Anthropic’s $1B to $19B growth run: how Claude became the fastest-growing AI product in history | Amol Avasare
Amol Avasare, Head of Growth at Anthropic, shares how his team scaled the company from $1B to $19B ARR in 14 months. He discusses automating growth with AI, the evolving roles of PMs and engineers, and balancing rapid growth with Anthropic's core mission of AI safety.
Deep Dive Analysis
21 Topic Outline
Anthropic's Unprecedented Growth Trajectory
Amol's Cold Email Strategy to Join Anthropic
Leading Growth at a Hyper-Growth AI Company
The Role of Anthropic's Growth Team
Activation Challenges in AI Products
Improving Mercury's Onboarding Experience
Adding Strategic Friction to Onboarding Flows
Anthropic's Growth Org Structure and Big Bets
Automating Growth Experiments with Claude (CASH)
How AI Identifies Experiments to Run
Future of PM, Engineering, and Design Roles
Why More PMs Might Be Needed
Using AI to Prototype Ideas and Skip PRDs
Automating Daily Data Review with Cowork
AI for Managerial Coaching and Misalignment Detection
Anthropic's Focus on AI Coding and B2B
Balancing Growth with AI Safety Mission
Advice for Thriving in an AI-First Future
Anthropic's Unique Culture and Notebook Channels
Lessons from Shutting Down a Startup
Overcoming a Traumatic Brain Injury
5 Key Concepts
Success Disasters
Situations where rapid, exponential growth causes other systems or processes to break down, leading to intense firefighting despite positive overall metrics. This is a common challenge for hyper-growth companies like Anthropic.
Capability Overhang
The phenomenon in AI where models improve so rapidly that product development struggles to keep pace in diffusing those benefits to users. This makes it challenging to guide users to fully leverage new model capabilities before the next iteration arrives.
Freedom Through Constraints
A principle where having limitations or restrictions (personal or professional) can paradoxically lead to greater clarity, focus, and effective decision-making by eliminating excessive choices. This was a key learning from Amol's personal and professional experiences.
Agentic Coding
A concept where AI models can autonomously perform complex coding tasks, going beyond simple code generation to actively understand, plan, and execute development processes. This represents a significant leap in AI's impact on engineering productivity.
Notebook Channels
An internal Slack practice at Anthropic where every employee maintains a personal channel to share thoughts, updates, and provocative ideas. This fosters an open culture, facilitates knowledge sharing, and scales leadership's beliefs across the organization.
9 Questions Answered
He cold-emailed Mike Krieger, Anthropic's CPO, expressing interest and highlighting the company's need for a growth team, which perfectly timed with their internal discussions about hiring for growth.
AI models are improving exponentially, leading to a 'capability overhang' where products struggle to diffuse new benefits to users, making it hard for people to understand and leverage the full potential of the AI.
Anthropic's growth team prioritizes larger, core product-type bets (70/30 split) over micro-optimizations, believing that the exponential growth of AI capabilities unlocks new, significantly larger markets.
Anthropic's 'CASH' initiative uses Claude to automate the entire growth experimentation loop, from identifying opportunities and building features to testing and analyzing results, with human approval currently in the loop.
Engineers are gaining the most leverage from AI, potentially leading to a need for more PMs to manage effectively larger 'groups' of engineers, or for engineers to take on mini-PM responsibilities for smaller projects.
Tools like Cowork, integrated with Slack, can analyze internal communications to proactively identify potential areas of misalignment among team members, reducing wasted effort and improving coordination.
Anthropic is structured as a public benefit corporation, legally prioritizing public benefit and AI safety over maximizing shareholder value, and is willing to forgo metric impact for safety, brand, and user experience.
Every employee has a personal Slack channel to share thoughts, updates, and provocative ideas, fostering an open culture, scaling beliefs, and providing context for AI agents.
This principle suggests that limitations, whether personal or professional, can paradoxically lead to greater clarity, focus, and effective decision-making by reducing overwhelming choices.
16 Actionable Insights
1. Prioritize Big Bets in AI
If your product’s core value is AI-driven, shift your growth strategy to focus more on large, transformative bets rather than small optimizations. The exponential growth of AI capabilities constantly unlocks new, significantly larger markets, making big swings more impactful.
2. Automate Growth Experimentation
Leverage AI tools, such as Claude, to automate the entire growth experimentation process, from identifying opportunities and building features to testing and analyzing data. This can significantly increase the volume and efficiency of experiments, even if initial win rates are comparable to a junior PM.
3. Deputize Engineers as Mini-PMs
For projects requiring two weeks or less of engineering time, empower product-minded engineers to take on mini-PM responsibilities, including cross-functional coordination. This leverages their increased productivity from AI tools and frees up PMs for higher-leverage strategic work.
4. Add Strategic Friction to Onboarding
Don’t shy away from adding friction, such as quizzes or extra steps, in your onboarding flow if it helps users understand why the product is for them and guides them to relevant features. This can lead to higher conversion, better activation, and improved long-term engagement.
5. Automate Daily Data Review
Set up AI tools like Cowork to automatically review 20-25 charts and metrics every morning, distilling key insights and flagging concerns. This saves significant time and provides proactive monitoring of a wide range of data.
6. Use AI for Managerial Coaching
Utilize AI to analyze direct reports’ activities, team goals, and discussions to generate insights and feedback. You can also ask an AI version of your manager for feedback on your own performance, leveraging their public and internal knowledge.
7. Focus on Quality for Growth
Prioritize fixing quality issues in critical user flows, such as onboarding, even if it means temporarily setting aside metric targets. A significant investment in improving the user experience can lead to substantial and lasting uplifts in completion and retention.
8. Minimize Bureaucracy, Prototype Ideas
Reduce reliance on extensive documentation like PRDs for most projects, especially smaller ones, to accelerate action and minimize bureaucracy. Instead, prioritize prototyping ideas to articulate concepts and streamline the development process.
9. Detect Misalignment with AI
Leverage AI tools, such as Cowork with Slack integration, to proactively scan internal communications for potential areas of team misalignment. This helps identify and address issues early, reducing wasted effort and improving cross-functional coordination.
10. Automate Personal Admin Tasks
Delegate routine administrative tasks like booking meeting rooms, clearing email inboxes, and filing reimbursements or expenses to AI tools. This frees up significant personal time for higher-value, strategic work.
11. Be Comfortable Leaving Money on the Table
As a growth team, prioritize safety, brand, quality, and user experience over squeezing every last dollar or metric. This long-term perspective builds trust with users and ultimately drives more sustainable growth.
12. Continuously Use AI Tools
Stay on top of the latest AI tools and model releases, actively experimenting with their capabilities and applying them to your job. This improves personal productivity and develops crucial product sense for AI-first products.
13. Double Down on Unique Strengths
Identify your unique skill sets or ‘spikes’ (e.g., craft, stakeholder mediation, interdisciplinary knowledge) and relentlessly develop them. This makes you an indispensable ‘unicorn’ in an AI-accelerated world by increasing your overall impact.
14. Embrace Freedom Through Constraints
View constraints, whether personal or professional, as opportunities to adapt, focus, and make clear choices. This mindset can reduce excess choice and lead to more impactful decisions and personal growth.
15. Be Adaptable to New Playbooks
Recognize that traditional playbooks and operating methods may become irrelevant in an AI-first environment. Be willing to discard old approaches and adapt quickly to new realities to avoid friction and thrive.
16. Perfect Cold Email Outreach
To secure opportunities, master cold email by crafting high-open-rate subject lines, finding personal contact information, keeping messages short and direct, and following up persistently.
7 Key Quotes
50, 60, 70% of how you operate in the past just throw it out the door.
Amol Avasare
The product value that we will deliver in two years' time is probably like 1,000x what it is today.
Amol Avasare
Linear charts are just not cool. Everything is log linear. Just show me a log linear scale.
Amol Avasare
We will have AGI. And it will still be impossible to get six people in a room to get to align.
Joel (Head of Design at Anthropic)
If you're building a product where it's an AI-first product, then I would definitely operate in this way.
Amol Avasare
One of the biggest mistakes I feel like I see growth teams make and particularly just like hardcore growth practitioners is, is just trying to squeeze every last dollar.
Amol Avasare
The true freedom in life is learning how to be content when you don't get what you want.
Amol Avasare (quoting a meditation teacher)
4 Protocols
Cold Email Strategy
Amol Avasare- Craft a subject line with a very high open rate.
- Find personal emails to avoid crowded work inboxes.
- Keep the message very short, stating who you are, why you're a good fit, and suggesting a chat.
- Follow up persistently until told to stop.
Managing Projects with AI-Accelerated Engineers
Amol Avasare- For projects requiring two weeks or less of engineering time: The engineer is on the hook to effectively be the PM, handling security, legal, and cross-functional stakeholders. The PM will get looped in and advise if needed.
- For projects requiring more than two engineering weeks: The PM should continue to be on the hook for making that go well, delegating more to engineering but remaining squarely accountable.
- For highly controversial one-week projects: The PM should probably still drive it, despite its short duration.
Daily AI-Powered Data Review
Amol Avasare- Go onto Cowork and download the desktop app.
- Connect the Slack MCP (may require team/enterprise admin permissions).
- Create a scheduled task in Cowork to look at a bunch of hex links or other data sources (e.g., using the Chrome extension or MCP).
- Receive a summary from Claude every morning highlighting concerning items and interesting insights from 20-25 different charts.
AI-Assisted Managerial Coaching
Amol Avasare- Connect Cowork with the Slack MCP (may require team/enterprise admin permissions).
- Ask Claude to look into what direct reports have done this week, review team goals/OKRs, and analyze discussion transcripts.
- Receive key takeaways, observations, and suggested feedback to give direct reports.
- For self-coaching: Ask Claude, based on what it knows of your manager (publicly and internally) and your own actions, for feedback on your performance.