Sahil argues that poor results do not always mean AI coding is broadly ineffective. Current tools are much stronger on popular front-end technologies and open-source component ecosystems, so teams using less represented stacks may spend more time correcting the model than benefiting from it.

AI coding capability varies sharply by language, framework, and task
React and common front-end libraries benefit from abundant training examples
A legacy or uncommon stack can hide the productivity available elsewhere
Technology migrations may become an AI productivity decision

“if you're not using those sorts of tools, you're not gonna get the value.”
Sahil Lavingia

“It's not that good, it makes a lot of mistakes.”
Sahil Lavingia

#ai coding#tech stack#react#open source

Myth Buster

Full Automation Keeps Receding Up the Abstraction Ladder

Sahil has become less confident that companies will reach full automation soon. As AI absorbs one layer of execution, people move to a higher layer of judgment, research, strategy, and invention, creating more valuable work instead of simply ending the function.

Automation removes tasks before it removes entire functions
Humans move toward higher-level judgment as execution gets cheaper
User, design, and market research still require people to ask questions
Radical ideas may remain outside the model's most probable suggestions

“I actually was probably more optimistic on like full automation. I don't think we're gonna really get there for a long time.”
Sahil Lavingia

“There's just always like a higher level abstraction that you get to operate at.”
Sahil Lavingia

#automation#future of work#strategy#research

Hot Take· 3

Hot Take

Human Engineers May Become the Platform Team for AI Builders

Sahil predicts that much human engineering work will shift from directly shipping features to removing technical debt and preparing the environment in which AI can ship. Standards, linting, CI, infrastructure, and developer setup become the enabling layer, while designers gain more ability to implement detailed product interactions themselves.

Technical debt becomes a direct constraint on AI-generated delivery
Engineers increasingly own foundations, standards, and architecture
Designers can move closer to shipping complete features
An environment that is easy for AI is also easier for new hires

“the majority of human engineering will be removing tech debt such that AI engineers can actually ship features.”
Sahil Lavingia

“Basically, like designers will be shipping features”
Sahil Lavingia

#engineering#tech debt#design#developer experience

Hot Take

If AI Removes Scarcity, Prioritization Changes Meaning

Product prioritization exists because engineering time and attention are limited. Sahil asks what happens when agents can clear every issue cheaply: teams may spend less time ranking known tasks and more time discovering radical ideas, talking to users, and inventing entirely new product directions.

Prioritization is a response to limited resources
Cheap execution could drive the issue backlog toward zero
Teams would redirect time from ranking tasks to finding new problems
The remaining work becomes research and non-obvious invention

“Prioritization is a function of like limited resources.”
Sahil Lavingia

“Like, because every issue is is solved.”
Sahil Lavingia

#prioritization#resource constraints#product strategy#innovation

Hot Take

AI Will Raise the Creative Bar From a Painting to a Movie

Sahil expects abundant AI-generated content to raise what audiences consider noteworthy. Work that once earned attention on craft alone may need far more narrative detail, novelty, and production value, with creators using the new capacity to make richer artifacts rather than simply publishing more of the old format.

Cheap production increases the baseline supply of competent content
Audience expectations rise with the available creative capacity
More elaborate work becomes normal rather than exceptional
Creators must use AI capacity to deepen the idea, not only increase volume

“It's just like you have to like up the game more and more and more.”
Sahil Lavingia

“But like in five years, it's gonna be like you need to like post the freaking movie.”
Sahil Lavingia

#marketing#content creation#creative work#attention

Explainer· 1

Explainer

How AI Could Turn Reactive Support Into Proactive Sales

Most AI support systems wait for a customer to ask a question. Sahil imagines the same context and personalization being used proactively, from recognizing a visitor's situation to opening a useful conversation before they file a ticket, which moves support closer to sales.

Current AI support is mostly reactive
Signup and browsing context can trigger tailored outreach
Personalized conversation can surface customer problems earlier
Department boundaries may blur as tools gain more context

“But this is all like reactive, you know.”
Sahil Lavingia

“I think sales like making it more making support more about sales, making it more proactive.”
Sahil Lavingia

#customer support#sales#personalization#automation

Story· 1

Story

Gumroad Lets Its Top Creators Weight the Roadmap

Gumroad sent a proposed feature list to its top 200 creators and ranked the roadmap using what those customers wanted. Sahil sees an opportunity for AI to combine that feedback with customer sales volume and engineering effort, producing a more informed priority order than any single person can hold in their head.

Gumroad asks high-value creators directly what they want
Customer feedback is weighted by the people funding the business
Sahil currently compares expected creator value with engineering hours
AI could join feedback, revenue data, and technical complexity

“we just sent this Google Doc to like our top 200 creators in 2024.”
Sahil Lavingia

“we kind of like rank this based on what they wanted from us, because it turns out like they're the ones paying our bills, right?”
Sahil Lavingia

#roadmapping#customer feedback#prioritization#gumroad

Tool· 2

Tool

Pick an AI Tool by Role, but Start With V0

Sahil's recommendation changes with the user's role: product people can learn fastest through V0, engineers may prefer Cursor's agent mode, and CEOs may find Devin's Slack-based delegation most impressive. He still calls V0 the lowest-hanging fruit because anyone can turn an idea into a working interface and share it through a URL.

Product builders benefit from V0's visual feedback
Engineers can start with Cursor agent mode
Executives may value Devin's asynchronous delegation
V0 makes AI capability visible before requiring deeper coding skills

“But V0, I think it's just like the lowest hanging fruit.”
Sahil Lavingia

“And then, you know, the cool thing about V0 is it shows you what's possible”
Sahil Lavingia

#v0#cursor#devin#ai tools

Tool

Two Tiny Prompt Tricks: Capital Letters and ‘Etc.’

When an AI system keeps ignoring a critical instruction, Sahil uses capital letters to signal emphasis. When he wants breadth or creativity, he gives two or three examples and adds “etc.” so the model continues the pattern rather than stopping at the supplied list.

Use capitalization to mark the instruction that must not be ignored
Give a few examples to establish the desired pattern
Add “etc.” to invite the model to extend the list creatively
Use these tactics as lightweight signals, not substitutes for a clear prompt

“Like, please do not ignore this specific part.”
Sahil Lavingia

“So if you want a list of things, you you can name like two of them and then just say etc.”
Sahil Lavingia

#prompting#ai tips#creativity

Takeaway· 1

Takeaway

AI Adoption Creates Both Job Security and Job Insecurity

The fear surrounding AI adoption is not just resistance to learning a new tool. Sahil says people are confronting the possibility that the work currently making them valuable may stop mattering, while avoiding the tools can make that insecurity worse as the surrounding standard keeps rising.

Organizational change consumes energy and creates real discomfort
People fear losing value, not merely learning unfamiliar software
Avoidance does not preserve the old competitive baseline
Trying the tools is part of adapting to the new standard

“There's like a fear of change. It's like job security, right? But at the end of the day, I think it's sort of also job…”
Sahil Lavingia

“Like, we don't know if like what we do will continue to be valuable.”
Sahil Lavingia

#job security#change#ai adoption#careers

Frameworks in this episode

Human Takeoff and Landing Model

Lead-Train-Incentivize AI Adoption Sprint

V0-Devin-Cursor Build Flow

Insights & moments

⚡Myth Buster· 2

Why AI Coding Feels Weak on the Wrong Stack

Full Automation Keeps Receding Up the Abstraction Ladder

◆Hot Take· 3

Human Engineers May Become the Platform Team for AI Builders

If AI Removes Scarcity, Prioritization Changes Meaning

AI Will Raise the Creative Bar From a Painting to a Movie

✶Explainer· 1

How AI Could Turn Reactive Support Into Proactive Sales

❝Story· 1

Gumroad Lets Its Top Creators Weight the Roadmap

✦Tool· 2

Pick an AI Tool by Role, but Start With V0

Two Tiny Prompt Tricks: Capital Letters and ‘Etc.’

▲Takeaway· 1

AI Adoption Creates Both Job Security and Job Insecurity

Myth Buster· 2

Hot Take· 3

Explainer· 1

Story· 1

Tool· 2

Takeaway· 1