The Product Growth Guy 🚀 | Helping you land your next job + succeed in your career | Newsletter Writer and Podcast Host

TL;DR: I write a newsletter for product managers, product leaders, and aspiring product managers. I focus on actionable and deep tips to improve your career.

The Longer Version:

After a long career in PM - rising to VP of Product at a Unicorn - I have made the shift to writing about it.

In particular, I write about product growth. That is, the art of growing via your product. 

This means I also talk a lot about product-led growth (PLG), the SaaS B2B motion.

My newsletter doesn't just focus on growth though. 

I also break down technology, product management, product leadership, and how to get a PM job. 

It's all the things that make for a great use of your learning & development budget. 

Or even personal spend. 

The math is pretty great. 
↳ $150 for an annual subscription
↳ $211K average PM salary in the US
↳ 0-10% average range of raises offered

If the newsletter helps you get a 0.07% better raise, it pays itself. 
That's 1/142th of your raise range.

It's a pretty smart investment for every level of PM.

http://www.news.aakashg.com

Aakash Gupta

Claude Code is the most powerful AI today.

(Especially with yesterday's Opus-4.6 release)

Here's why it reached $1B ARR in 6 months and how to master it:

1. It changes everything

Full system access, agent swarms, hours-long tasks. Chat Claude is browser-only. Cowork is a limited skin.

Guide to Claude: https://lnkd.in/eZ6UBbAQ 
Guide to Cowork: https://lnkd.in/g6MhB3YJ

2. What Claude Code can do

Process 10,000 files in minutes. Run bash without learning syntax. One interface for 200+ tools. Delegate like you have 5 junior employees.

Basic Guide: https://lnkd.in/eUyPEAma 
Advanced Guide: https://lnkd.in/gvA6rZtX

3. The Claude Code workflow

Analyze → Plan → Create → Scale

Meeting transcript → PRD → Jira tickets → Slack summary → dashboard. One tool.

AI PRD Guide: https://lnkd.in/eMu59p_z 
AI Roadmap Guide: https://lnkd.in/eEuNcj-m

4. Getting started

Use Cursor. Open terminal, type "claude", press Enter.

macOS 13+, Ubuntu 20+, or Windows 10+. Pro ($20/mo), Max ($100-200/mo), or API.

Pro Guide: https://lnkd.in/eadQ5fN6 
Official Setup: https://lnkd.in/gPthuahX

5. Connect everything (MCP)

USB-C for AI. One protocol, hundreds of connections. GitHub, Sentry, Linear, Notion, Slack, Jira, PostgreSQL, Gmail, Typefully, Buffer, Discord.

PM Operating System: https://lnkd.in/gsH9RSfJ
Anthropic MCP Servers: https://lnkd.in/g-sx7vXP 
MCP Collection: https://mcp.so

6. Essential commands

/help → all commands 
/clear → reset context 
/compact → save tokens 
/model → switch models 
/mcp → check connections

Use @ to reference files (@filename, @folder/).

Prompt Engineering: https://lnkd.in/d_qYCBT7 
Best Practices: https://lnkd.in/gNSNWmKB

7. Skills & CLAUDE.md

Skills are reusable automations Claude auto-loads. Built-in for docx, xlsx, pptx, pdf.

CLAUDE.md is project memory. Commands, style guides, context that persists.

Skills Tutorial: https://lnkd.in/eRidpGSR 
Steal 6 of My Skills: https://lnkd.in/eZyJzMnH 
Context Engineering: https://lnkd.in/eUUPMmJK

8. Prompting techniques

Pro pattern: Checkpoint + Iterate.

Plan first. Review before execution. Create checkpoints. Rewind (Esc twice) if it breaks. Iterate with specific feedback.

Prompt Engineering: https://lnkd.in/d_qYCBT7 
ChatGPT for PMs: https://lnkd.in/eyfUi8ZN

9. Resources

All links above
Podcast: https://lnkd.in/eyt7agKj
MCP Reference: https://lnkd.in/gPUWGa7S

🔖 Save this. Repost so your network sees it too.
➕ Follow Aakash Gupta for daily AI tips.

Claude code is the future. 

This is your guide.

Helping you succeed in your career + land your next job

Anthropic PMs don't write PRDs.

They use Claude Code to build the first version of the feature themselves. The whole company dogfoods it for weeks. Then they ship an experiment.

Anthropic went from $1B run-rate revenue at the start of 2025 to over $9B by year end. Claude Code alone generates over $500M in run-rate revenue with usage growing 10x in three months. The company signed a term sheet at $350B, nearly doubling from $183B just months earlier.

This is what happens when you eliminate handoff drift. In a traditional org, a PM writes a spec. Design interprets the spec. Engineering interprets the design. Each handoff introduces drift. Each handoff adds a week. The 2-4 week cycle of PRD → design → ticket → sprint → build → QA compresses to days when the PM can prototype in the codebase.

The prerequisite is making your codebase AI-friendly. Tailwind instead of scattered CSS files. Microservices architecture. Reduced codebase size. That infrastructure investment is what lets PMs vibe code directly against production components.

The math is simple. If your experimentation cycle is 8 weeks per test and your competitor's cycle is 3 days, they're running 13x more experiments per quarter. Compound that over a year and the learning gap becomes insurmountable.

For more on this way of working:

1. My Berkeley keynote: https://lnkd.in/g6tNMu2j
2. Claude code basics: https://lnkd.in/gij3jSvH
3. Claude code advanced: https://lnkd.in/gr9sPtJ7
4. Claude code OS: https://lnkd.in/gsH9RSfJ
5. AI PRDs: https://lnkd.in/eMu59p_z
6. AI prototyping: https://lnkd.in/eJujDhBV

There’s places where this doesn’t work. But they’re less than most people realize. And the benefits are tough to argue.

The PM skill that matters in 2026 is taste at speed. Boris Cherny just showed everyone what that looks like.

His Claude Code team at Anthropic doesn’t write PRDs. They build hundreds of working prototypes before shipping a single feature. Boris personally ships 20-30 PRs a day running 5 parallel Claude instances. They built Cowork, a full product for non-engineers, in about 10 days.

Everyone is debating whether PRDs should die. They shouldn’t in big companies and established products, where understanding the hypothesis and success metrics is critical. But the real question is: what happens to the PM who can’t evaluate 15 prototypes and pick the 3 worth shipping?

Because here’s what changes when building costs near zero: the bottleneck moves from “can we build it” to “should we ship it.” PRDs existed because building was expensive and you needed sign-off before committing resources. When a prototype takes 45 minutes instead of 6 weeks, nobody needs a document to authorize exploration. They need someone who can look at working software and say “this one, not that one” in real time. If anything, the PRD now comes after the prototype.

On the Claude Code team, PMs code. Data scientists code. User researchers code. In fact, now everyone has the same title (“Member of Technical Staff”), and it’s by design. Boris said productivity per engineer grew 70% even as Anthropic tripled in headcount. The coordination cost of translating specs into code disappears when everyone can build. And that changes what a PM is actually good for.

Boris said it himself: “There’s just no way we could have shipped this if we started with static mocks and Figma or if we started with a PRD.” The old process would have spent more calendar time documenting Cowork than his team spent building it.

This is the Claude Code team today. It will be most fast-moving teams within 18 months. The PMs who thrive will be the ones reviewing prototypes at 9am, killing 80% of them by noon, and shipping the survivors by end of week. Pattern matching across user research, technical feasibility, and business model simultaneously while staring at working software.

So here's how to get there:

1. Understand Modern PRDs: https://lnkd.in/eMu59p_z
2. Master Claude Cowork: https://lnkd.in/gwhgmSfw
3. Learn Claude Code: https://lnkd.in/eUyPEAma
4. Build your OS: https://lnkd.in/gsH9RSfJ
5. Watch Boris interview: https://lnkd.in/g8fXRPRR

The PMs who struggle will be the ones still writing 15-page specs for features that could be prototyped, tested, and validated before the doc hits its first review cycle. Taste at speed is the new moat.

No alternative text description for this image

Look at job postings from OpenAI, Anthropic, and Google DeepMind. They want PMs who can:

- Write and run evals
- Prototype with code
- Understand model architecture tradeoffs
- Automate their work with AI agents
- Use LLMs with MCPs

The title change is symbolic, but the signal is real. The line between PM and engineer is dissolving in AI-first teams.

This doesn't mean PMs are dead. It means the PMs who survive will be the ones who can do the technical work themselves.

Here’s how to master each:

- Evals: https://lnkd.in/eGbzWMxf
- Prototype: https://lnkd.in/eJujDhBV
- Model information: https://lnkd.in/eKhmKYNp
- Practical AI Agents: https://lnkd.in/eyamsW7a
- LLM with MCP: https://lnkd.in/gsH9RSfJ

The PM’s who can’t? They’ll soon have to, or be replaced by those who can.

The math on this project should mass-humble every AI lab on the planet.

1 cubic millimeter. One-millionth of a human brain. Harvard and Google spent 10 years mapping it. The imaging alone took 326 days. They sliced the tissue into 5,000 wafers each 30 nanometers thick, ran them through a $6 million electron microscope, then needed Google’s ML models to stitch the 3D reconstruction because no human team could process the output.

The result: 57,000 cells, 150 million synapses, 230 millimeters of blood vessels, compressed into 1.4 petabytes of raw data. For context, 1.4 petabytes is roughly 1.4 million gigabytes. From a speck smaller than a grain of rice.

Now scale that. The full human brain is one million times larger. Mapping the whole thing at this resolution would produce approximately 1.4 zettabytes of data. That’s roughly equal to all the data generated on Earth in a single year. The storage alone would cost an estimated $50 billion and require a 140-acre data center, which would make it the largest on the planet.

And they found things textbooks don’t contain. One neuron had over 5,000 connection points. Some axons had coiled themselves into tight whorls for completely unknown reasons. Pairs of cell clusters grew in mirror images of each other. Jeff Lichtman, the Harvard lead, said there’s “a chasm between what we already know and what we need to know.”

This is why the next step isn’t a human brain. It’s a mouse hippocampus, 10 cubic millimeters, over the next five years. Because even a mouse brain is 1,000x larger than what they just mapped, and the full mouse connectome is the proof of concept before anyone attempts the human one.

We’re building AI systems that loosely mimic neural networks while still unable to fully read the wiring diagram of a single cubic millimeter of the thing we’re trying to imitate. The original is 1.4 petabytes per millionth of its volume. Every AI model on Earth fits in a fraction of that.

The brain runs on 20 watts and fits in your skull. The data center required to merely describe one-millionth of it would span 140 acres.

If you want to stay up on AI, do not miss my AI newsletter: www.aibyaakash.com

I don't think most PMs realize how fast the "AI-native PM" bar is rising.

Six months ago, knowing how to write a good prompt was enough. Now the PMs getting hired at top AI companies have full tooling stacks: custom GPTs for PRD drafts, Claude Projects loaded with their company's design principles, Gemini Gems configured for competitive analysis, Claude Code OS with context.

Lisa Huang created Gems at Google. She showed her complete setup on the podcast and the pattern is clear. The best PMs aren't prompting AI. They're configuring AI. Once.

The setup cost is 2 hours. The ongoing return is 5+ hours per week. The compounding return is that your AI gets better at your job the more you use it.

Every PM interview I've seen in 2026 asks some version of "how do you use AI in your workflow." The answer they want isn't "I use ChatGPT." They want to hear about your system.

Here's how to ace those questions:

1. Watch Lisa's ep on Gems: https://lnkd.in/gbTJTt_3
2. Learn Claude Projects: https://lnkd.in/eZ6UBbAQ
3. Grab my OS: https://lnkd.in/gsH9RSfJ
4. Grab my prompt library: https://lnkd.in/eHU88esH
5. Practice your interviews: https://lnkd.in/eZZ_vvQc

Build the system.

The creator of Claude Code, which hit $1B ARR faster than almost any dev tool in history, just told you the endgame: the title “software engineer” gets replaced by “builder” or “product manager.”

This is Anthropic’s head of Claude Code saying his own product eliminates the skill that defined his entire career. Cherny rose to Principal Engineer (IC8) at Meta over seven years. He wrote and reviewed code for a living. Now he runs 10-15 concurrent Claude sessions and treats AI like compute you schedule, not a tool you prompt.

In a podcast with YC, Cherny posted that engineers are “more important than ever.” When someone pointed out the contradiction with Dario’s Davos prediction that models will do all of SWE end-to-end in 6-12 months, Cherny replied: “Dario is talking about what’s next.”

This tells you everything about Anthropic’s internal positioning. They’ve gone from $1B to $7B revenue run rate in nine months. They have 50+ engineering roles open right now. And the guy building the tool that automates coding is simultaneously hiring more engineers than ever.

The contradiction dissolves when you realize what those engineers actually do. Cherny uses a “verify-app” agent to run end-to-end tests, a “code-simplifier” agent for architecture cleanup, and a PostToolUse hook that auto-formats the last 10% Claude misses. He’s building systems that make AI reliable, not writing application code.

Anthropic is betting that the value shifts from people who can write code to people who can define what to build, sequence the work across parallel agents, and catch the 10% of failures that compound into production disasters. That’s the “builder” Cherny is describing.

5 resource to move to this way of working:

1. New tool stack: https://lnkd.in/gtJWybA5
2. Claude code for PMs: https://lnkd.in/eUyPEAma
3. Advanced claude code: https://lnkd.in/eadQ5fN6
4. Claude code PM OS: https://lnkd.in/gsH9RSfJ
5. Boris' interview: https://lnkd.in/ge2JGgCd

Every PM reading this should understand: the person who built the most successful AI coding tool in history just said your job title is the future of software. The question is whether you’re building the judgment and technical fluency to claim it.

The creator of Claude Code just mass-hired PMs.
Well, not literally. But close.

Boris Cherny built Claude Code at Anthropic. He runs 5 AI agents in parallel from his terminal. His workflow thread hit 7.7M views. He ships like a 10-person engineering team.

And yesterday he posted this:

“Someone has to prompt the Claudes, talk to customers, coordinate with other teams, decide what to build.”

Read that list again. That’s a PM job description, word for word.

Here’s what most people miss about AI eating software engineering: the bottleneck doesn’t disappear. It moves upstream.

When code generation takes hours instead of weeks, building the wrong feature doesn’t save you three weeks anymore. It costs you three hours of wasted compute and the opportunity cost of what you should have shipped instead.

The faster your engineering velocity, the more expensive bad prioritization becomes.

Anthropic went from $1B to $7B in annual revenue run rate in 9 months. Small teams amplified by their own AI tools. The constraint was never lines of code per day. It was product decisions per week.

Every PM who learns to prompt, prototype, and ship with AI tools just became 10x more valuable. Every PM who thought “I’m safe because I don’t code” just became 10x more replaceable.

Boris said the quiet part out loud. The tool builders know exactly where the human bottleneck is moving.

And it’s moving straight into product judgment.
Want to get ahead of this shift? Start here:

∙AI PM roadmap: https://lnkd.in/e3akMRm8
∙Prompt engineering for PMs: https://lnkd.in/eyfUi8ZN
∙AI prototyping: https://lnkd.in/eDVWcaPz
∙AI PRDs: https://lnkd.in/eMu59p_z
∙ Complete AI PM course: https://lnkd.in/eqFzK6tr

PM isn’t dead. The job description just changed completely.

If I were to learn PM again, I would start here. 

15 steps to a PM job paying $200K+:

1. Understand the PM role

Start with how to break in: https://lnkd.in/erSRWwUF
Watch Marty Cagan: https://lnkd.in/eb5hbA28

2. Learn PM fundamentals

Master PRDs: https://lnkd.in/ef5Yde54
Watch Dan Olsen: https://lnkd.in/gWgUjVpB

3. Master product strategy

It's the key PM skill: https://lnkd.in/e8EUQHAW
Hear from Roger Martin:  https://lnkd.in/ecVF-y6i

4. Learn product discovery

Master continuous discovery: https://lnkd.in/eYsTAnrq
Hear Teresa Torres: https://lnkd.in/gdVqf2bV

5. Understand growth

Master growth models: https://lnkd.in/dEfhHTMP
Hear from Brian Balfour: https://lnkd.in/eef_tipn

6. Learn analytics & experimentation

Master metrics: https://lnkd.in/ea8sWSsS
Watch Amplitude's CEO: https://lnkd.in/esuAeUUK

7. Study company growth case studies

Start with Cursor: https://lnkd.in/eq8xvCAK
Then read Linear: https://lnkd.in/e_R9ruDG

8. Build your PM Portfolio

Start with basics: https://lnkd.in/eQd32YUA
Then vibe code: https://lnkd.in/eT3RvQiM

9. Optimize Resume & LinkedIn

Learn to customize: https://lnkd.in/eyMAjsNc
Make LinkedIn your billboard: https://lnkd.in/egeXb7Sc

10. Master PM interviews

Learn each interview: https://lnkd.in/e8WkHDbK
Hear from Lewis Lin: https://lnkd.in/e4MkWiyD

11. Execute your job search

Learn how to get interviews: https://lnkd.in/eqn5Eqq6
Hear from Phyl Terry: https://lnkd.in/eVpXWuRN

12. Target companies

Build a company list: https://lnkd.in/eBCcuHy7

13. Nail team matching

Learn the process: https://lnkd.in/eG_YqMEn

14. Negotiate

Master negotiation: https://lnkd.in/e8ZJmvMe
Hear from Satish Mummareddy: https://lnkd.in/epwZ3Dfw

15. Grow your career

Learn product leadership: https://lnkd.in/ewTn7W6z
Hear the PM to VP journey: https://lnkd.in/g5wp7snd

🔗 All links: https://lnkd.in/g59eb3Zy

Want me to DM them to you? Comment 'PM roadmap' + DM me and I'll reply (repost to cut the line).

➕ Follow Aakash Gupta for daily PM content.

Replit's gross margins went from 36% to negative 14% in two months.

Same product. Same pricing. Same team. The only thing that changed: they launched a more autonomous AI agent that consumed more LLM resources than their pricing covered.

Traditional SaaS has 70-80% gross margins because one more subscriber costs almost nothing. AI products pay for compute on every prompt. Your best users are your most expensive users. That single fact breaks every pricing model designed for the SaaS era.

I mapped pricing across the top 50 AI startups by valuation with Moe Ali. Six patterns emerged.

The scariest finding: in most AI products, the P90 user costs 10-40x more than the P50 user. Both pay the same subscription. You're subsidizing your heaviest users with revenue from your lightest ones. And that subsidy grows as power users discover more ways to use the product.

Cursor learned this the hard way. They switched from flat 500 requests/month to a credit pool system. A developer burned the entire monthly allocation in a single day. $7,225 invoice. The CEO published a public apology on July 4th. The plan description quietly changed from "Unlimited" to "Extended" twelve days after launch.

Anthropic took a different approach. Their $17/$100/$200 tiers map to genuinely different user personas. A casual user, a power user, and a developer replacing an IDE. Those are different products with different willingness to pay. Then weekly rate limits targeting less than 5% of subscribers to push the heaviest users toward the API, where per-token pricing covers actual compute.

The pattern across all 50 companies: pure flat pricing is dying. Nearly half use two or three models simultaneously.

Here's the full breakdown:

1. Complete AI pricing guide: https://lnkd.in/gdKaQSMk
2. Replit guide: https://lnkd.in/gmA_c_AG
3. AI product strategy: https://lnkd.in/egemMhMF
4. AI agents guide for PMs: https://lnkd.in/eeey5Cxr

If you can't estimate your cost distribution across P10 to P90, you're not ready to set a price.

Only 24% of PM candidates have a GitHub.
 
Every PM I've placed at OpenAI, Anthropic, and Meta AI in the last year had one.
 
They didn't have 78K stars. They had three things: one project that solved a real problem, documentation that showed product thinking, and consistent activity.
 
Shubham Saboo went from Dev Rel to Senior AI PM at Google because his GitHub marketed him. Google reached out to him. When companies come to you, the interview-to-offer rate jumps from 22% to 37%.
 
But most PM GitHubs I review are anti-signals. Copy-pasted AI code with no tradeoffs section. Joke projects that prove nothing. Green square farms with 50 empty commits.
 
I put together the web's first guide on PM GitHubs:
 
• The 6 elements that made Shubham's GitHub get him hired
• Step-by-step setup with Cursor and Claude Code
• A starter kit repo you can fork today
• 50 project ideas mapped to PM skills
• The 6 mistakes I see candidates make
 
Full guide for paid subscribers:

Open article: This Github Got a PM Hired at Google by news.aakashg.com

This is tool stack of the best AI PMs I've talked to:

(You just need one in each category)

𝗕𝗨𝗜𝗟𝗗𝗜𝗡𝗚

Vibe Coding

• Cursor, Claude Code, Windsurf, Replit, or Warp
• You want to be able to build your own operating system
• My guide: https://lnkd.in/gsH9RSfJ

Prototyping
 
• Lovable, Bolt, v0, Magic Patterns, or Base44
• A prototype is a key pair to the PRD these days
• My guide: https://lnkd.in/eJujDhBV

𝗣𝗥𝗢𝗗𝗨𝗖𝗧𝗜𝗩𝗜𝗧𝗬
 
Dictation
 
• Wispr Flow, superwhisper, Tactiq, Speechify
• You talk 2x faster than you type. Use that
• My guide: https://lnkd.in/g3bRR7e6
 
Meetings
 
• Granola, Fathom, Otter.ai, tl;dv, or Fireflies
• Never take manual meeting notes again
• My guide: https://lnkd.in/gWk_t49C
 
General LLMs
 
• Claude, ChatGPT, Gemini, Kimi, or Grok
• Find your favorites and master prompt engineering
• My guide: https://lnkd.in/d_qYCBT7
 
𝗔𝗨𝗧𝗢𝗠𝗔𝗧𝗜𝗢𝗡
 
Simple Agents
 
• Zapier, Lindy, Relay, Bardeen, or Parabola
• Start here if you've never automated a workflow
• My guide: https://lnkd.in/eyamsW7a
 
Full-Featured Agents
 
• n8n, make, Activepieces, Workato, or Tray
• Graduate here when simple feels limiting
• My guide: https://lnkd.in/eeey5Cxr
 
𝗗𝗜𝗦𝗖𝗢𝗩𝗘𝗥𝗬
 
User Research
 
• NotebookLM, Perplexity, Elicit, Consensus, or Grain
• Use AI at every step to make it faster
• My guide: https://lnkd.in/gcD8_wNS
 
Customer Intelligence
 
• Dovetail, Unwrap, Enterpret, Monterey, or Viable
• Go from unstructured data to usable inputs
• My guide: https://lnkd.in/evCwj4rU

Want a pro version of each? 
↳ Grab my bundle: https://lnkd.in/eYxxVhZr

What would you add to the list?

The psychology of why people fail with Claude Code is the same psychology behind why people buy gym memberships in January and quit by February.

The tool works. The person didn't build the habit infrastructure around it.

Claude Code is the most powerful professional software I've used. Cowork is right behind it. But both have the same adoption problem. The first experience determines everything. And the first experience with a blank setup is always mediocre because Claude is guessing about every variable that matters. Your role. Your product. Your metrics. Your writing preferences. What "good" looks like at your company.

The fix is a one-time setup that takes about 60 seconds.

A memory prompt that gives Claude your professional context. A CLAUDE.md file with your role, north star metrics, PRD structure, and writing rules that loads automatically every session. Skills that encode your exact workflows so quality stays consistent across dozens of uses.

Then you add the compounding loop. Every time Claude makes a mistake, you add a rule to the CLAUDE.md. Claude proposes the edit, you approve it, and next session it's already loaded. Your setup gets smarter with every use. Most people's stays static.

Cowork extends this into automation territory. Scheduled competitive intel reports. Meeting transcripts turned into action items in 30 seconds. Browser-based research using your actual logged-in Chrome session. Spreadsheet creation from messy files.

And one habit that separates people who stick with it from people who don't: /clear between unrelated tasks. Context bleed is the number one failure mode. You write a PRD then ask for a status update in the same session. The update comes back written like a PRD.

Here’s everything you need for this:

1. Claude code quickstart: https://lnkd.in/gRsTZ6wH
2. Claude cowork guide: https://lnkd.in/gwhgmSfw
3. My PM OS: https://lnkd.in/gsH9RSfJ
4. 90 min video: https://lnkd.in/eUyPEAma

This is all the infrastructure you need to build the habit. And not quit the gym by February.

Your PMs should be building in the codebase.
 
Not full features. But prototypes on your actual design system. Polish on simple front-end changes. Sometimes the initial version of a feature. If your product team doesn't have access to these three tools, you're already behind.
 
1. Claude Code
 
This is the most powerful way to use an LLM. It connects your PM directly into the codebase. Not a separate prototype environment. Your actual elements, your actual design system. PMs building on top of production components, doing polish on simple front-end work, sometimes shipping the initial version themselves. Anthropic hit $14B in annualized revenue in February 2026, up from $1B fifteen months ago. Claude Code alone went from zero to $2.5B ARR in nine months. That adoption curve tells you how much value teams are extracting. If your PMs aren't connected to the codebase, they're living in the past.
 
2. Claude Cowork
 
Once you connect Cowork with MCP and API to your data sources (analytics, user research, customer support, CRM) and your output systems (Linear, Jira, your engineering team's coding agent), it becomes the best harness for creating strategy documents, PRDs, and roadmaps. Nothing else compares right now. It works in Google Slides, Keynote, PowerPoint. It builds Excels and impact sizing models. When Anthropic launched Cowork plugins, Thomson Reuters dropped 15.8% in a single day. LegalZoom fell 19.7%. The market is telling you what this tool means for knowledge work. Your team shouldn't be using ChatGPT or Gemini chat. They should have Cowork.
 
3. MCP or API into all key tools
 
People give their teams Code or Cowork access and then don't connect them to systems because of IT constraints, privacy, regulatory, compliance. It totally handicaps the team. If they're not plugged into analytics, business intelligence, actual revenue metrics and not just engagement data, they can't make strategic decisions. The best product leaders are working through the security hurdles and getting their teams connected to everything.
 
Here's your guides:
 
1. Claude Code: https://lnkd.in/gCevZspr
2. Claude Cowork: https://lnkd.in/gwhgmSfw
3. MCP + PM OS: https://lnkd.in/gsH9RSfJ
 
PMs already want to use these tools. The bottleneck is leadership. Work through the IT and security hurdles. The teams that do will be winning in the market.

75% of AI PM interviews are behavioral.

Here's the full breakdown (save this):

I help 100s pursuing AI PM jobs, so I see the inside. And here's the reality of what questions you're going to hear over the course of an AI PM job search.

1. CASE INTERVIEWS (15%)

These get all the attention, but they're mainly in use at the top companies with lots ex-big tech folks like OpenAI, as well as big tech itself.

a. Product Sense (5%)

The AI twist on the classic product sense case is here, and many candidates I have mocked with struggle to adapt without significant practice.

Guide: https://lnkd.in/ekWty_bw

b. Product Design (5%)

The classic 'design a pencil for the blind' has evolved to AI variants like 'design an AI device to communicate with pets', bringing with it its own set of challenges.

Guide: https://lnkd.in/g3gg2pct

c. Success Metrics (5%)

The classic execution interview hasn't left, because this is one of those areas PMs are expected to have a pointy edge compared to their dev counterparts.

Guide: https://lnkd.in/gfffr_zC

2. BEHAVIORAL INTERVIEWS (75%)

a. Leadership & Drive (40%)

This is the entire genre of questions about handling stakeholders and past experiences, by far still the largest bucket of questions. 

Guide: https://lnkd.in/eGhgVNyh

b. Values & Culture (10%)

Every company is their own unique wallflower, and, mostly as startups, they strongly feel that way. Expect to show that you fit that company's philosophy.

Guide: https://lnkd.in/gS-CKH9P

c. AI-Specific Experience (25%)

This is the area every company wants to go deeper beyond the resume: "Did you actually do AI PM work in the past or are you making it up?"

Guide: https://lnkd.in/edKrAAFA

3. TECHNICAL INTERVIEWS (10%)

a. Technical Deep Dive (5%)

These are surprisingly rare. Expect the rare technical question if it's a technical team or role, though.

Guide: https://lnkd.in/dG9EFhiR

b. Strategy / XC Presence (5%)

The PM and designer role are merging. So they want to see how well you can handle it.

Guide: https://lnkd.in/dve8it6D

Be strategic with your practice. 

Want coaching? Apply to the cohort: https://www.landpmjob.com/

📌 Want my free framework to success metrics interviews? Comment 'success metrics framework' + DM me (must do both). I'll reply to those who repost faster.

The AI PM interview is like any test. Break it down into the parts and grade yourself on each. Then bring up your worst grades.

I built this entire deck without ever leaving Claude.

Step by step (swipe through):

Gamma just launched as a Claude Connector.

You can now brainstorm, generate a presentation, and keep refining it all in one conversation. No copy-paste. No rebuilding from scratch.

The combo plays are even crazier:

- Gmail + Gamma: emails become decks
- Notion + Gamma: docs become presentations
- GitHub+ Gamma: repos become project overviews

Your deck finally keeps up with your thinking.

Try it: https://lnkd.in/g7YNKrJY

How to Build Presentations with AI

Acquiring Inbound Leads for B2B Founders with Personal Content Programs • Explaining Content, one Ugly Drawing at a time

Content steals focus, I know. Writing, formatting, visuals, posting cadence, and keeping up with comments. It becomes a weekly tax, then it gets dropped, then inbound stays unpredictable.I build an Authority Engine for B2B Tech Founders. It turns your existing knowledge into consistent content that makes you look like the category leader, without you writing a word.What you get:1) 100% custom AI voice cloneWe extract your voice from your real materials. Whitepapers, sales calls, internal docs, Slack messages, past posts, and anything else that reflects how you think and speak. No generic templates.2) LinkedIn Autopilot5 high converting posts per week, ghostwritten, formatted, and ready to publish. You do not write, format, post, or search for visuals. You approve drafts with a simple yes or no.3) Search Dominance pack2 SEO optimized long form articles per month, derived from your LinkedIn content, so your authority compounds beyond the feed.First week of content can go live within 48 hours of onboarding. You can see engagement on day 1. Your only job is approvals.I will write 3 posts for you for free using your data, so you can see the voice match before you pay. If the first week does not sound exactly like you, or better, I rewrite it until it does, or you do not pay.Best for founders who have real insight and a real product, but do not want content creation to compete with building, selling, and hiring.If you want to test the voice clone, message me: “3 posts”.

Kuba Czubajewski

IMO, this is the only realistic "future" of content marketers:

Every week, I see here new tools popping up that promise my clients they can let all their content team go because AI can do it all for them.

Not once have I seen it actually work, btw.

Content is a much deeper thing than writing a bunch of paragraphs that sound good or an infographic that looks nice.

For me, it's all about connecting the dots that are usually VERY far apart and turning it into a fresh, impactful message.

This is why I'm in love how Dropbox Dash is approaching AI for creatives.

Instead of proposing yet another "I'll do it for you" tool, they're solving one major pain most of us feel:

Not being able to connect the dots because the dot you're looking for is in a file you forgot the name of.

Sounds trivial, but OH MY GOD, the amount of time I used to spend on scrolling through my G-drive is embarrassing.

And this is where I think the whole AI for content conversation should land.

AI is incredible at recall.

Humans are still the best at judgment what matters, what’s true, what’s new, and what’s worth saying out loud.

You won't win because you can generate 50 posts a day.

But if you can pull the right inputs fast, then make a call on the angle, and ship a message people actually remember, you will win.

PS. this is not sponsored by anyone, I just saw their launch and loved it.

I’ll write 3 LinkedIn posts for free, in your voice. If it sounds like you (or better), we install your AI Authority Engine in 90 days.

I started with no money, no contacts, no fancy city office.

Picture this:

A small apartment in Poland.
No team.
Nothing to “show off.”
No one backing me up.

Every day – me, my laptop, and clients who didn’t know my name.

Most days felt like:

Write posts.
Wait.
Hope for an inbound lead.

(Usually nothing.)

B2B tech leaders... yep, tough crowd.

I’m creative, talkative, maybe too much for some.

Every call made me sweat.

Every pitch was incredibly awkward.

Sometimes I wondered if I was nuts for picking this path.

But I couldn’t quit.

I learned to spot what they really needed.

I tried new ideas - most failed, a few worked.

I worked with scraps.

Fixed broken things.

Built trust, one honest message at a time.

I got used to hearing “no.”

I got used to waiting.

But I kept sending, kept building, kept showing up.

Then, slowly...

A “maybe.”
A client who said, “you get what we need.”
A few more.

Then referrals.

All because I'm too stubborn to stop.

Now, I help B2B teams turn strangers into real leads.

One content system, built from scratch.

Starting with nothing makes you sharp.

You learn what matters FAST.

If you feel lost in your work, or your niche feels impossible, keep at it.

Your “weirdness” or “wrong fit” could be the thing they need most.

PS. What did you learn from your first clients when you started from zero?

Here's a 10-min flow I use with founders to create weeks of content for them:

The goal is to create one small, high-integrity narrative object that can survive multiple contexts without losing its meaning. 

The easiest way to do that is to start with a 10-minute CEO video (use Tella or Loom), because spoken answers preserve something writing tends to bleach out: conviction, nuance, and those small qualifiers that signal real experience. 

I beg you, don’t ask for “thought leadership.” 

Ask one tight prompt that forces specificity, like what buyers are getting wrong about a problem right now, what that misunderstanding costs them, and what they should do instead.

Once you have the Loom, the extraction step matters more than the editing. 

You’re looking for a structure:

Pull out:

- one contrarian claim that creates tension
- one vivid symptom that makes the claim relatable
- one constraint that makes it believable
- one before/after framing that makes the shift obvious
- one proof detail that anchors it in reality

That proof detail can be lightweight. It can be “we heard this on 12 calls this month,” or “this showed up in every deal we lost last quarter.” 

Now you take those five ingredients and you write one “source paragraph” in plain language. Don't think of any platform yet. 

Just articulate the idea clearly as if you were explaining it to a smart friend over coffee. This paragraph becomes the anchor. Every other asset is a translation of it, not a reinvention. That’s how you keep integrity when you repurpose.

From there, I like to create three outputs that each do a different cognitive job:

First, a short text post that opens with the contrarian claim and immediately explains the symptom. 

The mistake people make is opening with the claim and then drifting into abstraction. Don’t. 

Earn the reader’s trust by describing what they already see in the wild, the call snippets, the internal debates, the marketing copy that “sounds right” but doesn’t move anything.

Second, a simple carousel where each slide is a single unit of meaning. Not “tips,” but progression. The before model, the cost of that model, the constraint that keeps people stuck, the after model, and then one practical move they can try this week. 

If you can’t fit the idea on five slides without it turning into fog, the idea isn’t ready yet.

Third, a longer form piece where you unpack the constraint and the proof detail. 

This is where nuance belongs. Why the obvious fix fails. What tradeoff you’re making. What to watch out for. 

This is also where you can safely include a real example, even if it’s anonymized, because you have space to set context.

Two quality checks before you hit publish:

• If the post could apply to any company, it’s too generic. 

• If the reader can’t repeat your point in one sentence, it’s too complicated.

If you want, reply with your niche and I’ll give you 5 tight CEO prompts that reliably produce these “weeks of content” anchors.

I recently changed my banner to this one below and you might think I went insane:

This is not a typical path content people choose for their businesses.

“Free value” beyond a regurgitated ChatGPT lead magnet is a rare thing here.

My thinking is this:

I built an Authority System for copying B2B founders’ voice with AI.

It’s GOOD. Took me about a year to get here. 

And the easiest way to prove it is to show my potential clients how it works.

Yes, building a sample assistant is harder than posting fancy carousels and “vulnerable” fake-ass stories

But I’ve always been more of a show-don’t-tell kinda guy and not planning to stop now.

Curious how it works? Dm me “3 posts”

"comment on other creators' posts" is sh!tty advice:

I used to spend 2 hours per day here commenting on probably all of them big creators.

Most of them don't give a rat's a*s about me or my content.

And that's perfectly fine. Why would they?

Not once have I reached out to them in a DM, they have no reason to see my comment and think:

"I wonder what this guy is about?"

But the folks I exchanged DMs with are consistently showing up under my posts.

Because we established a more meaningful connection than a [put an insightful comment here so they see you're trying hard]

See what I'm getting at?

Connection > tactics.

Build the first one and you'll never need to think about the second one again.

A year ago I thought I had it figured out:

Build a content agency for B2B founders. Use AI to do the heavy lifting. Keep costs low, scale fast. Simple.

What actually happened: 

I spent months building something nobody wanted to pay for.

The first version of the system was basically:

1. take the founder's old posts
2. dump them into a prompt
3. generate something that kind of sounded like them

The output was fine. Technically competent. The kind of content that gets 15 likes and zero leads.

I wasn't impressed with myself tbh. Neither were my clients.

So I overcomplicated everything:

A 12-step onboarding process, 5 different content formats, a pricing page that looked like a software feature comparison matrix...

I was trying to cover every possible use case for every possible client, and the result was that nobody could figure out what I actually did.

The solution was so stupidly simple, I facepalmed when I finally got there:

I started interviewing founders instead of just ingesting their existing content.

Turns out the best content a founder can publish isn't what they've already written.

 It's what they say when someone asks them the right questions:

- the unguarded takes
- the specific stories
- the things they'd never think to write down but say out loud constantly

Once I built the system around that – extract the thinking first, then write – the output stopped sounding like AI trying to be human and started sounding like the founder on their best day.

That took about a year to nail.

The offer got simpler too:

Interview you -> Write it up -> Publish it -> You approve the drafts. 

That's the whole thing.

If you want to see what it looks like, DM me "3 posts" – I'll write them for free before you spend anything.

A thing we rarely discuss in marketing is how customers perceive our narrative through preloaded meanings:

We think we’re publishing a simple story about a benefit. They’re interpreting a signal about power, intent, and control. 

That gap is what Ring ran into with its Super Bowl ad about a lost dog found through a neighborhood camera network. The creative aimed for warmth and community. 

The audience, already saturated with years of privacy discourse and surveillance headlines, watched the same 30 seconds and didn’t primarily evaluate the plot. They evaluated the infrastructure implied by the plot. 

“AI + cameras” isn’t a neutral prop anymore. It’s a symbol, and symbols arrive with cultural baggage attached.

This is the real mechanism behind “losing the narrative.” It’s rarely about execution quality but about the interpretive surface area. 

When a story leaves room for people to fill in the blanks, they will. And they’ll fill them with the strongest available frame in their mind, usually the one that best matches their anxieties. 

In this case, the counter-frame was waiting in the wings, mass surveillance disguised as safety. 

Once that frame establishes itself, everything that follows stops being “new information” and becomes “supporting evidence.” 

That’s why Ring’s sudden move away from a planned partnership with Flock Safety was immediately absorbed into the same storyline. Regardless of internal motives, it read like validation, not coincidence.

The takeaway isn’t “be careful with your messaging.” That’s too shallow. 

The takeaway is that narrative is not a message you transmit. It’s a negotiation between your intent and the audience’s prior beliefs. 

And prior beliefs are not something you overcome with a nicer tone. They’re something you have to design around. If your product or campaign touches an area where trust is already fragile, you should assume your audience will run an adversarial interpretation, even if they like you, even if the ad is well made.

A better move is to treat messaging like you treat product design:

Reduce ambiguity. Anticipate misuse scenarios. Make the boundaries legible. 

Don’t just tell a heartwarming story about what your system can do at its best. Show that you’ve thought seriously about what it could do at its worst, and what you’ve done to prevent that. 

In 2026, “we’re doing this for safety” is not a persuasive claim by itself. People want to know who gets visibility, who gets veto power, and what recourse exists when the system is wrong.

Curious how you think about this. When was the last time you watched a brand’s “feel good” story and immediately felt the darker subtext the audience would attach to it?

Quick tip if you hate when AI says "It's not X, it's Y..."

This rhetorical device is called "contrastive reframing using antithesis structure."

Hard-prompt it into your content assistant to avoid it (example below).

Also, from my tests, Claude 4.6 is much better at this that GPT models.

PS. Let me know any other annoying things AI uses - I'll show you how to remove them from the outputs.

How I helped a B2B consultancy business (very niche) start showing up in AI search:

We focused on turning their LinkedIn content into long-form SEO articles built around real opinions, specific frameworks, and language that maps to actual buyer problems.

AI tools cite sources that are clear and substantive. Since their LinkedIn content had this already, expanding it into articles became much easier.

The results measured over the last 90 days:

Overall: 1,619 AI bot hits → 4,163. That's 3x.
Homepage: 514 → 946. Almost 2x.

Every hit is ChatGPT or Perplexity pulling from their site while a buyer is actively researching. That's pipeline you'd never see in a cold outreach report.

If you want to see what this looks like for your niche, reply ENGINE and tell me what you sell.

200,000 people just watched a creator get stood up by a brand's AI chatbot:

Reformation used AI to manage its influencer outreach. It confirmed a collaboration, set up a store visit, and ran weeks of back-and-forth with a creator.

She drove an hour to the store, shot video in the dressing room... The manager had no idea who she was.

TikTok called it "an AI bot that went rogue." Reformation sent her a sweater as an apology.

Here's my read on what actually happened: 

AI made commitments nobody at the brand had authorized. And no human was watching the inbox it was running.

The bot did exactly what it was built to do - respond, confirm, follow up. 

But nobody told it where to stop. That's a systems failure.

AI doesn't know when it's overstepping. The tool runs until you build a stop - and plenty of AI marketing setups skip that step entirely.

There are places where automation works well: segmentation, sequences, content drafts.

And then there are contexts that require a human in the seat: real commitments, real relationships, anything a creator or partner will hold you to.

I put a lot of emphasis on that while working with my clients on their content assistants.

First draft can (and should) come from the AI. But the human editor before anything goes live is a must. 

Doesn't matter how good your AI tool is. If there's no one monitoring what it's agreeing to, you're one post away from a mess.

The CEO of impact put it well: "Human credibility becomes more valuable, not less."

That's the thing marketing teams running AI tools aren't accounting for right now, imo.

taplio

The Battle

Aakash Gupta

Kuba Czubajewski

Followers

Posts

Engagements

Likes

Performance ScoreAakash Gupta's LinkedIn Performance

Performance ScoreKuba Czubajewski's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Aakash Gupta

Kuba Czubajewski

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Aakash Gupta's LinkedIn Performance

Performance ScoreKuba Czubajewski's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreAakash Gupta's LinkedIn Performance