1. Name and current roles

Raghav Mathur, professionally known as Rakz

Chief Operating Officer, Pounce Agency

Technology advocate; AI/ML/software and human interpreter; curious creative

2. Professional profile

Rakz works where business, technology and creative practice meet. As Chief Operating Officer of Pounce Agency, he helps connect the problems a business needs to solve with the people, processes and technology that can solve them.

Pounce is a creative transformation agency operating across Australia and Malaysia. It combines management-consultancy thinking with the people who execute the work, taking a human-led, AI-enabled approach to improving time, cost and quality.

Rakz brings a practitioner's perspective to AI. He and the Pounce team built Janice, Pounce's Agentic Chief of Staff, also known as Pounce OS. It is used as an operating layer across business context, work routing, research, delivery, growth, approvals and improvement. His teaching draws on that experience: define the business problem, build something useful, test it and keep people accountable for the decisions.

3. Education background

• Master's in Arts & Entertainment Management, Australian Institute of Music.
• Bachelor of Computing in System Development, Western Sydney University.

This combination connects systems thinking with an understanding of creative work and its management. It reflects the territory Rakz works in today: translating between technical possibilities, business needs and the people expected to put a new way of working into practice.

4. Professional experience and distinctive practice

Rakz's work at Pounce connects agency operations, creative transformation and the practical use of technology. His focus is not simply what a tool can generate. It is how that capability fits into the work, what information it needs and who remains responsible for the result.

Building Janice/Pounce OS with the team has made those questions concrete. An operating layer needs more than a persuasive response to a prompt. It needs relevant context, defined responsibilities, ways to route work and clear points where a person reviews or approves what happens next.

His approach starts before automation. If a task is repetitive, he asks why it exists, what decision it supports and whether the process itself needs to change. Making a poorly defined process faster is not the same as improving it.

From there, the method is practical: define a manageable task, build, test, verify and improve. Humans retain responsibility for planning, briefing, prompting and review. AI expands what the team can explore and produce; it does not remove the need for judgement.

This is the distinctive perspective he brings to a classroom. Participants learn from someone applying AI inside an agency, while starting with a task they can understand and assess themselves. They are not asked to reproduce Pounce OS or trust a system they cannot explain.

5. Publications and industry contribution

Rakz is co-author, with Nick Beaugeard, of Make AI Happen: A Guide for Business Executives. The book is part of his contribution to helping business audiences engage with AI beyond the technical conversation.

He also writes for Pounce Insights, including on generative AI and why marketers need to think like technologists. His contribution connects creative and commercial questions with an understanding of how technology works: what it makes possible, what it changes about the work and what people need to understand before adopting it.

6. Speaking and media appearances

Rakz's speaking and moderation experience includes:

• BBX Connect Day 2.0, March 2026: panellist for “From Hype to Know-How: Understanding and Implementing AI for SMEs”.
• Intrigue MAdVerse Kuala Lumpur 2026: keynote speaker.
• AWS Unicorn Day Malaysia, 2025: panel moderator.

His media appearances and coverage include a Solve 2 Evolve podcast episode featuring him, Marketing-Interactive coverage of Pounce's Malaysia expansion and AdNews coverage naming him as Pounce's Chief Operating Officer.

These contributions place his work in conversations about AI adoption, marketing, business leadership and the practical challenges of change.

7. Recognition

Rakz is part of the Blutui Founder's Circle, Founding 5. Blutui recognises him as an early adopter whose feedback helped develop AI Components. This is industry recognition of his contribution to product development and practical adoption, rather than a competitive award.

8. Personality

I'm the person who asks why we do it that way, then gets curious enough to build an alternative and see whether it actually works.

9. Values

Curiosity with a purpose

I want to understand the problem before I decide what to build. Curiosity means questioning the request, exploring alternatives and being willing to change my mind when the evidence changes.

Human responsibility

People provide context, make decisions and take responsibility for the consequences. I want AI to make room for better thinking, creativity and relationships, without disguising who is accountable.

Useful over impressive

A demonstration can look brilliant and still be wrong for the business. I value work that helps someone make a better decision or handle a real task, with results they can check.

Build, test, learn

I believe in making an idea tangible early enough to learn from it. Testing is part of creating the work, not something added after we have decided it is good.

Honest communication

I want people to understand what a system does, what it needs and where its limits are. Plain language makes it easier to question an output, spot a problem and decide what to do next.

10. Goals

I want to help business leaders move from experimenting with AI to making informed choices about where it belongs in their work. That starts with being able to describe a problem clearly, recognise a useful result and challenge an answer that only sounds convincing.

Through this course, I want participants to experience the full process on a small scale. They should leave having built and tested something, with enough understanding to explain how it works and enough judgement to know what still needs attention.

My broader goal is to keep developing human-led, AI-enabled ways of working at Pounce and share the practical lessons. The ambition is to improve time, cost and quality while protecting the human work that gives a business its character: relationships, ideas, judgement and care.

11. Module name and subtitle

AI in Action: Build Your Agentic Chief of Staff

Ten weeks to build and test a starter AI assistant, then begin adapting it to your own work.

Duration: Ten weekly sessions across six modules.

Course fee: $1,500 per participant.

Turn meeting notes into a useful leadership brief. Build a shared starter first, then adapt it to your own work.

Here, “agentic” means working through a defined task towards an outcome, not permission to act on someone's behalf. The assistant organises supplied information and prepares work for human review. It does not run the business.

12. Mechanism / learning approach

Weeks 1–8 guide everyone through the same starter build: turning sample meeting notes and project updates into a weekly leadership brief. Each week adds one manageable capability or improvement, from identifying decisions to handling missing information. A common task gives participants a shared basis for comparing results and learning from mistakes.

Early exercises establish the instructions and context. Later sessions assemble and test the prototype on the selected course platform. Demonstrations lead into guided practice and revision, with time between sessions to revisit each step.

Week 9 combines structured refinement of the starter with choosing, scoping and mapping a personal business use case. Week 10 begins that adapted build, tests its first result and sets out the next 30 days. Participants leave with one tested starter prototype and the beginning of their own application, not two finished systems.

All exercises use sample or anonymised information in a practice environment, including the personal-use-case work. Participants start each run themselves and check the output. No live production integrations, autonomous external actions, multi-agent operations, scheduled production workflows or recovery engineering are included. A “weekly” brief describes the output, not an automatic schedule.

The course draws on the foundations behind Janice/Pounce OS, not its full scope. Extra weeks allow practice and improvement, not production complexity.

13. What participants leave with

• One tested starter Chief of Staff prototype that prepares a leadership brief from supplied sample information.
• A task map showing inputs, steps, the expected output and human review.
• A role and boundary brief, a small sample context pack and a reusable output format.
• A test checklist documenting results, improvements and remaining limitations.
• One scoped personal business use case, with its own task map and success criteria.
• The beginning of an adapted build and a reviewed first result using sample or anonymised material.
• A practical 30-day continuation plan covering further tests, a reviewer and limits on use.

Participants can demonstrate the starter and explain its limitations. “Tested” means checked against course examples and recorded criteria, not guaranteed accuracy or readiness for unrestricted use.

14. Topics covered

• What an Agentic Chief of Staff can support, and where human judgement belongs.
• Moving from a one-off response to a defined, repeatable task.
• Mapping inputs, steps, outputs and review responsibilities.
• Supplying relevant context and separating facts from suggestions.
• Writing instructions, defining boundaries and flagging missing information.
• Building a prototype and checking it against straightforward, incomplete and out-of-scope examples.
• Choosing a personal use case and adapting the starter without expanding its permissions.
• Planning further testing before considering real business use.

15. Six-module outline across ten weeks

Module 1: Understand the Possibilities

Week 1

Focus: Understand the starter assistant's job and what makes its output useful.

Activity: Compare a one-off AI answer with a leadership brief prepared from a defined task and supplied notes. Follow a guided exercise to identify decisions in the sample material. Discuss what the assistant can prepare and what a manager must still decide.

Participant output: A clear description of the shared starter task, an initial decision list checked against the notes and a shortlist of possible applications to revisit in Week 9.

Module 2: Map the Work

Weeks 2–3

Week 2: Identify the information the brief needs

Focus: Connect the desired result to the information available.

Activity: Sort sample meeting notes and project updates into useful inputs, irrelevant detail and gaps. Map the path from receiving the material to preparing a brief. Add a simple input checklist so the assistant has a consistent starting point.

Participant output: A first task map and an input checklist identifying the information needed, what is available and what is missing.

Week 3: Define a useful result

Focus: Make the brief easy for a person to assess and act on.

Activity: Design sections for priorities, progress, unresolved decisions and missing information. Work through a sample by hand, distinguishing stated facts from suggested priorities. Identify where a reviewer must check or decide, and keep owners or deadlines blank when the source does not provide them.

Participant output: A completed one-page task map, a leadership-brief template and clear criteria for a good result.

Module 3: Design Your Agent

Weeks 4–5

Week 4: Set the role and boundaries

Focus: Give the assistant a specific job rather than an open-ended instruction to help.

Activity: Write its role, responsibilities and limits. Add instructions for asking for missing information, labelling suggestions and declining requests outside the task. Try a short sample request and revise any instructions that leave room for invented facts or actions.

Participant output: A role and boundary brief, with a tested instruction for handling a gap or an out-of-scope request.

Week 5: Add relevant business context

Focus: Help the assistant interpret updates against the supplied priorities.

Activity: Prepare a compact sample context pack covering business priorities, project names and relevant roles. Combine it with the instructions and output template. Compare a short response with and without that context, checking whether proposed priorities have a basis in the material.

Participant output: A sample context pack and a complete set of starter instructions, including the output format and human-review requirement.

Module 4: Build a Prototype

Weeks 6–7

Week 6: Assemble the first complete brief

Focus: Turn the preparation into a working starter prototype.

Activity: Follow a guided build on the selected course platform. Combine the role, context and template, then manually run the task with the shared sample notes. Review the complete brief against the Week 3 criteria. Correct unclear instructions before adding anything else.

Participant output: A first working prototype and a marked-up leadership brief showing what is useful and what needs correction.

Week 7: Make the task repeatable

Focus: Produce a useful brief from another set of updates, not just the first example.

Activity: Run a second sample through the same prototype. Improve the consistency of its sections and require important statements to point back to the supplied notes or updates. Check that changed inputs change the result, and that missing information remains visible.

Participant output: A revised prototype, a second reviewable brief and a short record of changes to the instructions or context.

Module 5: Test and Refine

Week 8

Focus: Find weaknesses before mistaking a fluent answer for a dependable result.

Activity: Test a straightforward example, an incomplete example and a request outside the assistant's role. Check factual accuracy, unsupported priorities, invented owners or deadlines, and respect for boundaries. Improve the instructions and repeat failed tests, keeping a record of anything unresolved.

Participant output: One tested starter prototype, a completed test checklist and a clear account of its remaining limitations.

Module 6: Apply It to Your Work

Weeks 9–10

Week 9: Refine the starter and design your own use case

Focus: Consolidate the shared build and choose one manageable personal application.

Activity: Review the Week 8 results, fix one important weakness and retest it. Then revisit the personal-task shortlist. Choose a recurring business task, such as preparing an internal project handover, and define why it matters. Map its inputs, steps, output and human review. Separate reusable parts of the starter from changes the new task needs.

Participant output: A refined starter and a one-page personal-use-case design, with a bounded scope, sample input, success criteria and a list of required adaptations.

Week 10: Begin the adapted build and plan the next 30 days

Focus: Test a first piece of the personal use case without attempting a complete business system.

Activity: Adapt the starter's role, context or output format for one part of the chosen task. Run it manually with sample or anonymised information and check the first result against the Week 9 criteria. Present what works, what needs revision and a continuation plan: further examples to test, improvements to make, who will review them and when to reassess readiness.

Participant output: The beginning of a personal-use-case build, one reviewed first result and a practical 30-day plan. The personal application is not promised to be complete or production-ready by Week 10.

16. Intended participants

The course is for business owners, leaders and managers who turn scattered information into priorities, briefs, follow-ups or decisions.

The emphasis is business judgement and guided practice, not advanced engineering. Participants adapt a common starter rather than design from scratch. They should understand a recurring task and be ready to question and improve the results.

Participants use sample or anonymised material, not confidential live business records.

17. Advanced follow-on course

Build Your Agentic Chief of Staff: Advanced Programme is a separate next course for developing beyond the starter. It covers deeper business context, live tools, procedures, schedules, specialist agents, approvals, monitoring and recovery.

Live production integrations, multi-agent operations, scheduled production workflows and recovery engineering belong in that advanced programme, not this entry course. Any external action requires explicit permissions and approvals.

AI in Action provides a tested starter, the beginning of a personal application and the judgement to decide what needs more work.