Jeremy Fritzhand leading a hands-on masterclass with founders
Jeremy FritzhandAI-Native Workshops
01 Workshop proposal

AI-Native Ways of Working.

A three-day, hands-on AI workshop for ONDC teams. Prepared by Jeremy Fritzhand.

Three days In person or remote Built on your own material
Jeremy Fritzhand · AI-Native Ways of WorkingPrepared for ONDC
02 The shift

AI can now do the work — while a person stays in command.

Agentic AI reads files, runs real commands, researches, analyses, and builds working software. The skill that matters is no longer doing every step by hand — it is directing the work, reviewing it, and approving what ships. This program builds that skill inside your teams, on your own documents.

Delegate Direct Review Approve
Jeremy Fritzhand · AI-Native Ways of WorkingThe shift
03 The arc

Three days, three altitudes.

The program climbs deliberately: understand the tools, then build with them, then turn the method into how a function runs. Each day ends with something the team has made.

Day 01 · Understand

The 50,000-foot view

What these tools are, how they work, and how to use them safely — a clear mental model, no keyboard required yet.

Day 02 · Apply

Hands on the keys

Build something real, from a structured knowledge base to a live webpage, using your own documents.

Day 03 · Operate

An operating model

Turn the method into how functions run: SOPs, a workbook, and a one-page executive console.

Jeremy Fritzhand · AI-Native Ways of WorkingThe arc
04 How the work happens

A loop you can trust.

Every task runs the same supervised loop. The AI drafts; a person reviews; corrections go back into the source, not just the output; then it runs again — sharper each pass. A human approves every consequential action.

1
Draft
The agent produces a first version from your material and instructions.
2
Review
A person checks it against what they know. Nothing ships unread.
3
Correct the source
Fixes land in the knowledge base, so the next run starts from better ground.
4
Iterate
Repeat until it is right. The system compounds; the human stays in command.

Permissions, data access, and human approval stay explicit throughout — safety is part of the method, not an afterthought.

Jeremy Fritzhand · AI-Native Ways of WorkingThe working loop
05 Day one — Understand

What these tools are, in plain language.

On the agenda
  • Agentic AI, explained — terminals, IDEs, and how agents actually run, without the jargon.
  • How Claude works — models, and the effort or thinking settings that change results.
  • What an MCP is — how an AI tool connects safely to email and Google Drive.
  • Safety by design — permissions, data access, and keeping a human in the loop.
  • Where AI earns its keep — the everyday work worth delegating first.
  • Tool-agnostic — the same approach on Claude, Gemini, or ChatGPT.
Leave with

A working mental model

A clear picture of agentic AI — and a personal AI workspace connected to the tools you already use.

Claude Gemini ChatGPT
Jeremy Fritzhand · AI-Native Ways of WorkingDay one
06 Day two — Apply

Hands on the keys — from a PDF to a live site.

On the agenda
  • Build a knowledge base — turn existing documents into structured, AI-readable source.
  • Install the toolchain — Git, GitHub, a free host, Claude Code, and VS Code.
  • Deep research — direct AI to gather and structure what you need.
  • Fan out agents — run several in parallel on parts of one task.
  • The daily loop — draft, review, correct the source, iterate.
Git GitHub VS Code Claude Code Free hosting
Worked example

A 100+ page scheme, made usable

Convert a long government-scheme PDF into a live, searchable website — the exact move behind builds like the Startup India Guide.

Leave with: a knowledge-base structure and a deployable build you made yourself.

Jeremy Fritzhand · AI-Native Ways of WorkingDay two
07 The multiplier

Don't start from scratch — start from what's proven.

The fastest builds begin on someone else's foundation. Before writing anything, we look for an open-source project that already solves most of the problem, then adapt it to the task and the brand. It is the difference between a week and an afternoon.

How to find one
  • Name the job in plain words — then search where the work lives: GitHub, and each tool's own ecosystem.
  • Judge fit fast — is it active, is the licence permissive (MIT or Apache), does it already do most of the job?
  • Fork it and make it yours — re-skin it, wire in your data, and keep a person reviewing each step.
Foundations already in play
Open source · MIT

site2deck

An engine that turns a website's design system into branded, offline slide decks. This proposal is built on it.

Day two's worked example

Startup India Guide

A 100+ page government scheme, turned into a live, searchable website — the same reusable pattern, applied.

Jeremy Fritzhand · AI-Native Ways of WorkingOpen foundations
08 Day three — Operate

From tasks to an operating model.

On the agenda
  • SOPs from real inputs — team interviews, transcripts, and your knowledge base.
  • An Excel workbook — drafted by AI, reviewed in Google Drive, iterated with the team.
  • One executive console — each function's metrics on a single monthly page.
  • Governance — principles for safe, accountable adoption at scale.
  • A 30–60 day pilot — a plan for two or three functions.
The principle

Documents before dashboards

Get the writing and the source right first. A dashboard is just a view over documents that already hold the truth.

Leave with: a repeatable operating model and an executive-console design.

Jeremy Fritzhand · AI-Native Ways of WorkingDay three
09 What you walk away with

By the end, ONDC could have…

Not a slide deck about AI — working artefacts, made with your own material, that a team can keep using on Monday.

A focused pilot
Two or three functions chosen for an AI-enabled pilot.
Knowledge bases
Structured, AI-readable sources for those functions.
Initial SOPs
Drafted from team interviews and the documents you already have.
Working builds
One or more internal tools, webpages, or operational workbooks.
An executive console
A one-page monthly view designed for leadership.
An adoption roadmap
A practical 30–60 day plan, with governance and human approval built in.
Jeremy Fritzhand · AI-Native Ways of WorkingSuggested outcomes
10 Format & logistics

Built around your real work.

The workshop is designed around practical application. Wherever possible, participants work with ONDC's own documents and operational challenges — not generic exercises.

Jeremy Fritzhand · AI-Native Ways of WorkingFormat & logistics
11 From workshop to pilot

A 30–60 day path, not a one-off.

During the three days

Build the foundation

  • Choose two or three functions
  • Stand up their knowledge bases
  • Ship one working build
Next 30–60 days

Run the pilot

  • Turn interviews into SOPs
  • Put each function's metrics on the console
  • Keep the review loop running
From there

Scale the method

  • Extend to more functions
  • Hold the governance and approval standard
  • Documents before dashboards
Jeremy Fritzhand · AI-Native Ways of WorkingThe pilot path
12 Who runs it

Built by someone who builds.

Jeremy Fritzhand

Jeremy Fritzhand builds the programs, relationships, partnerships, and operating structures that help founders, institutions, and communities move from idea to durable outcome — and the AI tools and focused products that support them.

  • He builds the tools too — AI tools, intelligence platforms, and focused websites, made with founders and teams.
  • Credibility from doing — the workshop teaches the exact loop he uses in his own work, not a generic framework.
Ecosystem architecture Founder coaching University innovation AI tools & digital products
Jeremy Fritzhand · AI-Native Ways of WorkingWho runs it
13 Next step

Let's run it with your teams.

A short call is the fastest way to scope the pilot — which two or three functions, in person or remote, and the dates that fit your calendar.

Jeremy Fritzhand
fritzhand.com linkedin.com/in/fritzhand calendly.com/jfritzhand
Jeremy Fritzhand · AI-Native Ways of Workingfritzhand.com
01 / 13
← → navigate