Subscribe now

Growth Product Managers focus on improving value from an existing product and helping it grow. They make optimizations and introduce new product experiences that help users find value faster, come back often, and drive better business outcomes.

Facebook created one of the first dedicated Growth teams back in 2007. A lot of how that team operated, such as working cross-functionally, experimenting constantly, and using the product itself to drive growth, has since become common ways of working for growth teams today.

Growth PMs differ from other types of Product Managers in how they operate:

• Focus on outcome metrics: Growth PMs are usually more directly tied to a business outcome. They start with the metric they want to move and use user problems and behaviors to understand what is holding the metric back. Traditional PMs are more likely to start with the user problem first and then connect solving it to the business outcome. Growth work is therefore less focused on feature output and more on identifying the product changes most likely to improve acquisition, activation, retention or monetization.

• Very cross-functional with limited surface ownership: Growth PMs often work across multiple parts of the product rather than owning one surface end to end. That means working closely with the teams that own those surfaces to identify and make improvements.

• Data-driven, with a focus on experimentation and learning fast: Growth PMs tend to be especially data- and experimentation-driven. They run experiments across the customer journey to quickly understand what works, what doesn’t, and where to invest next.

An analogy might help here: Imagine your product is an airport. Core Product Managers build the terminals, gates, and services that give travelers a reason to use the airport in the first place. Growth teams focus on making the journey through the airport easier — helping people find the right place faster, reducing friction at key steps, making relevant services easier to discover, and creating an experience that makes people more likely to come back.

Table of Contents

• How to create a Product-Led Growth Strategy and Roadmap

  • Identify the right growth opportunity

  • Identify growth levers

  • Develop Solutions

    • Make it effortless

    • Target the right user, at the right time, in the right place

    • Guide and educate

    • Let users experience value before they pay

  • Balance optimization with bigger bets

  • Define how to measure each solution

  • Prioritize what to work on

  • Test, learn, and productize

• Final thoughts

How to create a Product-Led Growth Strategy and Roadmap

Now that we have established what a Growth PM does, let’s dive a bit more into how to create a growth strategy and experimentation roadmap.

Before you start

Make sure these are true before building a growth roadmap:

• The underlying product has product-market fit. Without it, you may end up trying to grow something customers don’t really want.

• The company has enough data and data fluency to run and interpret meaningful experiments, and the organization values experimentation as part of how decisions are made.

• The decision is worth experimenting on. Experiments take time and resources. If the cost of testing is higher than the value of testing, then it might be better to make the decision and move on. For example, a small bug fix might not be worth running an experiment on.

Principles to keep in mind:

• Don’t optimize for the goal metric at the expense of the broader product experience. As you improve one area, make sure you’re not inadvertently hurting another.

• Growth work should be long term focused and shouldn’t focus on short term hacks.

A hacky growth strategy can look like this. You will have some spike but ultimately it will drop.

You ideally want a step change or even better, growth that compounds over time.

1. Identify the right growth opportunity

Start with the company strategy and the business outcome you are trying to improve.

Pick a metric that reflects the real outcome, not just an activity along the way. For example, clicks, page visits, or time spent on a page can help you understand user behavior, but they are not always the end goal. Growth PMs focus on metrics around acquisition, activation, retention, and monetization, including CAC, sign-ups, free trial starts, TTV (Time to Value), activation rate, conversion/purchase rates and ARPU.

You can find opportunities using a combination of product data, differences between user segments, customer research, and patterns in how your best users behave.

Then prioritize the broader potential of the opportunity. Could it create a compounding rather than linear impact? Could it strengthen a growth loop, where one user’s action brings in more users or activity, which brings in more again? (Adding friends on Facebook fills your feed, which makes you post more, which gives other people a reason to add you.) Will it create long-term value rather than just a short-term lift? Does it unlock other opportunities? And importantly, does it move the company strategy forward?

For example, if Uber sees that customers who use multiple products use Uber more frequently, have higher retention, and generate higher gross bookings than customers who only use rides, a growth opportunity could be:

Convert more Mobility-only customers into cross-platform customers (i.e. get customers who only use Uber for rides to start using another Uber product)

2. Identify growth levers

Once you know which growth opportunity to focus on, identify the behaviors, use cases, or product experiences that are most likely to move the outcome. These are your growth levers.

You may find several potential levers. Look at how improving activation of these levers are correlated with better acquisition, retention or monetization, while being careful not to assume that correlation means causation.

Going back to the Uber example, the opportunity is getting mobility-only customers to try another Uber product. Possible levers could be:

• A first Uber Eats order, for example through an offer shown after a ride

• Joining Uber One, which gives benefits across rides and delivery

• Using Uber for a new need, like groceries or package delivery

Once you identify the potential levers, diagnose where users are getting stuck and why.

• Is the feature hard to discover?

• Is the value unclear?

• Is it difficult to use?

• Is it irrelevant to that user?

• Is functionality missing?

• Or is there another source of friction?

Use funnel analysis, segmentation, cohort analysis and research to understand which frictions are actually blocking users in the journey.

The table below provides a starting set of funnel metrics you can look at. It needs to be adapted to the nuances specific to the feature or product you’re working on.

Look for the biggest gaps and sources of friction. A lever may have strong potential, but if users already adopt and successfully use it at high rates, there may be limited room to improve. Conversely, a promising lever with low adoption or poor completion may have more room for growth.

Using the Uber example, you could look at whether customers who activate on each lever go on to have better retention, frequency, or gross bookings. You could also look at how much room there is to improve each lever. If using Uber for a new need, like groceries, is strongly associated with better outcomes but only a small percentage of mobility-only customers currently do it, it may be the most promising growth lever.

3. Develop Solutions

Now brainstorm solutions based on the growth opportunity and growth levers you have finalized.

Below are details on some ways I have found to effectively bucket these solutions.

A. Make it effortless

Make the path to value effortless by removing unnecessary friction without stripping out important features people rely on.

B. Target the right user, at the right time, in the right place

Sometimes the feature isn’t difficult to use but the user isn’t aware of it at the right time and context. Introducing the feature when the user is most likely to use it improves the chances of adoption.

C. Guide and educate

Even if users are aware of a particular feature they might not know why they should use it or how to use it successfully.

D. Let users experience value before they pay

Let users try the product, see the value for themselves, build a habit, and become more willing to pay. This can also make acquisition more efficient. The tricky part is not giving away too much for free. You want to give users enough value to keep using the product, while still giving them a reason to move up the monetization escalator. Some options here include Freemium, Free Trial, and Reverse Trial.

When users are ready to upgrade, make sure the paid version feels worth it, the upgrade process is easy, and they actually have authority to make the purchase.

Also think about whether the monetization model itself is creating friction. Pricing and monetization can have a large impact on growth. Depending on the product, the right model might be advertising, subscription, transaction-based pricing, or a combination of these. The key is to make sure the monetization model fits how users experience value from the product.

Advertising: The product can stay free for users, while the company makes money from advertisers. Instagram and Facebook are common examples.

Subscription: Users pay monthly or annually to keep using the product. This makes sense for products people use regularly, like Netflix.

Transaction-based: Users pay when they actually use or buy something. Airbnb is an example. You pay when you make a booking.

Many products have a combination of different monetization models. Example: Spotify uses both advertising and subscription based models.

4. Balance optimization with bigger bets

Growth work isn’t just about optimizing existing experiences. Growth work needs to focus on innovation and 0-1 experiences as well. In a large company where products have had product-market fit for a while, optimization can be a great strategy. A 0.02% improvement on $10 billion in revenue is still $2 million in additional revenue. But if you want to go big and stay ahead of the competition in an ever-evolving software industry, you need to innovate as well. The returns can be much larger when you get a new product, use case, or growth loop right.

As a growth PM, you need to seamlessly move between optimization and innovation based on the stage of the product and company. My personal take is that even at an established company, a meaningful part of the growth roadmap should be focused on innovation, alongside optimizing what already works.

5. Define how to measure each solution

Before prioritizing and testing a solution, decide how you’ll measure whether it’s working, using these metrics:

• Primary/Outcome metric: The main outcome you expect the solution to move. (covered in Section 1).

• Secondary metrics: Other important outcomes the solution could affect.

• Tradeoff metrics: Metrics you don’t want to hurt while improving the primary metric.

• Leading indicators: Earlier behaviors that can tell you whether the solution is moving in the right direction.

6. Prioritize what to work on

You will probably end up with more solutions than you can work on. Use prioritization techniques like ICE + Time (Impact, Confidence, Effort/Cost, Time to get results) to rank the solutions and size them as Small (S), Medium (M), or Large (L) across the following dimensions. ICE isn’t perfect but it gives you a good direction on what to work on.

(I) Impact: How much do you expect the solution to move the outcome metric?

(C) Confidence: How confident are you that the solution will actually work and have the expected impact?

(E) Effort/Cost: How much engineering and other cross-functional effort will be required to build and test the solution?

Time to get test results: How long will the experiment need to run after launch before you have enough data to make a decision? This is largely determined by the required sample size and the amount of eligible traffic.

Once you have sized the ideas, compare them based on potential upside and risk. Ideally you would want to focus on high upside experiments with low downside.

Also think about sequencing. The highest-impact idea may not always be the first thing you work on. Sometimes a smaller initiative needs to happen first because it unlocks a larger opportunity later.

7. Test, learn, and productize

Once you have prioritized the solutions, test whether they actually work before investing in the full product experience. (Running experiments well, including sample size, statistical significance, novelty effects, and holdout groups, deserves its own post.)

Start with a clear hypothesis. Be clear about what you are changing, who you are testing it with, and what you expect to happen. You should already know the primary, secondary, tradeoff, and leading metrics you want to track.

Build enough to learn. Some ideas may have a lot of upside but require significant investment or rely on unvalidated assumptions. In those cases, test a smaller version first. Build the simplest version that can give you a reliable answer before investing heavily.

Run the test and evaluate the results. Look at whether the primary metric moved, but also check secondary metrics, tradeoff metrics, leading indicators, and any important differences across customer segments.

Learn and iterate. Don’t think of every test as simply a win or loss. Try to understand why users behaved the way they did and use those learnings to improve the solution or rethink the assumptions behind it.

Productize what works. If the solution shows enough promise, turn the test into a production-ready experience. This may mean improving the UX, building the proper backend and integrations, supporting more platforms or use cases, and rolling it out more broadly.

Keep measuring after launch. Make sure the impact you saw in the test holds up once the solution is exposed to more users, and capture the learnings so future experiments can build on them.

Final thoughts

Growth isn’t about one big win. It’s about building a steady system: find what’s holding your metric back, test fixes quickly, and build on what you learn. Start small, stay curious, and let each experiment shape the next.

Subscribe now

As a product manager, one has to constantly deal with ambiguity while making many decisions: what to build and why, when to cut scope, and what to shut down. The goal is to make informed decisions with all the information available while understanding the consequences of the decision.

Here are certain methods I have incorporated that have helped me make decisions faster and align teams.

1. Calibrate effort to importance

The same effort of analysis is not needed for every kind of decision. A simple change that can be quickly tested and easily reversed if it doesn’t work does not require the same amount of scrutiny as a decision that would be difficult to reverse or one that has long-term consequences. Although some ambiguity can be de-risked through experimentation, every experiment you run comes with costs, time, and other implications.

Before deciding how much rigor to apply, I try to answer three questions:

How costly is it to be wrong?

Try to understand what happens if the decision fails. Will we lose weeks of engineering and cross-functional partners’ time, will we hurt the customer experience in a way that they might not come back, will it create regulatory risk, or will it move the org in the wrong strategic direction?

The greater the consequences, the more important it becomes to understand the assumptions and risks before committing.

How easy is it to reverse?

Some decisions are easy to test and undo. Others create technical, customer, partner, or organizational commitments that are expensive to unwind. The harder a decision is to reverse, the more scrutiny it deserves upfront.

What is the cost of waiting?

Not making a decision and putting more effort into analysis before making a final call has its cost too. Waiting can mean slower learning, lost revenue, missed opportunities or delaying other work that depends on the decision.

These 3 questions, taken together, help determine how much rigor a decision deserves.

2. Structure the options and drive the decision

2.1 Option mapping

After going through the “Calibrate effort to importance” step above, if you decide the decision warrants more rigor, create an options table that lays out the main choices and their tradeoffs. Then draft the first version yourself and circulate it to a small group of relevant partners for feedback. Strengthen the analysis with qualitative input, data, and first-principles thinking. Make sure to time-box your decisions so there is a hard deadline by which the decision needs to be finalized. Otherwise, you run the risk of analysis paralysis.

Try to keep the final set to three options. To make the tradeoffs easier to compare, summarize them in different colors across different criteria. Green highlights a relative strength, yellow flags a meaningful tradeoff that needs more scrutiny, and red points to a weakness or risk.

The colors are not meant to create a mechanical score but the purpose is to make the tradeoffs easy to see and focus the discussion on which ones matter most for the decision.

Example trade-off table

2.2 Leadership buy-in

Once you have the right set of options mapped, get buy-in from the right people and leadership. Some firms have formal product reviews, others have a more informal approach. Either way, make sure you have the right people and leadership to finalize the decision. Only involve people who are absolutely necessary to finalize the decision. Everyone else gets the summary. Try to use frameworks like DACI to assign roles:

D=Driver: The person responsible for coordinating stakeholders, getting all the information and pushing the decision forward.

A=Approver: The one person who makes the final call. Ideally, you want a single approver, but in complex cross-functional orgs, there may be multiple approvers.

C=Contributors: The cross-functional partners who are subject matter experts and can provide essential input.

I=Informed: Partners who need to stay in the loop about the decision.

At the end, make sure you capture the decision rationale, key trade-offs, and revisit triggers so the team does not reopen it unless the underlying assumptions change.

3. Operationalize and communicate the decision

The work is not finished yet. One needs to operationalize the decision and make sure stakeholders are aware what decision was taken and why a certain decision was taken.

• Communicate the decision through the right channels based on the audience and impact: 1:1s, team meetings, Slack, leadership or cross-functional forums, and internal announcement channels where appropriate.

• Document the decision and rationale so people can refer back to what was decided and why.

• Assign clear owners, follow-up actions, and expected timelines so the decision translates into execution.

Final thought

The goal is to understand which uncertainties matter, match the rigor of the process to the stakes, make the trade-offs explicit, and learn quickly when the underlying assumptions change.

As a PM’s scope grows, you can’t be in every decision. The real leverage comes from helping teams make good decisions on their own by giving them the right context, principles, and clarity around the trade-offs.

Subscribe now

There’s a lot of noise around agents and it gets confusing fast: Claude Code, Codex, OpenClaw, Hermes, Cowork, RAG, MCP. New tools and terms seem to appear every week.

It’s easy to feel overwhelmed and not know where to start. I spent some time figuring out what everything does, how the pieces fit together, and how to build a custom agent of my own. I can promise you, it’s not as complicated as it first appears.

To prove it, I built a travel-planning agent. I give it a destination, a budget, and my travel dates, and it gives me back a day-by-day itinerary with real prices, recommendations and map links pulled together from Google Maps, TripAdvisor, YouTube, and tourism websites.

Here’s exactly how I built it, step by step.

What is an Agent?

OpenAI has the clearest definition I’ve found:

An agent is a system that independently accomplishes tasks on your behalf.

An agent can take a task and work through the steps needed to complete it. It can decide what to do next, use tools, look at the result, and continue until the job is done.

Underneath, there are three parts.

Instructions are the rules and workflow you write. This is the only part you actually make, and it’s where nearly all your time goes.

The model is the LLM doing the thinking. Claude, GPT, Gemini.

Tools are how it reaches outside itself. Web search, your files, Google Drive.

The agent keeps going until the task is complete.

For example, if you ask an agent to research a trip, it might search for flights, compare hotels, check restaurant reviews, and put everything into an itinerary. You give it the goal; the agent figures out how to get there.

First: Decide which tools to use

Not every task needs a custom agent. There is plenty of overlap here and the boundaries will keep blurring as the tools evolve, but in broad strokes you have two choices:

1. Configure an existing tool

Tools such as Claude Cowork, Hermes Agent, OpenClaw, Zapier Agents, Gumloop and n8n already provide the basic system. Some work like ready-made assistants. Others let you connect apps and arrange the steps yourself. You can use them to review emails, create reports, process files, run scheduled routines or update a CRM.

The amount of setup varies, but you are starting with a system that already exists.

2. Build your own custom agent

Use Claude Code or Codex when you need to create the system yourself and the task is more complex. This gives you more control over the steps, rules, workflows, data sources, interface, and behavior.

An agent like this might check whether a sales lead meets your requirements before anyone contacts them, research a topic and stop if the evidence is weak, or review an invoice, flag what is missing, and request approval above a certain amount. In each case, the agent needs to know not just what to do, but when to stop, check, or ask for approval.

The agent I created is a travel planner. I give it a destination, my dates, and a budget, and it researches the trip across Google Maps, TripAdvisor, YouTube, and local tourism sites. It checks things like ratings, opening hours, travel times, and current prices, then pulls everything together into a day-by-day itinerary with suggested areas to stay, restaurants, attractions, and activities.

I mostly used Claude Code. Many of the tools above ultimately rely on the same underlying models—Claude, GPT, or Gemini—so I preferred working closer to the model itself. Claude Code gave me more control over the instructions, context, tools, and workflow, instead of having another product layer decide how the model should behave.

The rest of this post is about how I built it. You can follow the same steps for whatever agent you actually need.

Step 1: Install Claude Code

Check Step 1 in my previous blog post about how to create an app and publish it.

One thing to know up front: Claude Code requires a paid Claude plan. It isn’t included in the free plan.

Step 2: Install VS Code

VS Code is a free code editor from Microsoft. You can build your agent straight from the terminal if you want (on a Mac, hit Cmd+Space, type Terminal, press enter), but VS Code has a much nicer interface and shows you every file the agent writes.

Download VS Code from https://code.visualstudio.com/download and download the Mac version. Install and open it.

Once opened, go to Extensions in the left sidebar, search for “Claude Code,” install it, and sign in.

Step 3: Set up the project and write your CLAUDE.md

In VS Code, go to File → Open Folder and create a New Folder for the project. I called mine AI Agent.

Inside that folder, click the new-file icon at the top of the Explorer and name the file CLAUDE.md. The name matters—this is the file Claude Code looks for when it starts working on a project.

CLAUDE.md is a simple Markdown file that stores the instructions and rules you want Claude to follow every time it works on that project. Think of it as the project’s instruction manual: write the guidance once instead of explaining it again in every session.

Keep it concise. Anthropic recommends targeting under 200 lines because longer files consume more context and are followed less reliably.

Also keep it structured. Use Markdown headers and bullets instead of dense paragraphs.

Here is the CLAUDE.md I used for my travel agent. Feel free to copy it and adapt it for your own:

# Project Context

Travel research workspace. I use it to plan trips: cost research, timing, and detailed itineraries.

## About Me

Not a travel professional. I want honest numbers over aspirational ones, and I’d rather know a trip doesn’t fit my budget early than read a beautiful itinerary I can’t afford.

## Before Starting Work

- **Open workflows/travel-research.md and follow it. Every time, including short-notice and single-day trips.** It contains a source-completeness gate — run it before showing me anything.

- Ask the clarifying questions in the workflow before researching. Don’t assume budget, dates, or travel style.

- Show a plan and wait for approval before writing files.

- If the budget can’t support the trip as described, say so directly and propose what would work instead.

## When There Isn’t Time

- Urgency does not lower the research standard or silently remove required checks.

- If you cannot complete the full source checklist before the deadline:

• Say so before presenting the plan.

• List exactly what you checked and what you did not check.

• Label the output as preliminary or incomplete.

• Do not present an incomplete plan as fully researched or finished.

## Research Standards

- Prefer sources from the last 18 months. Travel info goes stale fast — prices, closures, visa rules, transit changes.

- Cite every price and every claim with a link.

- Give price ranges, not single numbers, and say what the range is based on (season, booking window, source).

- Distinguish what you verified from what you estimated. Label estimates as estimates.

- Discount sponsored YouTube content and note when a recommendation appears mainly in listicles rather than firsthand accounts.

## Files

- workflows/ — reusable process instructions

- output/ — finished plans, named YYYY-destination-plan.md

- resources/ — screenshots, booking confirmations, saved research

Edit files in place rather than regenerating them.

Step 4: Write the workflow

Now create the workflow. You do not need to write the file yourself. Describe what you want in plain English, and Claude Code can write it and save it in your project.

I gave Claude this:

I want to build a reusable workflow for researching travel plans.

The goal: I give you a destination and my constraints, and you produce a costed, day-by-day itinerary I can actually book from.

Before researching, ask me clarifying questions: budget per person, trip length, departure airport, who’s travelling, what pace I want, how flexible my dates are, and any must-dos or dealbreakers.

Two things about the order of the research. Figure out the best month to go before you price anything, since flights, hotels and crowds all depend on the season. And check whether the total cost actually fits my budget before you write the itinerary. If it doesn’t, tell me and stop rather than trimming the plan to fit.

For things to do, cross reference Google Maps, TripAdvisor, YouTube and the official tourism site, and tell me where they disagree. Build the itinerary grouped by neighborhood so no day crosses the city twice, and end with coordinates I can import into Google Maps.

Never present historical price ranges as if they were live fares.

Save the workflow to workflows/ and ask me questions if any of this is unclear.

Claude created a file called workflows/travel-research.md.

That file gives it a repeatable process: what questions to ask, what to research, when to check the budget, and how to structure the final itinerary.

After that, I no longer need to repeat all the instructions. I can simply say:

“Create an itinerary for me for Madrid, Spain”

Claude loads the project instructions (CLAUDE.md), follows the travel workflow, and starts from there.

Step 5: Give the agent access to the tools it needs

The workflow tells the agent what to do. But if it needs information from other services, you also have to give it access to them.

You may not need any extra tools. For my travel agent, I did.

I asked Claude Code to help me connect the services one at a time. My instructions were roughly:

Google Maps: “Set up Google Maps so the travel workflow can check ratings, opening hours, whether a place is still open, coordinates, and actual travel times. Tell me what API I need to enable and where to add the API key.”

Amadeus: “Set up Amadeus so the workflow can look up current flight and hotel prices. Walk me through getting the API credentials and then add them to this project so you can use them.”

YouTube: “I want the workflow to use YouTube videos as a research source and actually read the transcripts. Find a way to pull public YouTube transcripts and set it up for this project.”

Claude handled much of the technical setup from the terminal. I still had to sign up for some of the services and get the required API keys, but Claude told me what I needed and where to put them.

Step 6: Iterate and improve

Once everything was connected, I gave the agent a real trip to plan.

“Create an itinerary for me for Madrid, Spain”

The first run immediately exposed things I had not thought about. Some instructions were too vague, some steps happened in the wrong order, and some outputs needed better checks.

Instead of rewriting everything myself, I told Claude what went wrong and asked it to update the project.

If the correction applied to how Claude should always work, I added it to CLAUDE.md.

If it was specific to travel planning, I added it to the workflow.

Then I ran it again.

That is really how the agent gets better. You do not need to design the perfect system upfront. Build a first version, test it on a real task, see what breaks, and keep improving the instructions.

What the agent actually produced

The final output was much more than an itinerary. For a Madrid trip, the agent:

• chose the best travel window before pricing

• estimated the full trip cost, including taxes, parking and tipping

• compared areas to stay based on cost and travel time

• researched attractions and restaurants across Google Maps, TripAdvisor, YouTube transcripts and local sources

• showed where sources disagreed

• grouped each day by neighborhood to avoid unnecessary travel

• created the final Markdown itinerary

• generated a KML file I could import into Google My Maps to see the full itinerary organized by day

The full plan is much longer, so I put the complete output here

My agent happens to plan trips. But the same setup of persistent instructions, a reusable workflow, and the tools it needs could work for competitive research, expense reports, or any other task.

Subscribe now

TL;DR

You can build and publish an iOS app without knowing how to code:

• Plan it with Claude: Pressure-test the idea, define the first version, and create a milestone-based build plan.

• Design it with Claude Design: Turn the requirements into screens and prepare the files for development.

• Build it with Claude Code: Give Claude Code the plan and design files, then let it create and run the app.

• Test it: Open the app on your iPhone, test each milestone, and fix problems early.

• Ship it: Build the production app, distribute it through TestFlight, and submit it to the App Store.

The basic workflow is simple: plan → design → build → test → ship.

Steps to Build and Publish an App

Note: This guide is written for Mac users publishing to Apple’s App Store. The same process works on Windows, but a few setup steps, like opening and using the command line, will be slightly different.

The distance between “I have an idea” and “it’s on my phone” feels enormous. Claude closes that gap enough for anyone to ship a working app to the App Store. This playbook walks through the process step by step, with each step designed to be practical and easy to follow.

What you’ll need

Set these up before you write a single prompt. None of them require coding. It’s just clicking through onboarding flows, and getting it out of the way first keeps you in flow later.

Step 1: Install Claude Code

Here are the steps to install Claude Code onto your computer.

1. Install Node.js: Go to https://nodejs.org/en/download and click the macOS Installer (.pkg) button. Run it like you install any other software on your Mac: double-click, click through the prompts, done. (Node is what the Claude Code install command below runs on.)

2. Open the terminal: Press Cmd+Space, type “Terminal,” and hit Enter. A plain text window opens.

3. Install Claude Code: Paste this and hit Enter:

npm install -g @anthropic-ai/claude-code 

If you get a permission error, point npm at a folder you own, then install again:

npm config set prefix ~/.npm-global 

npm install -g @anthropic-ai/claude-code

If you used that second version, run these two lines once so your terminal can find the claude command:

echo 'export PATH=~/.npm-global/bin:$PATH' >> ~/.zshrc 
source ~/.zshrc

4. Type claude --version and hit Enter. A version number means you’re set. If it says “command not found,” close the terminal, open a fresh window, and try again.

Step 2: Develop your strategy (Claude chat)

Before any code, run a real analysis on your product idea, then turn it into a concrete list of what the app actually has to do. Open a fresh chat at claude.ai and treat Claude like a skeptical strategy consultant you’ve hired to poke holes in your idea.

The sequence that worked for me:

First, the pitch and the pressure test. Describe the idea in a few sentences, then ask for a full, skeptical analysis: who the customer is, who the competitors are, what the market looks like, why now, and what would most likely kill this.

Then, attack your own assumptions. After reading the first pass, push back on yourself. Ask what you might be wrong about and what the strongest version of the case against the idea looks like.

Next, pin down the requirements. Once you’ve settled on what you’re building and why, get specific about what the app actually has to do before you plan how to build it. Ask Claude to write the requirements for a first version: the core features, the screens the app needs and what each one does, the data it stores, and the rules that govern it, such as who can do what, what happens on sign-up and how the app works.

Finally, turn the requirements into a plan. With the feature list agreed on, ask Claude to turn it into a build plan that takes you from nothing to a real iOS app you can put in front of people, broken into three milestones. For each milestone, have it list the screens to build, the backend pieces, and the tests to run. Milestones are useful because they break a large app into smaller, testable stages.

Together, the requirements and the plan are your master spec. Copy the whole thing into a document you’ll keep returning to throughout the build.

Step 3: Design (Claude Design)

Now you turn your requirements into screens you can actually see.

Open Claude Design at claude.ai/design (or the palette icon in the desktop app). It’s part of your paid plan. Start a new project, separate from your Step 2 chat.

Then, one screen at a time:

1. Give it a style reference. Drop in some mobile design examples so Claude knows what good design looks like. Figma’s Community has plenty to pull inspiration from: https://www.figma.com/community/mobile-apps

2. Paste in your milestone plan from Step 1. This tells Claude which screens to make and what goes on each.

3. Ask for the designs. Describe your brand, colors, fonts, and overall mood. Ask for a few versions so you can compare.

4. Refine. Tell it what to change (spacing, headline size, fonts) until the screen looks right. This takes a few rounds. That’s normal.

5. Repeat for every screen (welcome, onboarding, main view, paywall, settings).

6. Ask for the handoff. Once the screens look right, tell Claude Design to package everything for Claude Code: “Now create a prompt I can use in Claude Code, along with the downloaded files, to build these screens.”

That handoff will usually give you a downloadable folder with everything inside:

• PROMPT.md — the prompt you’ll paste into Claude Code to build the screens.

• README.md — the full spec: layout, components, colors, fonts, and copy.

• The design files — your screens as an HTML file you can open in a browser, plus the stylesheet.

Before you leave Claude Design, download the folder and unzip it.

Step 4: Build (Claude Code)

This is where the mockups become a real app on your phone. It all happens in the terminal, where you’re talking to an AI engineer that can read, write, and run code on your machine.

Getting started:

1. Open a terminal: On a Mac, press Cmd+Space, type “Terminal,” hit Enter. A window with a blinking cursor appears.

2. Make a project folder: type mkdir myapp, then cd myapp to step inside it. Nothing visible changes, but you’re now “in” the folder. (”myapp” is just the folder name. Call it whatever you like.)

3. Take the files you unzipped in Step 3 and drop them manually into your new myapp folder that you just created, so everything sits in one place.

4. Launch Claude from inside that folder: type claude and hit Enter. It greets you right there in the terminal. Then give it this:

"Read PROMPT.md and README.md in this folder, then follow PROMPT.md to build the app."

Claude Code runs the actual commands in front of you, installs everything, builds the app, and stops.

Step 5: Get it onto your phone

You don’t really know what you’ve built until it’s in your hand. Here are the steps to test the app on your phone.

1. Install Expo Go from the App Store. It runs your code on your phone in dev mode.

2. In a new terminal window (Cmd+Space, type “Terminal,” hit Enter), run:

npx expo start --tunnel

It prints a QR code.

3. Scan the QR code with your phone’s camera. Expo Go opens with your app inside it.

That’s how you test the app as you build.

Step 6: Ship it to TestFlight and the App Store

This is the final stretch: moving your app off your computer and onto a real phone—first through TestFlight, Apple’s testing platform, and then through the App Store.

Before you begin, make sure you have the two accounts listed in the setup section:

• Apple Developer Program: $99 per year. Approval may take anywhere from a few hours to a couple of days, so register early.

• Expo account: Free.

Most of the technical work happens through EAS, Expo’s cloud build service. EAS builds your iOS app on Apple hardware in the cloud. Let EAS manage the certificates for you. Signing is one of the most confusing parts of iOS development for beginners, and EAS handles most of it automatically.

Building and uploading your app. Run these commands from inside your project folder in Terminal. In Terminal you use the cd command (”change directory”). Since you made your project in the myapp folder back in Step 4, you go there with:

cd myapp

• Install EAS and log in: (this is your Expo login and password)

npm install -g eas-cli 
eas login 

• Prepare the project:

eas build:configure

This is a one-time setup that tells EAS how to build your app. It’ll probably ask you for a bundle identifier during setup. A bundle identifier is a unique ID Apple uses to tell your app apart from every other app in the world. It looks like a website address in reverse: com.yourname.yourapp. You set it once and it’s permanent, so pick it now.

• Build the app in the cloud:

eas build --platform ios

It asks you to sign in with your Apple account, handles the certificates, and hands you a finished build.

• Send the build to Apple:

eas submit --platform ios

After ten to thirty minutes of Apple processing, the build appears in App Store Connect under TestFlight.

Testing on TestFlight. TestFlight is Apple’s tool for trying your app on real phones before it goes public. The quickest way is to add yourself and a few people as testers by email. You all get the app right away, no waiting. To share more widely, TestFlight lets you invite up to ten thousand people through a public link. Then actually use the app, and get a few others to use it too. This is where you catch the things that only show up on a real phone in real use: the sign-up that fails, the screen that loads blank, the button that goes nowhere.

Submitting to the App Store. TestFlight only tests, it doesn’t publish. To actually go live, you fill out your store listing in App Store Connect and submit it for review. Here’s the whole path:

1. Go to appstoreconnect.apple.com and sign in. Click Apps, then the blue plus button, then New App.

2. Fill in the basics: platform (iOS), your app’s name (it must be unique across the entire App Store), primary language, the bundle identifier you set earlier, and an SKU (any short code you make up for your own records, like app001).

3. Add your screenshots. Apple requires them at set sizes, the 6.9-inch iPhone size is mandatory, plus iPad sizes if your app supports iPad. These are your storefront, so make them count.

4. Write the listing text: a description, keywords, your category, and a support URL (a webpage people can reach you at, even a basic one).

5. Fill in privacy details. You must provide a privacy policy URL and answer a questionnaire about what data your app collects and why.

6. Set the age rating by answering a short content questionnaire.

7. Attach your build. The one you uploaded with eas submit will be waiting under the build section once TestFlight finishes processing it. Select it.

8. Click Submit for Review.

Then you wait. Apple’s review usually takes one to three days. Once approved, your app will be published on the App Store.

Don’t panic if you’re rejected. Most first submissions bounce once over something small: a missing permission sentence, a support link that doesn’t load, or placeholder text left in by accident. Apple tells you the exact reason. Fix what is needed and resubmit. You don’t have to start over.

The gap between an idea and a real app on your phone is smaller than it looks. It mostly comes down to doing things in the right order: pick the idea, argue it out, design it, build it screen by screen, and get it in your hand early and often.

A dinner is a forcing function. You can't swipe away from a table. You can't ghost someone passing the bread. Two hours, five people. Whatever happens, happens in person. Available in Bay Area, LA, Chicago, and NYC.

Backend matching that does the work for you. Onboarding is answering a short personality questionnaire, sharing your interests, and connecting your social accounts so we learn more about you. You never write a bio.

𝐓𝐡𝐞 𝐜𝐚𝐝𝐞𝐧𝐜𝐞. Two dinners per month for a small subscription, plus the cost of your meal. Each Monday our matching engine runs city by city and notifications go out, so you always know whether you're seated for the next dinner or waiting in the pool for a future seat.

𝐖𝐡𝐚𝐭 𝐲𝐨𝐮 𝐝𝐨. Once your dinner is finalized, tap I'll be there on the notification. Show up. After dinner, group chat unlocks within the app. Keep the conversation going if it was a good night, or let it be a good night that doesn't repeat. Your call.

𝐓𝐡𝐞 𝐚𝐝𝐦𝐢𝐧 𝐩𝐚𝐧𝐞𝐥. Where the operation actually runs. We match strangers, set up dinners, and keep a running waitlist of people who are interested but couldn't be seated this round. From here we see every attendee and every RSVP.

𝐖𝐡𝐚𝐭'𝐬 𝐜𝐨𝐦𝐢𝐧𝐠 𝐩𝐨𝐬𝐭-𝐏𝐌𝐅. Post-dinner chat, deeper matching with LinkedIn, Instagram, and Facebook signals, a map view of the restaurant, post-dinner ratings and feedback, user verification, email notifications, and more cities.

𝐓𝐡𝐞 𝐬𝐭𝐚𝐜𝐤. Expo + Supabase, built solo with Claude Code and Claude Design.

You’ve no doubt experienced the impacts of the Great Resignation. Several colleagues have likely left your organization in recent months, maybe you are thinking of leaving as well. The pandemic has given people the headspace to think more deeply about their lives. Where are we going with our careers? What skills do we need? What are we good at? There are many questions to answer.

A few decades back, it was commonplace to have the same job throughout one's life. Not only was this a societal norm, there was much less knowledge available about other careers. Knowledge in the past was guarded by those who had access to it; now anyone with an internet connection has access to more knowledge and information than many experts of the previous generations. If you’re interested in having multiple careers, now’s a great time! That said, changing careers is much easier for those with the resources and means to do so. But read on, for ideas that might help anyone looking to think about alternate career paths.

Subscribe now

Are you a rockstar or a superstar?

Let’s begin by exploring career trajectories in the modern knowledge economy. They often take two forms. Figure 1 is an example of someone finding their calling early and moving up the ladder with each new role in a given field. You can also call these type of people the “rockstars,” a term Kim Scott in Radical Candor uses – rockstars are experts in their field, or specialists. I must say I am envious of those who have found their calling and have built on their strengths in one domain. For many, you might fall into Figure 2, which has a lot of loops and setbacks, but ultimately can lead you to where you want to be. These types of people are the “superstars” per Kim Scott – they have different passions and like trying out different jobs and careers. The world needs both rockstars and superstars, so it is a matter of finding out what interests you most. Also, you might not want to climb the career ladder; and that's fine too!

Figure 1: Rockstar Career Figure 2: Superstar Career

How to make choices for the long term

The clichéd term of “your career is a journey, not a destination” is true. We need to think of our careers long term. There are so many skills I have acquired that I didn’t know I was good at and others I thought I would like but did not in practice. Annie Duke talks about the 10-10-10 framework when making a big decision. We can use the same framework when making a career decision. How will you feel with the decision 10 minutes from now, 10 months from now and 10 years from now? When making hard decisions or moves, it might be painful and difficult for you in the short term but if you think it will take you where you want to go 10 years from now, then probably you have made the right choice.

That said, your vision of a certain career might be completely different from the reality as you might have painted a wrong image in your head. To counter this, I recommend talking to as many people as you can about a certain path to learn if that career is right for you. Try to understand from them about how the organization is in reality and not the buzzy branding that you will usually find on the internet. Glassdoor is a great platform to find out more about the organization. One might argue that only those who dislike the organization or love it a lot will usually review the organization on Glassdoor. The actual truth of what the company is like is probably something in between. Use it as one lens of many to look at the organization. At the end of the day, you need to ask yourself, while keeping in mind that no job or career is perfect: Will you enjoy doing what others in the company are doing day in and day out?

As you try to reach your goals, enjoy the journey. It will probably shift and change and maybe you will find something completely different than what you set out to do. Be open to change and have a growth mindset.

Finding flow

So how does one find a perfect career? Daniel Pink, speaker of a famous TED talk on career and author of many books, talks about how we work best when we have autonomy and mastery. In his recent LinkedIn course on motivation, he puts our careers in four quadrants as shown in Figure 3.

Figure 3

Pink says that we perform our best when we have control of our jobs and find it challenging at the same time. We are in the top right quadrant and are in FLOW. If our jobs are challenging but we don’t have a lot of control we might hover into BURNOUT. If we have control but the job is not challenging, you might hover into DRIFT. The worst is where you neither have control nor challenge, which leads to MONOTONY. It’s hard to be in flow mode all the time and there will be aspects of your work that will hover around all the quadrants but one should try to work with your manager to carve a position where you have the most flow.

We can expand further into the topic of FLOW through psychologist Mihaly Csikszentmihalyi’s work, which focuses on how we can use challenge and skill to achieve flow. If you have a challenging project in which you are highly skilled, you will be in a state of flow. When we are over ambitious, the work might be challenging but we might not be as skilled in an area, which leads to ANXIETY. Explore Figure 4 below to see where you stand with your current projects and work.

Figure 4

Finding work you want to do

Then comes the question of what work do you want to do? Bill Burnett and Dave Evans in their book Designing Your Work Life explain how everyone works for Money, Impact and Expression at different levels. Your focus on each of these three levers will vary by who you are and the point in time in your life. One has to figure out what combination works best for them.

Let’s look at a few examples in Figure 5 of how the three levers are calibrated for different job profiles of an investment banker, a non-profit employee, and an artist. If a non-profit employee is dissatisfied because they are not getting paid like a newly minted investment banker, then their lever and perhaps career is not calibrated properly: one works at a non-profit mostly for its impact and not for the money. Similarly, as an artist your sense of expression is probably very high; money and impact not so much. If you have the levers calibrated properly and are conscious about it, you can substantially reduce your source of unhappiness at work.

Figure 5

Multiple ways to measure impact

Money and expression are easy to measure but what about impact? In my opinion every job has some sort of impact but how can it be measured? I suggest looking at it from three angles.

1. Is the job internally or externally focused? Within a large organization some departments are internally focused, like IT, HR, and Accounting. Some are a combination of internal and external like marketing and communications, and some are externally focused like the Corporate Innovation Arm or Business Development. Figure out what kind of roles you prosper in. You can create an additional lever for how internally and externally focused the job is like shown in Figure 6.

Figure 6

2. Is the unit where you’ll be working the cost or revenue making center? For example, when you are an accounting specialist working for one of the top auditing firms like KPMG or PWC you are generating revenue for the company; however, if you switch to a corporation like Microsoft whose bread and butter is not accounting, you fall out of the cost making center. Figure out from the level below where you stand in the spectrum.

Figure 7

3. Burnett and Evans talk about how your point of impact can be far (global) or near (personal), and the type of impact of your work can be focused on what is new or what needs to be renewed (repaired).

Let’s look at a few examples as shown in Figure 8 below:

Figure 8

• As a startup founder of an innovative healthcare company that helps with curing cancer, you fall in the quadrant of creating something new, and your point of impact is global as the startup may impact thousands of lives if successful.

• As a Product Manager working on a new feature in WhatsApp, you can affect billions of lives with a single tweak of a feature. You fall in the category of New/Far.

• As a professor at a university if you are creating a new digital marketing course, you are impacting probably 50 students at a time, so you fall in the category of New/Near.

• As a general physician you are not creating anything new and working with one patient at a time, so you might fall in the category of Renew/Near.

• A program manager at the Work Bank’s headquarters in Washington, DC who is working on an education project in Bangladesh. You are managing the program from afar but there are lots of resources working with you on the ground in Bangladesh to implement the project. Your role is at a high level making sure the millions of dollars being disbursed are being used properly, which falls into renew. Here, you fall in the category of Renew/Far.

Every role can be mapped in these 4 quadrants and your job can traverse many of these quadrants at a time. Try to figure out what kind of impact works for you.

When you choose a career path, remember you are saying no to many other things. In addition, what might work today, might not work for you in the future. Be judicious and conscious about your career path taking into consideration your finances and personal circumstances. I would recommend trying out many things when you are younger to see what works best for you. Small conscious changes will ultimately lead to your goal.

In Summary:

1. Figure out if you work best as a rockstar (specialist) or a superstar (generalist).

2. Use tools like the 10-10-10 framework and Glassdoor and talk to people in the field to think long term and find roles that fit.

3. Aim for a job that has challenge and control, i.e. that puts you in flow state.

4. Calibrate your money, impact, and expression levers with the type of field you choose.

5. Determine how you want to measure your impact through the angles of internal/external focus, revenue generation, and point and type of impact (global/near and new/renew).

The beauty of career mapping in the modern age is that there is no one right answer. None of the quadrants or levels are necessarily good or bad. Figure out what works best for you – it’s your career and no one else’s.

In closing, let’s remind ourselves that career is only one dimension of our multi-dimensional lives. Health, family, friends, hobbies…there are so many things to live a fulfilled life. It’s your individual journey and best to not be bothered by what others are doing and to compare yourself. We need to define what success means to each of us personally. Happy career mapping!

(Originally published on LinkedIn, March 12, 2020.)

We have all heard of India’s rapid development in the past few decades. Its GDP per capita, PPP (in current international $) has grown from $1,237 in 1990 to $7,763 in 2018. I left India in 2002, and every year during trips back home to Delhi and Kolkata I would witness the transformation. While the development in my new home, the United States, was more or less static, the changes in India were constant and “in your face” – new malls, new consumer brands, massive increases in car and mobile phone ownership, to name a few. There was a sense of almost endless growth.

As India has developed, the opportunities for tech start-ups have grown significantly. As of last year, the number of tech start-ups in India clocked in around 9,000. According to Goldman Sachs, India is home to 560 million internet subscribers, 354 million smartphone subscribers, and the cheapest mobile data rates in the world (1 GB for $ 0.26). By 2025, 60 percent of Indians are projected to have internet subscriptions and smartphones.

Subscribe now

How will technology advancement impact India in the coming decades?

To capture the full potential of technology, India’s national and sub-national governments will need to bridge the urban-rural digital divide by addressing barriers such as limited telecom infrastructure, language, slow internet speed, and low computer literacy. Nevertheless, the impact of technology will be undeniable.

While most of India will be connected in the coming decades, India is diverse beyond our imaginations. Start-ups should be careful of one size fits all approaches, as Indians fall in very different income segments than the West.

How do India’s income segments compare with other major players?

Hans Rosling’s work divides the world into four income levels based on household per capita income:

The image below gives an idea of how people in the different parts use transportation, cook, eat and sleep. People in Level 1 live in extreme poverty whereas people on Level 4 are the richest. Most of you reading this article are probably in Level 4.

To give us some perspective, let’s compare India’s population and earnings in 2019 with China, which has a developing story like India, and the United States, the most developed part of the world.

Most of India is still on Level 2 (earning between $730/year to $2,920/year), whereas the majority of China is on Level 3 (earning between $2,920/year to $11,680/year) and the majority of United States is rich on Level 4 (earning $11,680/year and above). Above $11,680/year might not be considered rich to many reading this article, but it gives you perspective on where you stand with respect to the rest of the world.

The population on Level 2, where most Indians fall, are the rural laborers, the rural landowners, the urban blue collar and migrant worker segments. Whereas, the Indians on Level 4 are the rich class of around 2.6 million households, a tiny percentage of didddIndia’s total population. These folks have white collar jobs, are small and medium size business owners or are corporate big shots. Most Indians in Level 4 are concentrated in the $11,680 to $30,000 annual income range.

How do India’s income segments impact tech start-ups?

Most tech start-ups in India have targeted their customer base to the Level 4 consumers and the richer Level 3 consumers (earning above $5,500/year). These two groups represent a population of close to 45 million. This demographic is probably the one watching Netflix, using ride hailing services, shopping online, or ordering food through delivery apps. While these customers are more likely to have disposable income, they represent just a small slice of India. Once this market is capped, start-up founders in India are likely to find it difficult to expand into the “bottom of the pyramid” markets, which by their sheer volume merit greater consideration.

Where should start-ups and VCs focus in the coming years?

As the graph above indicates, it will take another 20 years for most Indians to move towards Level 3. For comparison, most Chinese will remain on Level 3, with a significant portion moving to Level 4. In the United States, nearly 100% of the population will be on Level 4.

Given that start-ups do not have 20 years to wait, or they risk going under, VCs and tech start-ups should deepen their focus today on Level 2 consumer and the lower end of Level 3 consumers in addition to their focus on India’s richer segments. However, Level 2 and 3 consumers’ dreams, aspirations and incomes are very different from the consumers on Level 4. Only a few start-ups in the last decade have been able to reach a wide segment of society —Whatsapp, Google and Youtube are examples.

Start-up founders should focus on the right businesses and consumer segments and not create copycat models from the United States or other Western countries, which are too rich to serve as effective models. Indian start-up founders will need to be creative about how to expand their customer-base. One source of inspiration may be to look at how Chinese start-ups address Level 2 and 3 consumers currently and how this could be adapted to the Indian context.

The growth potential in India in the coming decades is huge. While India is a unique diverse market, with its inherent challenges, if tapped properly, the upside is tremendous.

The question has plagued many, especially the millennial generation. When we think of startups, we may conjure up images of successful tech startups like Facebook or Apple that propelled their way to mega company status. In reality, the road to a successful exit is often bumpy and not for the faint of heart.

Subscribe now

To know whether a large corporation or a start up is the better fit for you, I like to use the analogy of steering a big ship as opposed to a small canoe. Big ships are more complex and require more expertise to steer but when the waters are choppy, they can weather the storm. Small boats on the other hand are easier to maneuver but during a storm they are more likely to go down under.

Large corporations are like big ships. Change is harder, takes time and requires many approvals and people to steer the ship. Startups on the other hand are like small canoes. Making change within the startup is easier, takes less time and less people to execute. Big ships can travel long distances without need for supplies; startups have a limited runway to execute and need constant cash injection at every point.

Then comes the question of which water do you want to ride on. Some waters are choppy to begin with and the risks of going under are higher. The choppy waters are analogous to the support and ecosystem for building a company in a particular context or country. Large corporations (big ships) can most of the time weather the choppy waters as they are not dependent on external funding. Not so much for startups.

To illustrate this point, the graph below shows the number of startups which have received funding at different levels since 2001. (Note: the data does not cover micro-businesses, information for which is not publicly available, which make up a large percentage of businesses, especially in emerging markets; however, the data represented here is for scalable businesses.) The USA is by far the most entrepreneurial friendly country. The chances of getting funded and having a successful exit is way higher in America than other countries. The water there is relatively calm. That’s not to say that you should not open up startups in other countries, just know that the water is choppy, as the risks are higher there.

Adding to the analogy further, the Angel Investors, VC funds, private equity, large corporations and governments are like the kings and emperors of the olden times that funded explorers like Vasco da Gama and Christopher Columbus to set sail on the sea; whereas the explorers are the startup founders in today’s time. They are high risk takers. They will challenge the status quo to invent new business models and start entire new industries.

Explorers relish the challenge of creating something new, while their counterparts, prefer the safety of big corporations. To be clear, change also happens in big corporations, but it takes a long time for it to be put to practice.

So what kind of organization do you want to work in? Do you get seasick in choppy waters? Do you prefer the safety of big ships? Or are you ready to explore what awaits you in your nimble canoe?

It all depends on what kind of mentality and risk taker you are. Happy sailing!

Subscribe now