Skip to content
AI SchoolAI. Film. Agents and prompts.
Choose this schoolProject & Product courses

Long courses · Live online, or onsite in the GTA

Program In Generative AI and Agentic AI

A chatbot answers. You build what sits behind it: a model you can call, documents it is allowed to read, an agent that takes the next step, and a release you can stand behind. You leave with that system running.

150 hoursAdvancedDate on this page

Schedules for Program In Generative AI and Agentic AI

Each date shows the trainer, the session times, and the enrol or waitlist control. Pick London to see GMT or BST on the clock. Refund and transfer terms.

NGS-1032-2026-11

Program In Generative AI and Agentic AI

Enrolment closed
SG

Shailesh Gabhane

Founder · Lead Trainer

English

Trainer profile

Where and when

  • Sat, Nov 7, 2026 — Sun, Apr 25, 2027
  • Sat–Sun · 9:00 AM to 12:00 PM
  • Virtual (Eastern Time)
  • Live online, or onsite in the GTA · source zone America/Toronto
More details

$2,500

This date is closed. Bookings end two days before the class starts. Join the waitlist.

Price

What this seat costs

Program In Generative AI and Agentic AI. For Software engineers. 150 hours. No date is open to buy. Class $2,500. Practice $3,750. Mentor $5,000.

Prices and the next date were checked on 19 November 2026. A date that has started is not bookable.

Who it is for
Software engineers
Hours
150 hours
Where
Canada
Next date
None open

Choose your package

Pick a package, then add this course. The price is the package you select. Dates are listed below.

Class · CAD 2,500

Add to cartEnrol now

Course overview

You are tired of being the person who pastes into a chat box. Over six months of live weekends you build the system that has to be right when the chat is wrong.

Saturday and Sunday, three hours each, for twenty-five weeks. You write the Python. You call a model and you limit what it can see. You make it answer from documents you chose. You let an agent act, then a small team. You put the result where someone else can start it.

You leave with one system you scoped, built, and deployed, and the smaller pieces you made on the way. The certificate is for finishing this program. It is not a university degree.

Course journey

Twenty-five weeks

The live path is eight blocks. A primer and three self-paced paths sit beside it. They are not extra weekends.

  1. Start

    Python that can carry a service

    Weeks 1–2. You write code that finishes, moves a table of data, and opens a small door other software can call. The editor may suggest the next line. You still decide.
    01
  2. Then

    Say what you mean to the model

    Weeks 3–6. You learn the call well enough to set it. You stop handing the model a context that has gone stale.
    02
  3. Then

    Answers from documents you chose

    Weeks 7–11. The model stops inventing the source. It reads what you indexed, and you can tell when the hit was the wrong page.
    03
  4. Then

    An agent that is allowed to act

    Weeks 12–16. A reply is not the job. The job is a next step, a tool, and a person on the step that can hurt.
    04
  5. Then

    A contract with the rest of your software

    Weeks 17–20. The agent reaches a file, a repository, or another agent through a door you can log. Not through a script you hope still works.
    05
  6. Then

    The step that should not be code

    Weeks 21–22. Some jobs belong on a canvas an operator can see. The model still reasons. A person can still find the step.
    06
  7. Then

    After the demo

    Weeks 23–24. Your laptop is not where this lives. You box it, you watch the cost and the drift, and you write the rule.
    07
  8. Finish

    The system you will defend

    Week 25. You scope one system, build it, and deploy it. You can explain every part you left in.
    08

Quick overview

What this fee covers

  • Who it is for

    Engineers, analysts, technical leads, and product managers who will build, not only prompt.

  • When it meets

    Live, Saturday and Sunday, three hours a day, for 25 weeks.

  • Hours

    150 hours in the live room. The primer and the three extra paths are self-paced and sit outside that count.

  • What you need

    A degree. Some code already, or the primer before week one.

  • In the fee

    The live weekends, the work between them, and a certificate of completion for this program.

  • Not in this fee

    A university degree. Model, cloud, and vendor accounts.

Why this work, now

The work has already moved

A chat window can draft a paragraph. The job in front of most teams is a system that takes the next step, reads a private file, and stops when it should. People who can build that are being asked for faster than teams can grow them. The figures below are published estimates. They are not a job, a salary, or a promise from us.

$1.3T

A published projection for the generative AI market across the coming decade.

Bloomberg

97 million

Roles one published outlook ties to AI and machine learning, with technology leading the count.

World Economic Forum

$193,000

A published average yearly pay for a generative AI engineer. An offer you receive will not be this number.

Glassdoor

Where this shows up at work

  • A multi-step job handed to an agent that can use a tool.
  • An answer drawn from the company's own files, not from the public internet.
  • A whole function moving, not one person's inbox.
  • A prototype in front of a customer while the idea is still warm.
  • A shared picture a team can look at before it decides.
  • A new way of connecting agents, adopted without waiting a year.

The path

How the six months move

  1. 01

    Write Python a service can stand on

    A file that runs, a table you can defend, and a small door other software can call.

  2. 02

    Give a model an instruction it can follow

    The request, the limit, and an output a program can read.

  3. 03

    Build an application that answers from your files

    Search first. Then the model. Then a check that the sentence came from the page.

  4. 04

    Let an agent take the next step

    A tool it may call, a tool it must not, and a person on the sharp step.

  5. 05

    Design a system of agents

    More than one worker, a shared state, and a rule for who may overrule whom.

  6. 06

    Put it where other people can run it

    The same box twice, a watch on the cost, and a morning-after report.

  7. 07

    Ship the system you will defend

    One path, from the document to the action, that someone else can start.

After the program

The work this prepares you to do

Kinds of work, not a placement. Finishing the program does not confer a title, a salary, or a seat on a team.

Starting out

  • Build the first AI feature on a product
  • Write the instruction and the check beside it
  • Keep a live assistant from falling over
  • Call a model from code, with someone senior nearby

In the middle of the work

  • Own a generative feature, not a demo
  • Build agents that are allowed to act
  • Test whether the model is telling the truth
  • Specify an AI product so other people can build it

Later

  • Architect the system other people run
  • Lead the AI work for a team
  • Be the principal engineer on this kind of system
  • Own the decision of what the company will not automate

The portfolio

What you will have built

A portfolio of working pieces, not a folder of notes.

The last week

One system, from a blank page to a start button

You scope it, build it, and deploy it. It can read, use a tool, hand work to more than one agent, and stop. You can explain every part you left in.

  • A reader that chooses what to fetch

    It decides which document to open, and it corrects itself when the first page was the wrong one.

  • A small research team

    One looks things up, one drafts, one reviews. You say who may overrule whom.

  • A support agent that keeps a secret

    It answers from the approved file, and it will not repeat a customer's private details.

  • A reader for the paperwork

    Contracts, invoices, and reports, searchable, with the page named in the answer.

  • A door you built for one model

    Your tools and your data sit behind a server you run. The model gets a contract, not the database.

  • A workflow that starts itself

    A trigger, a decision, and an action in a system you already use. A person can see the step that stopped.

How a class runs

Five moves. Then you have the piece.

A live class. You see the move, you try it on your own work, and you hear what is wrong before you leave.

  1. See the move

    The instructor shows the idea on one picture before anyone opens a tool.

  2. Bring the real task

    A roster, a story, a shot, a backlog. Not a case written for a slide.

  3. Do it in the hour

    You try the step while the instructor is still in the room.

  4. Hear what is wrong

    You get a correction on your work, in the class. Not a score from a quiz.

  5. Take the piece home

    What you made is what you use on Monday. The recording is there if you need the hour again.

Why this school

Why people choose Next Gen Skills Builders.

A live class, practice before it counts, and a recording for 90 days. The last two are the reason most people finally book.

  • A live instructor

    The hours on this page are spent in class. This is not a video you are left to finish alone.

  • Practice before it counts

    You do the work in the room. Where the course leads to an exam, you also sit a timed practice before the real clock.

  • Hear the class again

    The live hours are recorded for 90 days. That replay is included. A second live seat is a new booking.

  • The lessons stay here

    After you enrol, the chapters for this course stay on this site. There is not a second system to learn.

  • Three ways through the same course

    Class, Practice, and Mentor. Practice adds a set you can redo. Mentor adds one private session and a written plan.

  • Built to the work it names

    Exam courses follow that exam's outline, including the hours you can show on an application. Every other course follows the job in its title.

  • A result you can point to

    A film someone can watch. An agent that does one job you already do. A plan you can take into Monday's meeting. Or a practice score that tells you whether to pay for the exam. You leave with a result, not a certificate.

  • The date is the decision

    You have already thought about this long enough. The date on this page is the moment you stop postponing it.

  • You are not in the room alone

    Other people are doing the same hard thing, on the same day. You can ask the question you would never type into a video.

On Google

What a student wrote

These people wrote about the AI classes. Their words, from Google. Not a promise about your result.

  • “I learned business analysis, project management, change management, and some uses of AI tools. I joined while I was still an undergraduate.”

    Trushil Patel

    Google review

  • “I do not come from a technical background. The AI for business analysts course and AI for Delivery Professionals helped me learn the subject and move toward AI leadership.”

    Mohit

    Google review

  • “The AI Transformation for Business Leaders course exceeded what I expected. I already knew generative AI and prompt engineering. I wanted the rollout: stakeholder dynamics, ethical frameworks, and how an AI strategy sits inside platforms such as ERP and CRM.”

    Mayur Parekh

    Google review

  • “I could finally tell the roles apart: business analyst, change manager, project manager, product manager, and release train engineer, including how that work meets current AI. I also got one-to-one help with job hunting and interview preparation.”

    Krishna Singh

    Google review

  • “I completed AI for Delivery Professionals. I was already a Scrum Master. The course covered generative AI, agentic AI, agents, RAG pipelines, and workflow automation. After it, I built agents that automated backlog creation.”

    Preena

    Google review

Benefits

What is included

What is in the seat. The skills you leave able to use are under Outcomes.

  • Twenty-five weekends of live class. Six hours each weekend.

  • A system you can start: the documents, the tools, and the rule it will not cross.

  • Smaller builds along the way: a reader, a support agent, a tool server, a workflow.

  • A primer, and three self-paced paths, if you want the extra depth.

  • A certificate when you finish this program.

Outcomes

What you will be able to do when the course ends. This is not the list of things you take home. That list is Benefits.

  • You can write the Python a service needs, and put a small API and a screen in front of it.

  • You can call a model, limit the context, and demand an output a program can use.

  • You can make an answer come from documents you chose, and notice when it did not.

  • You can let one agent act, then a small team, and keep a person on the step that can hurt.

  • You can box the system, watch it, and say what it is not allowed to do.

Curriculum

12 parts, in the order you will use them

  1. 01

    Python that can carry a service

    You write code that finishes, moves a table of data, and opens a small door other software can call. The editor may suggest the next line. You still decide.

    • A file you can run again
    • Work that does not stand in a queue
    • A table you can defend
    • A door and a screen
  2. 02

    Say what you mean to the model

    You learn the call well enough to set it. You stop handing the model a context that has gone stale.

    • Two ways to call
    • A request, not a wish
    • What stays in the window
    • Did that change help
  3. 03

    Answers from documents you chose

    The model stops inventing the source. It reads what you indexed, and you can tell when the hit was the wrong page.

    • A paragraph you can search
    • Where those searches live
    • An answer that names the page
    • The second look
    • A page that is not only words
    • A small adjustment, not a new model
  4. 04

    An agent that is allowed to act

    A reply is not the job. The job is a next step, a tool, and a person on the step that can hurt.

    • Look, decide, act, look again
    • The tools it may touch
    • A state you can read tomorrow
    • A second worker
    • Memory that is allowed to last
  5. 05

    A contract with the rest of your software

    The agent reaches a file, a repository, or another agent through a door you can log. Not through a script you hope still works.

    • A door you wrote
    • The same door for a second tool
    • When two agents write to each other
    • A trace and a rail
  6. 06

    The step that should not be code

    Some jobs belong on a canvas an operator can see. The model still reasons. A person can still find the step.

    • A workflow that starts on its own
    • The same pattern on a second canvas
    • A canvas that can stop
  7. 07

    After the demo

    Your laptop is not where this lives. You box it, you watch the cost and the drift, and you write the rule.

    • The same box, twice
    • What you will not scale
    • The morning after
  8. 08

    The system you will defend

    You scope one system, build it, and deploy it. You can explain every part you left in.

    • The job, and the thing it must not do
    • From the document to the action
    • Ship it once, on purpose
  9. 09

    If the maths is thin

    Self-paced, and not part of the live weekends. The numbers, a classical model, and a small network, so the later weeks are not magic.

    • The numbers you will actually meet
    • A classical model
    • A small network
  10. 10

    The editor as a pair

    Self-paced. The editor drafts a line. You reject it or you keep it. A second tool does not change that rule.

    • Draft, then reject
    • A second editor, the same rule
    • The assistant in the cloud console
  11. 11

    Text, before the large model

    Self-paced. How words are cleaned, counted, and attended to, so a transformer is a step and not a brand.

    • Clean the words
    • Represent them without a chatbot
    • Attention, in one example
  12. 12

    One vendor, end to end

    Self-paced. You take a single model family from the instruction, through the API and a tool call, to a door that vendor can see.

    • The model and the instruction
    • The API and a tool call
    • The editor that stays with that vendor
    • A door that vendor can see

Who can attend

Built for people who will ship the system

You will get the most from this ifSoftware engineers, data people, analysts, technical leads, and product managers who will build the system. Also someone early in the work who can already write a little code.

What to bringA degree. You will do better if you already write some code. If the maths is thin, do the primer before the first weekend. Bring a laptop. The model and cloud accounts are yours to open.

Instructor

Faculty for the next cohort is confirmed closer to the start date.

With this course

What you can add

These are the extras mapped to this course. An admin chooses them. Nothing here is tied to the course by name.

  • This courseIncluded

    Live Program In Generative AI and Agentic AI

    150 live hours, the recordings, and the work from class. The seat price is above, not on this card.

  • IncludedIncluded

    Course certificate

    A certificate of completion for these hours. It is not the exam credential.

Twenty-five weekends. You leave with a system you can start, and a line it is not allowed to cross.

FAQ

What is the difference between Class, Practice, Mentor?
Class is $2,500. It includes Live class on Saturday and Sunday for twenty-five weeks.; The work between those weekends.. Practice is $3,750. It includes A practice set you can redo after the class.; The recording stays available for 90 days.. Mentor is $5,000. It includes One private coaching session.; A written plan for the work you will do next.. A higher package is the one below it, plus what its card lists.
Do you guarantee a job after a cohort?
No. We publish dates, prices, refund terms and consented stories. Placement and income claims are not part of this academy.
What time zone are live sessions in?
Source zone is America/Toronto unless a cohort says otherwise. Session times always name that zone. Convert in your own calendar from the ICS on the cohort card.
Can I join a waitlist if a cohort is full?
Yes. Waitlist captures name, email and optional phone. It does not take payment. Seats that free up are offered in join order.
How does early-bird pricing work?
If a cohort has an early-bird price and the cut-off has not passed (server date at payment), checkout charges the early-bird amount. The cart may still show list price until quote.
What is the refund window?
Full refund before the published cut-off relative to the cohort start; after that the published bands on the Refund page apply. Checkout requires you to accept that policy.

Ask

About this instructor and this programme. A person replies.