Enzo Avigo

Amplitude

Hey, I'm Enzo. I worked in Product Management for a decade and learned something surprising. Most tech companies still don't know what people do with their products. Data tools exist, but they follow the same pattern: a steep learning curve, and ongoing work to extract some value. What if instead, your analytics could work for you? 

A few years ago I left my comfortable job at Intercom to build that, it was called June.so (YC W21). Recently Amplitude acquired it and so we joined them 💜 ! 

Now I'm working on turning our product analytics tool into a simple, all in one platform. It will help teams understand what people do, why they do it, and what to do next. Useful for both product and marketing teams, with the help of AI.

> Want see my face: https://bit.ly/EnzoOnYoutube 
> Twitter addict? twitter.com/0zne

June is the next-gen analytics for B2B SaaS (YC W21).

Let’s bring some magic back to analytics 🎉

Claude Code is exploding right now.

People talk about vibe coding, but the real unlock is "skills".

You can turn Claude into a PM, researcher, analyst, or automation engine just by loading the right skill set.

❌ No complex setup
❌ No agents to build
❌ No workflow tools to learn

Just drop a "skill", open chat, and go.

I put together a visual with vetted Claude Code skills you can use right now, with direct links.

Here are 9 powerful ones to start with

⸻

1.	Interview / meeting insights
https://lnkd.in/eG7BuMnp

Summarizes meetings, extracts decisions, and turns messy conversations into usable notes

⸻

2.	Research + writing assistant
https://lnkd.in/e7jHZDQw

Helps with structured research and content creation

⸻

3.	File / content organizer
https://lnkd.in/eC-4RF89

Useful for cleaning up folders, docs, and unstructured content

⸻

4.	Full agile workflow skill set repo
https://lnkd.in/e5yTuiUf

Brings planning, tickets, and delivery workflows into Claude

⸻

5.	Skill creator
https://lnkd.in/eaczJDFw

Lets you generate new Claude skills from scratch

⸻

6.	Claude skills integration templates
https://lnkd.in/efpmdAME

Templates that help Claude Skills work across different agent tools

⸻

7.	Scientific / data / analysis skill set
https://lnkd.in/eusfSUjH

For research, analysis, reporting, and technical workflows

⸻

8.	Large Claude Code workflow collection
https://lnkd.in/e3Pq7EM6

Big collection of Claude Code workflows and resources

⸻

9.	Skills search
https://lnkd.in/e8UCXiNf

A plugin-style skill for discovering and managing Claude Code skills

⸻

Skills are the real power of Claude Code.

They let AI:

✅  Follow real workflows
✅  Use structured methods
✅  Behave like a specialized teammate
✅  Automate multi step work

All inside one chat.

Which skill should I test next?
And which ones did I miss?

It’s March 2026. AI can write your PRD, analyze your data, and summarize research. The only real question left is whether you can guide it.

There is something many people misunderstand about AI.

It doesn’t replace most jobs ❌ 

Instead, it will replace "tasks" inside jobs ✅ 

1/ THIS HAS HAPPENED BEFORE

It did, with secretaries. 

When computers arrived, many predicted the role would disappear.

Instead, tasks changed.

Typing, scheduling & formatting documents were automated. 
But the job stayed ✅ 

It evolved into coordination, operations, and decision support.
And you could argue that the work became more interesting!

2/ THIS PATTERN SHOWS UP IN EVERY TECHNOLOGICAL SHIFT

During the industrial revolution, many professions remained, but machines automated "parts" of the work.

Craft jobs didn’t vanish  ❌ 
The workflow changed  ✅ 

We’re seeing the same thing now with knowledge work.

3/ AI IS DOING THIS TO PRODUCT MANAGEMENT

AI can now help with tasks like:

- writing documentation
- summarizing meetings
- synthesizing customer feedback

Etc...

That’s why many PMs suddenly feel 2× faster.

4/ THE TRAP

AI performs the task now but... 
you still need to understand how the task works. 

Otherwise you cannot tell if the output is good.

Let me repeat that:

AI can automate most of the work.
The hard part is knowing what's great. 

In other words, guiding it.

5/ THIS IS WHY FOUNDATIONS MATTER MORE THAN EVER

To pilot AI well, we need deep understanding of the craft, and there is no shortcut for that.

We have to learn from the best. 
Study the foundations, practice the work ourselves.

Oftentimes this means spending hours doing things ourselves. 
Without AI.

Hundreds of hours.
Only then can AI become a multiplier.

______

The graphic below shows:

- 10 core product management skills
- how AI is changing them
- some of the popular tools product managers use to automate them

But the most important part is the last column: 
Knowledge 🧠

That’s how you can guide AI.

The 10 books I listed changed how I think about those skills. 
They gave me some foundations to master some of them.
I highly recommend reading them!

What resources/books/courses am I missing?? :)

Hope this helps 💜
--Enzo

Nothing is ever supposed to stop anymore. I go back and forth between Paris and San Francisco. They are no longer living in the same time system.

Paris rests. 
San Francisco runs overnight.

I didn’t understand the mood in San Francisco until I realised nothing is ever supposed to stop anymore.

Every time I land there, the change is subtle but hard to miss. 

Not louder. 
Not faster. 
Just more… continuously active.

People leave dinners early, not because they are tired, but because something is processing at home.

Cafés are full, but half the tables are quietly monitoring outputs AI generates for them.

Morning routines start with reviewing the agent that ran overnight.

Weekends sound less like plans. 
And more like execution windows.

To the extend that a new social reflex has appeared:

- What do you have running right now?
- How many agents are active?
- Did you let it run unattended?
- How much did that workflow consume?

Continuous activity has become a form of status.

Walking through the city, you see physical clues everywhere:

- conversations that pause mid sentence because a process completed

- people describing models like collaborators with personalities and moods

- friends comparing automation stacks the way people once compared gyms or diets

There is a shift in atmosphere. 

Fewer visible pauses. 

Almost no moments where nothing is happening.

And when nothing is happening, people seem uneasy.

That’s the part that surprised me most. 

A new background tension. 

Not stress exactly. 
More like a low level alert state.

The feeling that idle systems mean missed opportunity.

Some people call it « token anxiety »

It’s not fear of AI. 
It’s not even fear of competition. 

It’s the persistent awareness that productive activity no longer depends on your presence. 

And therefore maybe it should never stop.

San Francisco feels like the first place where people are adapting emotionally to continuous digital productivity.

When I return to Paris, the contrast is almost disorienting. 

Proper lunches. 
Closed laptops. 
Café en terrasse.
Evenings that fully end. 

Entire hours where nothing is being optimised.

Then I go back to San Francisco and within a day I catch myself checking progress logs while waiting in line for coffee.

That’s when you realise it’s contagious.

Token anxiety might be the first emotional side effect of a world where work never actually pauses.

Only humans do. Sometimes.

No alternative text description for this image

Figma stock is down ~75% YoY. For months, the market has been expecting a new generation of design tools. Yesterday, it finally arrived.

For the past 15 years, building software meant designing the interface first.

A typical workflow looked like this:

1.	open your favourite design tool
2.	create screens/frames
3.	design flows
4.	handover to engineers

It worked in the SaaS era.

Not anymore.

When software can generate UI, the interface stops being the source of truth. The system becomes the source of truth.

Why? Simply because shipping becomes faster, cheaper AND a better source of validation.

We’ve seen this shift happening for 2 years now.
But we’ve been missing a new generation of design tools to go with it.

Sure there were a few attempts:

- Figma launching MCPs
- Photoshop allowing AI image generation

But nothing really native and purposely built for this new era we’re in.

I think this changed yesterday when "Paper" launched.

I call it a system design tool.
(for lack of a better term, help me! 😅)

With tools like Paper, teams stop drawing screens. Instead, they define primitives, like:
 
- layout (sidebar + content + panel)
- state (loading, empty, error, success)
- data (table, chart, list, form)

If you already have a code base you just pull it.
And if you don’t have a system in place, well you just ask AI to generate it for you.

All at once, the whole design system building process become automated. And frankly for the better. Besides a few branding arguments, SaaS apps structures are insanely standardized. 

What this means in practice:

❌  You don’t design a layout
✅  You define the screen structure in English

❌  You don’t design 20 validation/error modals
✅  You define the modal primitive

❌  You don’t design responsiveness for mobile
✅  You define a mobile primitive (or copy it)

Pretty neat, right? 😍

This is very different from how Figma + MCP works for instance.

Figma was built for humans moving pixels.
Paper & Cos are built for systems generating interfaces and that feels a lot more native.

And just like that, Paper might do to Figma what Figma did to Sketch.

Not because Figma is bad.
But because the primitives changed.

We’ve seen this shift before.

- Sketch killed Photoshop for web design by introducing vector based interface design
 
- Figma killed Sketch for product teams with rich components, multiplayer and handover baked in

- and so I bet Paper & Cos will start to replace Figma for AI native startups 

⸻

When I see Paper, one more thought comes to mind.

Adapting our SaaS to an AI-first world isn't only about adding new layers. Like MCPs, copilots, or plugins.

Sometimes it means changing paradigms.

When interfaces are generated, pixels stop being the product. Systems become the product, and tools built for the old primitive rarely survive the new one.

____

Are you bullish on Paper at all? 
What am I missing? 🙃

Teams can build anything now.
That’s exactly the problem.

Teams can ship in hours...
but speed never answered the real question:

Are you building the right thing? ⭐️  

Yes, we're in a era where millions can create products almost instantly.

However, 90% of startups still fail. 
Even with faster development...

Why is that, you may ask?

Because building was never the main constraint.
Understanding users is 💖

So we've been working on changing that!

I'm VERY VERY EXCITED to introduce the next evolution of Amplitude.

It comes with 3 core new features:

1. Global Agent

A general conversational analytics agent.
That achieves an overall score of 76% descriptive & diagnostic analytics tasks.

Think Claude, but natively built for data analysis. 

2. Specialized Agents

Purpose-built agents that run autonomously and async. 

For instance:

- an agent that watches 100s of session recordings, and give insights
- a CRO agent that analyses your website, and automatically optimizes it

3. MCP & integrations

MCP opens a new world of possibilities.
Together with them we're launching new integrations:

- Cursor
- Lovable
- Figma

... and more.
 
Think analytics, directly where you build, design and prototype.

With this launch, analytics is becoming agentic 🤖 🤝 🤖 

An agent continuously understands behavior.
Reasons about what matters.

Communicate with another agent when needed.
And even take action!

________

A few words about our launch video ☺️

I got to work on it with an incredible team.
(thanks for trusting me Amplitude 🫶)

We wanted something honest so we recorded the voice actor in a studio.
Iterated for weeks with creatives.

Refined every frame, every transition, every beat of the story.

Let me know what you think!

🔥🎁 Gift for PMs: The Claude Cowork playbooks!

Still doing release notes, triage, research, and interview scheduling manually?

I built 5 ready-to-use Claude Cowork playbooks for product managers

If you’ve seen Claude Cowork, you know how powerful it is.

If you aren’t sure where to start, I’ve put together a starter pack with real PM use cases.

Inside the sheet:

✔️  Market research digest (auto summary from Drive + browser)
✔️  Interview scheduling assistant (calendar + email)
✔️  Competitor monitoring report (weekly auto check using Chrome)
✔️  Support feedback prioritization (Intercom + CRM)
✔️  Release notes generator (tickets → clean notes)

Each playbook includes:

✅  The exact prompt to copy
✅  Required MCPs / connections to enable
✅  Your folder setup in Google Drive
✅  Scheduler instructions

& the expected output

You just paste → connect → run

Previews below, but full prompts don’t fit here, so I’ve added them to a sheet.

Comment COWORK (or anything 😅) and I’ll message you the spreadsheet with full prompts + setup

(and add me if we’re not connected yet)

Hope it helps 💜
Enzo

taplio

Enzo Avigo's Linkedin Analytics

Enzo Avigo

Performance ScoreEnzo Avigo's LinkedIn Performance

What is Enzo talking about?

Enzo Avigo's Best Posts (last 30 days)

Ready to grow your LinkedIn presence?

taplio

.css-6ytdur{background-image:linear-gradient(284deg, rgba(39, 43, 72, 1) 0%, rgba(33, 47, 183, 1) 100%);color:transparent;-webkit-background-clip:text;background-clip:text;}Enzo Avigo's Linkedin Analytics

Enzo Avigo

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);}Enzo Avigo's LinkedIn Performance

What is Enzo talking about?

Enzo Avigo's Best Posts .css-1emoo84{font-size:var(--chakra-fontSizes-md);font-weight:400;color:var(--chakra-colors-gray-600);}(last 30 days)

Ready to grow your LinkedIn presence?

Enzo Avigo's Linkedin Analytics

Performance ScoreEnzo Avigo's LinkedIn Performance

Enzo Avigo's Best Posts (last 30 days)