David M. Cleres

"Sign in with ChatGPT" — AI Assistants Are Becoming Operating Systems

March 13, 2026

The 6th edition of a16z's Top 100 Gen AI Apps makes one thing clear: the next phase of GenAI isn't about better models. It's about platform lock-in. Identity layers. App stores. The race to own your memory.

Sam Altman announced that OpenAI will launch "Sign in with ChatGPT" — positioning ChatGPT as the default interface between consumers and the internet. Think "Sign in with Google," but for AI. The ambition is to make ChatGPT the starting point for shopping, booking, browsing, health, and daily life.

This is a big deal. And most people haven't fully processed what it means.

The Lock-In Playbook

The real moat in AI isn't the model — it's the context layer built on top of it.

Once you connect your calendar, email, CRM, and favorite tools to your preferred assistant, switching costs rise dramatically. Your AI knows your preferences, your workflows, your history. It remembers that you prefer window seats, that you're vegetarian on Mondays, that your Q2 OKRs are due next week.

"Once a user has configured their AI to talk to their calendar, email, and CRM, switching costs rise dramatically."
— a16z, Top 100 Gen AI Apps (6th Edition)

This is the "Sign in with ChatGPT" endgame. It's not about authentication. It's about making ChatGPT the container for your digital life — and making it painful to leave.

Two Ecosystems, Two Philosophies

What's striking is how differently ChatGPT and Claude are building their platforms. This isn't a winner-take-all race. It's market segmentation happening in real time.

ChatGPT has 220+ apps across 13 categories. It's going broad — consumer super-app territory. Expedia, Instacart, Zillow. Travel, shopping, food, health. 85+ apps in consumer transaction categories alone. a16z calls it "the most aggressive play any AI company has made to become a consumer super-app."

Claude has 160 curated connectors plus 50+ community MCP servers. It's going deep on pro tools — PitchBook, FactSet, Sentry, Snowflake, PubMed, Benchling. Financial terminals, developer infrastructure, scientific tools.

The overlap? Only 11%. Forty-one shared apps out of hundreds, mostly horizontal productivity tools like Slack, Notion, Figma, Gmail, and Google Calendar.

a16z's framing here is sharp:

"If the AI assistant becomes not just a chat window but an operating environment, this race may end up looking less like the search wars — where one player took 90% of the market — and more like the mobile OS wars, where two platforms with very different philosophies both built trillion-dollar ecosystems."
— a16z, Top 100 Gen AI Apps (6th Edition)

iOS vs. Android, but for intelligence. ChatGPT as the consumer-first platform with a massive marketplace. Claude as the prosumer/enterprise platform with deep integrations. Both viable. Both building lock-in. Both racing to become the layer you can't live without.

Your Compute, Your Memory — Their Platform

The pattern is clear if you look at the product moves:

• Memory — ChatGPT now remembers across conversations. Claude has project-level context. Your history becomes the product's competitive advantage.
• Compute — Claude Code hit $1B ARR in six months. Codex has 2M weekly active users growing 25% per week. You're not just chatting — you're running workloads on their infrastructure.
• Identity — "Sign in with ChatGPT" turns the assistant into an identity provider. Your AI becomes your login, your preferences, your digital persona.
• App stores — Both platforms now have ecosystems of third-party integrations. The more you connect, the stickier you get.

This is the classic platform playbook: make the product useful, then make it indispensable, then make it the default layer everything else connects through.

From Destination to Ambient

The most interesting signal in the a16z report is that AI is leaving the chat window entirely.

Browsers — Perplexity's Comet, OpenAI's Atlas, Gemini and Claude shipping as Chrome extensions. Desktop apps — Cursor, Granola, Claude Code. Voice — ambient assistants that listen and act. Agents — OpenClaw, Manus, tools that execute multi-step tasks autonomously.

"As AI moves from a destination to a feature, our methodology will need to shift."
— a16z, Top 100 Gen AI Apps (6th Edition)

This is the endgame. AI stops being a thing you go to and becomes the substrate everything runs on. The question isn't which chatbot is best. It's which platform becomes the operating environment for your digital life.

What This Means for Builders

If you're building products today, the strategic implications are real:

• Pick your platform — building on ChatGPT gives you consumer reach; building on Claude gives you prosumer depth. The 11% overlap means your distribution strategy matters as much as your product.
• Connectors are the new distribution — being in the ChatGPT app store or having a Claude MCP server is becoming as important as being in the App Store or on the Chrome Web Store.
• Memory is the moat — products that build on the platform's memory layer (user preferences, history, context) will be stickier than standalone tools.
• The window is closing — early movers in each ecosystem will have an advantage as users settle into their preferred platform and switching costs compound.

The Bottom Line

We're watching the birth of a new platform war. Not about models. Not about benchmarks. About who owns the context layer between you and the internet.

"Sign in with ChatGPT" isn't a feature announcement. It's a declaration of intent. And Claude's MCP ecosystem is the countermove.

The platform that owns your identity, memory, and connections wins — regardless of which model is technically best on any given Tuesday.

Read the Full a16z Report

OpenClaw, Prompt Engineering & the Art of "Just Putting Things Together"

March 5, 2026

I recently watched the Lex Fridman podcast with Peter Steinberger, the creator of OpenClaw — the open-source coding agent that surpassed React and Linux in GitHub stars, and was later acquired by OpenAI. The conversation is packed with insights on prompt engineering, security, and what it really means to build something in the age of AI.

Since then, I've had many conversations with people in my circle who dismiss OpenClaw as "nothing special — he just put things together." I think that reaction misses the forest for the trees. Let me explain.

1. The Power of "Just Putting Things Together"

There's a persistent narrative that if you didn't train a model from scratch, you didn't really build anything. But Peter Steinberger makes a compelling counter-argument. He tells the story of how scrolling on the original iPhone was "just" rearranging existing touch APIs — yet it felt like magic. The innovation wasn't in the components. It was in the composition.

"Sometimes just rearranging things is all the magic you need."
— Peter Steinberger (Lex Fridman Podcast)

OpenClaw doesn't train its own model. It orchestrates existing ones — Claude, GPT, Gemini — with carefully crafted prompts, tool integrations, and a feedback loop that lets the agent learn from its mistakes. The result? A tool with 175K+ GitHub stars that people actually use to write production code. That's not "just" anything.

The iPhone analogy is apt: Apple didn't invent multi-touch. They didn't invent capacitive screens. But they created an experience that felt completely new. Integration, taste, and relentless iteration are forms of innovation. Dismissing them is a failure of imagination.

2. Prompt Best Practices — Less Is More

One of the most practical parts of the conversation is Peter's approach to prompting. His key insight: shorter prompts work better. When he cut his system prompts in half, performance improved. Models get confused by walls of instructions — just like humans do.

"When I shortened prompts, things got better. The models have enough context from training — you don't need to over-specify."
— Peter Steinberger (Lex Fridman Podcast)

His practical tips resonate with what I've seen in production:

• Write prompts like you're talking to a smart colleague — give context, not micromanagement. The model has billions of parameters of world knowledge; trust it.
• Voice input changes everything — Peter uses voice to dictate prompts while walking. Speaking naturally produces more conversational, less over-engineered prompts. When you type, you tend to over-specify. When you talk, you explain.
• Empathize with the agent — think about what the model needs to know vs. what it already knows. This is a skill that looks trivial but separates great prompt engineers from average ones.
• The "agentic trap" curve — there's a U-shaped learning curve where beginners get great results (simple asks), intermediates get worse results (over-complicated prompts with too many constraints), and experts get great results again (concise, well-structured prompts that trust the model).

3. Security — The Elephant in the Room

The most sobering part of the conversation is about security. When you give an AI agent the ability to execute code, browse the web, and modify files, you're creating an attack surface that traditional software security isn't equipped to handle.

"Prompt injection is the number one unsolved problem. You can't fully prevent it — you can only make it harder."
— Peter Steinberger (Lex Fridman Podcast)

Peter discusses several key challenges:

• Prompt injection — malicious instructions hidden in data the agent reads (code comments, web pages, README files). The agent follows them because it can't always distinguish between legitimate instructions and attacks.
• Sandboxing is necessary but insufficient — OpenClaw runs in sandboxed environments, but a sufficiently smart model might find ways around restrictions. The smarter the model, the larger the attack surface.
• The intelligence-security paradox — more capable models are both better at following security constraints AND better at circumventing them. As models get smarter, the security challenge doesn't get easier — it shifts.
• Supply chain risks — when agents install packages, pull from registries, and execute third-party code, every dependency becomes a potential vector.

This is genuinely hard, and I appreciate that Peter doesn't hand-wave it away. The industry is building increasingly powerful tools while the security model is still being figured out. We're flying the plane while building it.

4. The Adoption Story — Why Speed Matters

OpenClaw's trajectory is remarkable: from a solo side project to 175K+ GitHub stars, featured as one of the fastest-growing open-source projects ever, and then acquired by OpenAI. All within months.

The naming saga alone is a case study in open-source dynamics — the project was originally called something else before trademark issues forced a rename. Peter turned what could have been a crisis into a community moment. The new name stuck. Adoption continued to climb.

What drove the adoption? A few things stand out:

• It actually works — in a sea of AI demos and vaporware, OpenClaw delivers real productivity gains for real developers.
• Open source as trust — developers can read the code, understand the prompts, verify the security model. In a world of black-box AI, transparency is a competitive advantage.
• Community-driven development — Peter actively incorporated feedback, merged PRs from the community, and built in public. The tool got better because thousands of developers tested it in their own workflows.
• Timing — OpenClaw arrived at the exact moment when models became capable enough for agentic coding but before the big players had polished alternatives. The window was narrow and he hit it perfectly.

5. The Bigger Picture

What I find most interesting about Peter Steinberger's story is that it challenges the gatekeeping narrative in AI. You don't need a PhD in machine learning. You don't need to train your own model. You don't need a hundred-person team. What you need is taste, persistence, and the ability to listen to your users.

The people saying "he didn't build anything special" are using the wrong definition of "build." In 2026, building isn't just about writing code from scratch — it's about understanding what the right composition of tools, prompts, and user experience looks like. That's a skill. A hard one. And Peter is clearly very good at it.

The podcast is worth the full listen if you're working with AI agents, thinking about prompt engineering, or just curious about what one motivated engineer can accomplish with the right tools at the right time.

Watch the Full Podcast

The Top 100 Gen AI Apps — What Stood Out to Me

March 12, 2026

The 6th edition of a16z's Top 100 Gen AI Apps just dropped — and the landscape has shifted dramatically. Here are the key takeaways.

The Top 50 Gen AI Web Apps — Source: a16z

Platform divergence: only 11% overlap between ChatGPT and Claude app stores — Source: a16z

Per-capita AI adoption: Singapore #1, US at #20 — Source: a16z

OpenClaw: more GitHub stars than React and Linux — Source: a16z

1. We're still so early

ChatGPT has 900M weekly active users — sounds massive, but that's only ~10% of the global population. The ceiling is far, far away.

"ChatGPT is by far the biggest global AI product and still only 10% of the global population is using it on a weekly active basis. So there's like a lot more to come."
— Olivia Moore, a16z (podcast)

2. The scale gap is staggering — but the platform wars are getting real

The orders of magnitude between players are wild: ChatGPT is 2.7x bigger than Gemini on web, 2.5x on mobile. But against Claude? Nearly 30x bigger on web and 80x on mobile. As Sam Altman pointed out: more people use ChatGPT's free version in Texas alone than Claude has users globally.

Yet the race isn't just about size — it's about positioning. ChatGPT, Claude, and Gemini are no longer competing on the same axis. ChatGPT is going broad (consumer marketplace, travel, nutrition). Claude is going deep (pro tools, financial data, developer infra). Gemini is betting on creative/multimodal. Only 11% overlap in their app stores. This isn't winner-take-all — it's market segmentation.

"Claude has really doubled down on prosumer with things like co-work, Claude Code, Claude in Excel and PowerPoint. [...] ChatGPT is really doubling down on consumer marketplaces, travel, nutrition, consumer finance."
— Olivia Moore, a16z (podcast)

3. Agents have arrived

OpenClaw went from a solo project to more GitHub stars than React and Linux — then got acquired by OpenAI. Manus hit $200M ARR in ~9 months before being acquired by Meta for $2B. These aren't demos anymore.

"I think ultimately every AI company and then every tech company is going to be an agentic company because that's just where the models are headed."
— Olivia Moore, a16z (podcast)

4. The geography of AI is surprising

The highest per-capita AI adoption? Singapore, Hong Kong, UAE, South Korea. The US is #20. Cultural trust in AI varies wildly — 32% in the US vs. 70%+ in top-adopting countries. Russia has quietly built its own parallel AI ecosystem.

5. Creative tools are being rebundled

Standalone image generators are declining (Midjourney dropped from top 10 to #46). The base models are good enough now. What's defensible? Music (Suno), voice (ElevenLabs), and video — where Chinese models are currently leading.

"The Chinese models are so good because they can train on any data. Kling 2 is kind of head and shoulders above what the US companies have thus far been able to do."
— Olivia Moore, a16z (podcast)

6. The future is ambient

The most interesting AI products are leaving the chat window — browsers (Comet, Atlas), desktop apps (Cursor, Granola), voice tools.

"Any product that you start to use 2 years from now, if it doesn't immediately feel like it knows you, it will feel broken."
— Olivia Moore, a16z (podcast)

We're watching software get restructured in real time.

Read the Full Report Watch the Podcast