required inputs, reusable instructions, and output checks as a starting point. Keep the source page open in a separate tab and return here to write the formula. The slide link opens the library’s Start Here section and carries workshop attribution. The optional fictional handoff from the former walkthrough is preserved in the next practice slide’s notes and in the saved pre-review deck. Visual aid: Real navigation in the live public Fortune 100 Prompt Library on October 13, 2026: Start Here, search for weekly project, and open Weekly Project Status. No signup or AI request was submitted. Participants can pick a different recurring task; this visual demonstrates how to find the instructions and required inputs. The GIF is an animated sequence of authentic screenshots, not a continuous recording. The participant’s selected task remains the main exercise. Pause holds the current screenshot; Play resumes the sequence. Reduced-motion viewing starts with the static poster. Open the image for the original full-size GIF. All source frames and crop bounds are recorded in docs/sources/ai-workshop-visual-aids.json.

Your turn

Write your
AI Formula

Use the workflow you chose.

“How might we VERB OUTCOME
without CONSTRAINTS
based on SOURCES”

VERB: Name the action.
OUTCOME: Describe the result.
CONSTRAINTS: State what to avoid.
SOURCES: List approved inputs.

Template slide 31 of 42. Use the workflow selected on slide 32, whether the participant’s own recurring task or a prompt-library example. Ask participants to write the four parts in their own notes, then join them into one sentence using the exact How Might We AI Formula from /longhand-ai: “How might we VERB OUTCOME without CONSTRAINTS based on SOURCES”. VERB: name the specific action, such as summarize, compare, draft, or build. OUTCOME: describe the useful result and format so a person can tell when the work is done. CONSTRAINTS: state what the AI must avoid, such as inventing details, changing approved facts, or exceeding the task’s scope. SOURCES: list the approved notes, documents, examples, or systems it should use. Keep the selected job, tool setup, owner, and review checklist alongside the formula. When adapting a library example, retain the necessary context and output checks. Read the completed sentence once to confirm that it asks for the intended result from the selected sources. The next slide runs, checks, and improves this formula.

Your chosen AI tool

Test your workflow

Run: Paste your formula + approved inputs.
Check: Compare output with sources, outcome, and constraints.
Improve: Save corrections. Run again.

Animated actual Claude draft from fictional meeting notes, focusing on conflicting dates and missing owner and date values. — Claude draft checks · fictional notes ↗

Template slide 33 of 42. Use the formula written on slide 33 for the workflow selected on slide 32. Run: open the AI tool they chose, paste the completed How Might We AI Formula, and supply its approved notes, documents, or connected sources. Run it and inspect the actual output. Check: trace facts to the sources, compare the result and format to the requested outcome, and confirm that it respects the constraints. Flag missing information instead of inventing it. Improve: explain the specific gap and save its correction in the formula, inputs, tool setup, or checklist. Run it again to check whether the change helped. Capture reviewed corrections in the workflow so the next run benefits from them. Save the formula, source references, checked draft, and quality checklist before discussing what worked. Give participants hands-on time; if tool access or inputs are unavailable, finish the formula and use the illustrative fallback to practice the checks. Optional preserved fictional handoff demonstration: “How might we summarize actions, owners, and dates without inventing details based on these approved notes?” Source notes: Maya: send the checklist Thursday. Leo: review access exceptions. Illustrative draft, authored for teaching: Action / Owner / Due; Send checklist / Maya / Thursday; Review access exceptions / Leo / Not stated. Missing dates stay “Not stated.” Ask when Leo’s review is due; a person supplies missing information rather than the model inventing it. These notes and the illustrative draft are fictional teaching material, not a client output or evidence from a live model run. If demonstrating live, run the formula and review the actual output. Keep this fallback distinct from participants’ chosen workflows and the published healthcare case. Visual aid: A sequence from an actual Claude response to fictional N1/T1/N2 classroom notes, captured October 9, 2026. It shows the complete checked draft, then focuses on the conflicting date and missing owner/date. These are factual screenshots and focused crops, not a client result, a current new model run, or a substitute for the participant’s workflow. Use the visual to discuss output review rather than introduce another prompt format. The original response screenshot remains unchanged. The GIF is an animated sequence of authentic screenshots, not a continuous recording. The participant’s selected task remains the main exercise. Pause holds the current screenshot; Play resumes the sequence. Reduced-motion viewing starts with the static poster. Open the image for the original full-size GIF. All source frames and crop bounds are recorded in docs/sources/ai-workshop-visual-aids.json.

How could the workflow you tested save time and help your team make money?

How often? How much time back?

Template slide 34 of 42. Ask participants to reflect on the workflow they just tested. Which manual step could it reduce, and how could their team use the time for customer follow-up, delivery, or other work that supports revenue? Invite one concrete example in the chat. Connect the opportunity to their actual task and checked result. Treat time savings and financial value as opportunities to measure rather than established results from the exercise. The next slide asks what worked, what needs improvement, and what to change next. Before discussing the price, ask how many times per week this work recurs and the manual time per run. Estimate potential time recovered from their own observations, including checking and correcting AI output. Ask what they would do with that capacity. Keep estimates distinct from verified savings or revenue; do not invent an ROI claim.

Insights and actions

What worked?
What needs improvement?

What will you change
before running your workflow again?

Turn your tested workflow into a repeatable AI agent

Template slide 35 of 42. Introduce turning the workflow participants just tested and improved into a repeatable AI agent. Keep the same recurring job, useful outcome, approved inputs, and quality standard. The following slides cover saving the formula and setup, testing fresh inputs, assigning an owner, and measuring results before choosing scheduled or on-demand use. An implemented agent needs the saved instructions, relevant tools and access, and clear review points so it can perform the task again. Use reviewed feedback to improve the next run. Today’s tested workflow is the starting point for this build; do not present it as an already installed agent. AI 30 offers hands-on help building from that workflow and repeating the process across more tasks. The next slide now creates the agent in a supported tool using the tested formula as instructions, approved sources, and quality checks. Saving instructions is part of that setup, not proof that an agent is running.

Do this first

Create your agent

Save your formula as instructions.

Add approved sources and quality checks.

Animated official Microsoft Agent Builder screenshots showing reusable instructions, the Create control, and the saved Contoso Report Agent. — Copilot · vendor example ↗

Template slide 36 of 42. In the chosen AI tool, create an agent where the tool supports agents. Save the tested How Might We AI Formula as reusable instructions, including the expected output and constraints. Add approved sources or connectors, the required tools and permissions, and the quality checklist. Keep the checked draft as a reference and make the review points explicit. Saving a prompt alone is not an installed agent. If the chosen tool lacks an agent builder or setup is incomplete, preserve this instruction package for the build rather than claiming an agent is running. Bring this same workflow into AI 30 for hands-on co-building if help is needed. The next slide tests the configured agent on fresh inputs. Visual aid: Official Microsoft Agent Builder screenshot sequence, source-checked October 9 and documentation rechecked October 13, 2026. Contoso is Microsoft’s example, not a participant’s account or a completed workshop build. Configure instructions, add approved knowledge and checks where the selected tool supports them, then create and test the agent. Availability depends on account and license. No agent was created by this deck edit. The GIF is an animated sequence of authentic screenshots, not a continuous recording. The participant’s selected task remains the main exercise. Pause holds the current screenshot; Play resumes the sequence. Reduced-motion viewing starts with the static poster. Open the image for the original full-size GIF. All source frames and crop bounds are recorded in docs/sources/ai-workshop-visual-aids.json.

Then

Test your agent
on fresh inputs

Review the result. Save feedback. Test again.

Public AI 30 workflow illustration showing draft review, a saved feedback rule for the next run, and editable instructions in the handoff. — Longhand workflow · illustrative ↗

Template slide 37 of 42. Test the configured agent on fresh representative inputs with the same outcome and quality standard. Compare the output with the approved sources, expected format, and constraints. Capture the actual correction, save the useful lesson in its instructions or checklist, and run another input to check whether the change helped. Keep reviewed feedback with the workflow so each test can inform the next run. This is the practical learning loop used by the live AI 30 offer verified October 13, 2026: reviewed corrections become reusable instructions and checks. The participant controls what is saved; do not imply automatic model learning or unreviewed changes. If the agent is not configured yet, test the instruction package and name the remaining setup. One successful sample does not establish reliability. Visual aid: Actual public AI 30 page screenshots captured October 13, 2026. The source labels this meeting workflow illustrative and describes a chosen tool or Longhand workflow. Frames show the reviewable brief, a correction saved into instructions and checked on the next run, and the editable handoff. This is an illustration of the Longhand feedback process, not an authenticated Longhand product screen, an executed agent, or measured proof of improvement. Participant-controlled review and saved changes remain necessary. The screenshot sequence does not execute a workflow. Explain the practical loop: review, save the correction, check the next run. The visual has a reduced-motion still, Pause/Play, and a full-size GIF link; source crops and dates are recorded in docs/sources/ai-workshop-visual-aids.json.

And then

Assign an owner.
Measure the value.

Time saved · Quality · Token cost

Track one business outcome.

Template slide 38 of 42. Name who runs, reviews, and improves the agent and when the next input arrives. Compare time with comparable manual work, including preparation, review, corrections, and maintenance. Track output quality against the checklist and token cost where the tool reports it. Choose one business outcome appropriate to the task, such as customer response time or completed follow-ups. Distinguish potential capacity value from measured savings or actual revenue. Use repeated tests and this evidence to decide whether the configured agent is ready for routine use, what still needs improvement, and whether to expand to another task. The owner maintains the saved instructions and feedback loop. Record minutes per comparable run and runs per week before and after using the agent. Review quality and maintenance effort alongside time. Let the participant choose a useful business measure. Included program API usage does not eliminate the need to manage tokens or account for other subscriptions and API usage after day 30.

What could your team do
with the time back?

When your agent is ready

Schedule your agent
or use it on demand

Continuous improvement

Review results. Save feedback. Improve the next run.

Animated official ChatGPT Tasks screenshots showing scheduled work and the recurrence settings. — ChatGPT Tasks · vendor example ↗

Choose scheduled or on-demand use once the agent has been configured and tested. Match the usage mode to the recurring task, available inputs, the tool’s capabilities, and the review required. Set a named owner and the appropriate review points. For continuous improvement, review the actual result, capture corrections, save useful lessons in the instructions or checklist, and check the next run. Participants control what is kept; self-improving does not mean unreviewed automatic changes. Transition into AI 30 as hands-on co-building around today’s workflow, with help repeating the process across more recurring tasks. This slide describes implementation and does not configure a schedule or run an agent. Visual aid: Official OpenAI Tasks screenshots, source-checked October 9 and documentation rechecked October 13, 2026. The second frame focuses on recurrence controls; the original vendor sample remains unchanged in its source asset. These controls illustrate cadence, not a claim that any saved workflow can run in Tasks. Current OpenAI guidance says tasks created in a project cannot access its uploaded files. Choose a tool that supports the workflow’s approved sources and review points. No task was created or scheduled. The GIF is an animated sequence of authentic screenshots, not a continuous recording. The participant’s selected task remains the main exercise. Pause holds the current screenshot; Play resumes the sequence. Reduced-motion viewing starts with the static poster. Open the image for the original full-size GIF. All source frames and crop bounds are recorded in docs/sources/ai-workshop-visual-aids.json.

From our community

“…significantly reduce
the time I spend
on repetitive tasks.”

Marco Monteiro

Tech Leaders member feedback

Read member feedback

A nine-word exact excerpt from Marco Monteiro’s public testimonial, verified October 13, 2026 in the Tech Leaders participant testimonials embedded on the live AI 30 page. The source is https://embed-v2.testimonial.to/carousel/all/tech-leaders-mastermind-program?id=948cd5e8-1b47-43b4-8da1-a15b4bbdfb8e. His review discusses focusing on better solutions and reducing repetitive work through the program and AI tools. This is member feedback from the broader Tech Leaders community, not a measured AI 30 result, a 30-day guarantee, or a claim that every participant will obtain the same result. Keep the earlier published healthcare example identified as team training. Ask participants which part of their own recurring work they would like to get back before introducing hands-on AI 30 co-building.

Workshop Q&A

Questions about
your workflow?

Ask in the chat.

AI 30

Build unlimited,
self-improving AI agents
in 30 days.

Your first useful agent.
Built with you and for you.

Bring today’s workflow. No program cap.
Builds depend on complexity and participation.

The live AI 30 offer at https://www.technical-leaders.com/ai-30 was verified October 13, 2026 and now emphasizes unlimited, self-improving AI agents. AI 30 is a 30-day program for individuals with hands-on co-building around their actual work. Bring today’s formula, approved sources, checked draft, and quality checklist. We build the first useful agent with them and for them, including a practical learning loop: review results, capture corrections, save useful lessons in instructions, and check the next run. Then help them repeat the process across more tasks. Unlimited means no program cap on agents participants can create with the process. Co-building scope is agreed together; completed builds depend on complexity and participation. OpenAI, Anthropic, and Gemini API keys and unlimited usage are included during the 30-day program. Other tool subscriptions and API usage after the program are separate. Participants control saved changes and review whether they helped. Do not promise a completed agent count or financial return. Bridge explicitly from the participant’s tested workflow to co-building alongside us. The workshop supplies a checked starting point; AI 30 supplies help configuring, testing, improving, and expanding useful agents over the 30-day program.

Build with us.
Keep improving.

Hands-on co-building
Unlimited model API usage · 30 days
Curriculum + recording library
1 year of weekly AI Mastermind
24/7 AI Support Line

The live AI 30 offer at https://www.technical-leaders.com/ai-30 was verified October 13, 2026. Included support is hands-on AI agent co-building; the full agent curriculum with the masterclass, walkthroughs, examples, and templates; the full recording library; one year of continuation support in the weekly AI Mastermind with coaching, member hot seats, and peer feedback; and the 24/7 AI Support Line. Participants can begin building while learning and use reviewed feedback to improve their agents. One year refers to AI Mastermind continuation support after the 30-day program, not every service or unlimited one-to-one delivery. The support line is available for questions; no instant response-time guarantee is stated. Longhand is the platform for building and running workflows and carrying feedback into reusable instructions and quality checks. The live page now includes API keys and unlimited OpenAI, Anthropic, and Gemini usage throughout the 30-day program. This helps participants compare models on their own work without additional program API charges. Other tool subscriptions and API usage after the program are separate. It does not include every consumer app subscription or extend unlimited API usage through the full year of Mastermind support.

AI 30

Your 30-day build sprint

Start with today’s workflow.
Build in the feedback loop.
Expand to more work.
Measure. Improve. Keep building.

Template slide 41 of 42. The live AI 30 offer at https://www.technical-leaders.com/ai-30 was verified October 13, 2026. Explain the four-part progression using today’s workflow as the starting point. Week 1: map recurring tasks, choose a useful first build, begin co-building, and agree what good looks like. Week 2: build in the feedback loop by running real examples, capturing corrections, saving useful lessons in instructions, and checking the next run. Week 3: expand toward the next automation priorities and connect steps where useful with approved permissions and human review. Week 4: compare results with the starting point, check which improvements helped, and decide what to automate next. Adapt the pace to complexity and participation. Participants keep the scoped working agents, editable instructions, saved feedback and quality checks, an automation roadmap, a 30-day results summary, and an AI Bill of Materials recording tools, instructions, data, permissions, and review steps. Continue improving through the included year of AI Mastermind support.

Is this for you?

Ready to take recurring work
off your plate?

For individuals: owners, leaders, and operators.

No engineering background needed.
Your work, your team’s work, or both.

The live AI 30 offer at https://www.technical-leaders.com/ai-30 was verified October 13, 2026. AI 30 is for individuals, including owners, leaders, and operators, who have recurring work to automate, want hands-on help, and want a building process they can keep using. Bring today’s workflow, approved context, and feedback. Build alongside us, review results, and save useful corrections to improve future runs. No engineering background is needed. A company may fund an individual’s participation; this is distinct from a whole-team training package. Identify company-approval or access questions in Q&A and direct joining to the AI 30 page. Keep a person responsible for outputs and saved changes. Return to the opening A/B/C question. AI 30 is for an individual who wants help building around their own work or work they own for their team. Company-funded participation is still individual participation. For team-wide AI literacy and skill development, discuss AI 90 at https://www.technical-leaders.com/ai-for-leaders in Q&A while keeping AI 30 the workshop’s main joining invitation.

The investment

$2,500

One payment · 30-day build sprint

30 days of unlimited model API usage included.

Other subscriptions + later API usage cost extra.

Not the right fit? Tell us within your first 7 days for a refund.

technical-leaders.com/ai-30

The live AI 30 offer at https://www.technical-leaders.com/ai-30 was verified October 13, 2026. One payment of $2,500 covers the 30-day program with hands-on co-building and one year of AI Mastermind continuation support. OpenAI, Anthropic, and Gemini API keys and unlimited usage are included during the 30-day program. Other tool subscriptions and API usage after the program are separate. There is no program cap on agents participants can create using the process; co-building scope is agreed together, and completed builds depend on complexity and participation. Self-improvement uses reviewed feedback saved in instructions and quality checks, with the participant controlling what is kept. If the program is not the right fit within the first 7 days, tell us for a refund of the program payment. Use the live page as the authority for terms. Do not add a discount, urgency, or guaranteed savings.

Questions about
AI 30?

Bring today’s workflow. Join AI 30.

technical-leaders.com/ai-30

Outline adapted from The+Million+Dollar+Workshop.pdf, physical page 22: The Client Machine Workshop slide overview. The live AI 30 offer at https://www.technical-leaders.com/ai-30 was verified October 13, 2026. Invite questions about their workflow, hands-on co-building, the feedback loop, unlimited-agent definition and agreed scope, included program model API usage, other subscriptions and API costs after day 30, the 30-day sprint, one year of AI Mastermind support, and the 7-day refund condition. Explain that self-improving means reviewed corrections saved as reusable instructions and checks, then tested on the next run. Answer from the current page. Keep private company context for follow-up. Return to one invitation: bring today’s workflow and join AI 30 to build and improve more agents. Keep the program link visible throughout Q&A. Return to the participant’s chosen workflow and their A/B/C answer. Joining provides access to curriculum, recordings, and the support line; the first step with us is mapping their work and choosing the first build. For a broader team development need, discuss AI 90 separately; do not replace the primary AI 30 invitation with a call.

Your next step

Bring today’s workflow.
Join AI 30.

Build and improve more agents.
Automate more work in 30 days.

Join → Get access → Choose your first build

technical-leaders.com/ai-30

Outline adapted from The+Million+Dollar+Workshop.pdf, physical page 22: The Client Machine Workshop slide overview. The live AI 30 offer at https://www.technical-leaders.com/ai-30 was verified October 13, 2026. Invite participants to bring today’s formula, approved sources, tool setup, checked draft, and checklist into AI 30. We build alongside them around their recurring work, starting with a useful agent and a practical feedback loop, then helping them repeat the process across more tasks. Review results, capture corrections, save useful lessons in instructions, and check the next run. The 30-day program has no cap on agents created with the process; co-building scope is agreed together and completed builds depend on complexity and participation. OpenAI, Anthropic, and Gemini API keys and unlimited usage are included during the 30-day program. Other tool subscriptions and API usage after the program are separate. One year of AI Mastermind continuation support is included. The attributed public AI 30 link lets them review the offer and join. Leave the invitation visible; the deck performs no enrollment or payment. After joining, access to curriculum, recordings, and support begins. Start by mapping the work and choosing the first build with us. Keep the explanation short and practical; do not invent a fixed start date or response-time guarantee.

Thank you

Bring today’s workflow.
Join AI 30.

technical-leaders.com/ai-30

Template slide 42 of 42. Keep the thank-you, invitation, and public AI 30 link visible as participants leave. Bring today’s workflow into the 30-day hands-on co-building program, learn a repeatable process for building more agents, and use reviewed feedback saved in instructions and checks to improve future runs. Continue with one year of AI Mastermind support. Unlimited means no program cap on agents created using the process; co-building scope is agreed together, completed builds depend on complexity and participation. OpenAI, Anthropic, and Gemini API keys and unlimited usage are included during the 30-day program. Other tool subscriptions and API usage after the program are separate. Repeat the invitation without guaranteed financial returns or urgency. The attributed link opens https://www.technical-leaders.com/ai-30; the deck submits no information, enrollment, or payment.

Move Beyond Prompts With AI Agent Workflows

Your role.Your task.Your goal.

What AI tools areyou using today?

Who are you lookingfor AI help for?

We help leaders and teamsturn one-off AI experimentsinto repeatable workflows.

One task.One repeatable workflow.

Bring the work.Keep your judgment.

AI can now automatemulti-step knowledge work.

Everyone wants AI.Most are stuck experimenting.

Tool hype

News fear

“I’m not surewhat to say to AI.”

“I’m not surewhich AI tool to use.”

“Why can’t I getconsistentresults?”

“Why bother?But what ifwe fall behind?”

Automate work with AI agents.Save time. Make money.

Know whatto say to AI

Know the right AI toolfor the job

Get consistent resultswe can trust

AI agents dothe recurring work

AI forrecurring work

The foundation:repeatable workflows

What recurring work couldAI agents do for your team?

Three foundationsfor AI agents

The 4 Gears of AI Usage

Five steps to repeatableAI agent workflows

InstallCapabilities

Setup

Assign

Equip

How Might WeAI Formula

Which step do you needto complete next?

Let’s build one

Pick a workflowto agentify

Write yourAI Formula

Test your workflow

How could the workflow you tested save time and help your team make money?

What worked?What needs improvement?

Turn your tested workflow into a repeatable AI agent

Create your agent

Test your agenton fresh inputs

Assign an owner.Measure the value.

What could your team dowith the time back?

Schedule your agentor use it on demand

“…significantly reducethe time I spendon repetitive tasks.”

Questions aboutyour workflow?

Build unlimited,self-improving AI agentsin 30 days.

Build with us.Keep improving.

Your 30-day build sprint

Ready to take recurring workoff your plate?

$2,500

Questions aboutAI 30?

Bring today’s workflow.Join AI 30.

Thank you

Your role.
Your task.
Your goal.

What AI tools are
you using today?

Who are you looking
for AI help for?

We help leaders and teams
turn one-off AI experiments
into repeatable workflows.

One task.
One repeatable workflow.

Bring the work.
Keep your judgment.

AI can now automate
multi-step knowledge work.

Everyone wants AI.
Most are stuck experimenting.

“I’m not sure
what to say to AI.”

“I’m not sure
which AI tool to use.”

“Why can’t I get
consistent
results?”

“Why bother?
But what if
we fall behind?”

Automate work with AI agents.
Save time. Make money.

Know what
to say to AI

Know the right AI tool
for the job

Get consistent results
we can trust

AI agents do
the recurring work

AI for
recurring work

The foundation:
repeatable workflows

What recurring work could
AI agents do for your team?

Three foundations
for AI agents

Five steps to repeatable
AI agent workflows

Install
Capabilities

How Might We
AI Formula

Which step do you need
to complete next?

Pick a workflow
to agentify

Write your
AI Formula

What worked?
What needs improvement?

Test your agent
on fresh inputs

Assign an owner.
Measure the value.

What could your team do
with the time back?

Schedule your agent
or use it on demand

“…significantly reduce
the time I spend
on repetitive tasks.”

Questions about
your workflow?

Build unlimited,
self-improving AI agents
in 30 days.

Build with us.
Keep improving.

Ready to take recurring work
off your plate?

Questions about
AI 30?

Bring today’s workflow.
Join AI 30.