Welcome back!

This week we break down why letting users build for you beats building everything yourself, walk through n8n’s community-powered path to $2.5B valuation, explore an open-source tool for building private ChatGPT-style apps with your documents, and show how Intercom priced for trust while losing money on every sale.

Have suggestions? Send them to us at [email protected].

Founder’s Intel

Let users build what you can’t

Data Intel

• Platforms with 100+ community-built extensions report 3× higher retention than those with fewer than 10.

• Developer platforms that open contribution pathways see 35% higher engagement and 40% faster feature expansion without increasing headcount.

• Open-source projects with active contributor communities grow user bases 2.5× faster than closed alternatives in the same category.

Why It Matters

Every feature you build costs engineering time.

Every feature your users build costs you nothing.

The best platforms figured this out early and let users create templates, prompts, integrations, or assets that other users benefit from.

Each contribution makes a product more valuable, which attracts more users, who contribute more, and the flywheel spins without having to push it.

You see this pattern everywhere. n8n has 2,000+ community-built nodes, Figma has thousands of community plugins, and Notion templates spread across the internet with each one serving as a landing page pointing back to the product.

These companies didn’t hire armies of engineers to build all of it. They built systems that made contribution easy and rewarding, then got out of the way.

The math compounds fast when you think about it. One power user builds a template that helps 100 others, and ten of those users build their own templates.

Suddenly you have coverage across use cases you never planned for, built by people who understand those use cases better than you.

But this only works if contribution is frictionless.

If submitting a template takes 20 steps, nobody will do it. If sharing a workflow requires approval from your team, the bottleneck is you. The platforms that win make contribution as easy as using the product itself.

Quick Tip

Identify the most valuable thing power users create inside your product, then ask how hard it is for them to share that with others. If the answer is “impossible” or “very hard,” you’re leaving compounding growth on the table.

Build the share button, create the gallery, and let users see what others have made.

Contribution follows visibility.

Behind the Tool

How Killing Bias Built a 130K User Base

The Spark

Dileep was hooked on AI the moment ChatGPT launched. Like many technical founders, he immediately started listing ideas to build, but the problem was that he had no way to validate any of them.

Talking to friends and colleagues introduced bias. Everyone had their own perspective, their own blind spots. And not everyone had access to a mentor who could evaluate ideas objectively across industries and markets.

So Dileep built the validation tool he needed. AI could provide unbiased analysis of business ideas by drawing on knowledge across domains, without the personal filters that human advisors bring. That became Stratup.⁠ai (a wordplay on “strategy” plus “startup”).

The Build

Stratup.ai lets users brainstorm startup ideas with AI, generate detailed business reports, and ask follow-up questions to stress-test concepts. The platform tunes prompts to deliver polished, actionable outputs rather than generic responses.

The technical build was straightforward for Dileep, who comes from a pure tech background. The hard part was figuring out which features to ship, which to skip, and how users actually interacted with the platform. Measuring engagement and iterating based on real behavior was new territory.

What started as a solo ideation tool evolved into something bigger. Dileep added collaboration features so that users could work together on ideas, match with potential co-founders, and showcase their products to the community.

That shift turned Stratup.⁠ai from a tool into a network where aspiring founders could find each other.

The Breakthrough

Stratup.⁠⁠ai now has 130,000+ users with more than 200,000 ideas generated on the platform.

Dileep experimented with X (formerly Twitter) ads, Facebook ads, and Google ads, but while traffic came, conversions didn’t. That’s where TAAFT came in.

TAAFT was different because the audience consisted of AI enthusiasts actively looking for tools, and conversion rates justified the spend when other channels failed.

The combination of TAAFT visibility plus organic content created a steady acquisition channel without the content budget that would have been required otherwise.

The Next Chapter

Dilip is building Stratup. ai into an accountability platform for startup builders, with daily guided tasks that move users toward launching a product. AI acts as the mentor most founders never get access to.

A book is also in the works that’ll cover how to launch a business within 30 days in the age of AI. And the goal for the next five years is to see unicorns built through the platform.

Key Lessons

TAAFT converts when broad ads don’t. AI enthusiasts on discovery platforms have intent while generic social traffic often doesn’t. Testing TAAFT early with a small budget and measuring conversion rather than clicks proved to be the right approach.

Product sense is harder than tech. Engineers build features, but knowing which features to build and how users engage requires a different skill set entirely. Measuring behavior while iterating based on data became essential.

Bias is the real competitor. When users discuss ideas with friends or colleagues, personal filters distort feedback. But AI provides a perspective that draws on cross-industry knowledge without the emotional investment that humans bring.

Community compounds in unexpected ways. Co-founder matching and collaboration features turns solo users into a network, and people stay longer when they connect with others building alongside them.

Tool of the Week

n8n’s Community-Powered Growth Engine

What’s n8n?

A workflow automation platform that hit $40M ARR and a $2.5B valuation in October 2025. The teams serves more than 230,000 active users and 3,000+ enterprise customers. Revenue grew 10× in 2025 alone.

What Worked

• Fair-code licensing created a distribution moat: n8n isn’t fully open-source. Their “fair-code” model lets anyone self-host for free, but commercial hosting or embedding requires a paid license. This prevented cloud giants from cloning the product while still attracting a massive developer community. The result was nearly 2,000 community-built nodes that n8n never had to build themselves.

• Community became the product team: Instead of hiring engineers to build every integration, n8n opened the platform to contributors. Developers built nodes for their own use cases and then shared them. Each new node made n8n more valuable and it attracted more users, who built more nodes.

• Embedded partnerships drove 15% of revenue: Other software companies integrated n8n’s automation engine into their own products. These OEM deals generated recurring revenue while expanding distribution, with n8n powering workflows inside products whose users never knew they were running n8n.

• The AI pivot accelerated everything: n8n integrated LLMs directly into the platform in 2023, letting users build AI-powered workflows visually without writing code. 75% of customers use the AI features, and ARR grew 5× in a single year after the pivot because the timing aligned with enterprise AI adoption.

• Expert partner program formalized the ecosystem: In Q1 2025, n8n launched a program for consultants and agencies to build businesses around the platform. Partners get early access to features, commercial support, and co-marketing. n8n gets distribution through trusted implementers who recommend the platform to their clients.

• GitHub ranking serves as social proof: n8n ranks in the top 150 GitHub projects of all time with more than 70,000 stars. That credibility matters for developer tools because engineers trust projects with active communities, visible code, and real adoption. The ranking becomes a self-reinforcing signal that attracts more contributors.

Founder Quote

“Automation shouldn’t be a black box. Companies need transparency, customization, and cost efficiency. We’ve built more than a platform; we’ve built a community that loves and relies on us.” — Jan Oberhauser, n8n Founder and CEO

Key Lesson

Let your community build what you can’t. Open your platform to contributors, make it easy to extend, and the ecosystem will create value you never planned.

n8n’s community nodes would have taken years and millions to build internally, but instead developers built them for free because the platform made it worth their time.

Fresh Out of the Lab

AnythingLLM

What Is It?

An open-source desktop application that lets you build a private ChatGPT with your own documents. You upload PDFs, Word docs, CSVs, or code files, pick any LLM (local or cloud), and chat with your data. Running locally means no API costs and no data leaves your machine.

What’s New

AnythingLLM now supports MCP (Model Context Protocol) for tool integrations, along with automatic document sync that watches files and re-embeds when they change.

Workspace-level permissions enable multi-user deployments.

The latest updates added reranking options for better retrieval accuracy while streamlining the RAG pipeline so that you get relevant answers without configuration headaches.

Why It Matters

Founders building internal tools, research assistants, or customer-facing chatbots own the entire stack without API dependencies.

AnythingLLM runs on macOS, Windows, and Linux while supporting Ollama, LM Studio, OpenAI, Claude, and dozens of other providers.

Vector storage uses LanceDB by default with embedded zero-setup (swap to Pinecone, Chroma, or Weaviate if needed).

The killer feature is that it works offline with everything running locally, no signup required, and no telemetry unless you enable it.

For founders who need document chat without sending proprietary data to third parties, AnythingLLM is the fastest path to production.

Founder’s Edge

This Week’s Builder Toolkit

• Dev Tool: Void is a Cursor alternative that gives you full control over your code and data. You can use any model (Claude, GPT, Gemini, DeepSeek, or local via Ollama), transfer your VS Code settings easily, and connect directly to providers without a middleman.

• Free Dataset: Failory’s 100+ Startup Resources provides free Notion documents covering fundraising, product, people, and growth, plus post-mortems on failed startups to learn what not to do.

• No-Code App: Penpot is a design and prototyping platform that outputs CSS Flexbox and Grid, letting designers and developers work in the same file with no translation layer and it’s free forever.

• Productivity Hack: Raindrop.io organizes bookmarks, highlights, and full-page archives across devices so you tag, search, and share collections. The free tier handles most needs while Pro adds nested folders and permanent copies.

• Learning Resource: Elements of AI from University of Helsinki is a free, no-code introduction to AI that covers concepts, ethics, and real applications without requiring math or programming. It has over 1 million students enrolled.

Note: If you’ve found a tool that’s sped up your build process, hit reply and share it. We’ll feature the best submissions in a future issue.

AI Founder’s Journal

How Intercom Priced for Trust (and Won)

In late 2022, Intercom was dying. Five consecutive quarters of declining net new ARR, approaching zero growth, heading toward negative territory.

Then ChatGPT launched.

Within six weeks, Intercom had a working prototype of Fin.

McCabe priced Fin at $0.99 per resolved ticket while the cost to deliver was $1.21. Negative margin on every sale. But he wasn't guessing.

At the time, AI inference costs were already dropping fast, and the trajectory was predictable. He bet that costs would fall faster than competitors could react.

But the pricing did something else.

At $0.99 with a "pay only if it works" guarantee, customers had zero risk. No procurement battles, no pilot committees, no six-month evaluations. Just turn it on and see what happens.

This matters for AI tools. Most charge customers whether the output works or not. They prompt, they pay, even if they get garbage back.

Intercom took a different approach. If Fin didn't resolve the ticket, the customer paid nothing. That removed every objection.

And it paid off because adoption compounded fast.

More tickets meant more data, better models meant higher resolution rates, and higher resolution rates meant more customers. The flywheel started spinning.

Meanwhile, inference costs dropped exactly as McCabe predicted.

What cost $1.21 in early 2023 costs a fraction today. The margins flipped positive, and Intercom went from losing money to reporting strong gross margins.

Here's what made it compound:

Speed created the opening. Six weeks from ChatGPT launch to working prototype. While competitors were still planning, Intercom was already in market collecting data and iterating.

Outcome-based pricing removed objections. Customers only paid when Fin resolved tickets. No wasted credits, no paying for garbage outputs. The risk sat entirely with Intercom, which made adoption easy.

Betting on cost trajectory turned losses into margins. McCabe priced for where operational costs were going, not where they were. The short-term loss bought market share that paid off once costs dropped.

The results: nine consecutive quarters of increasing net new ARR, Fin growing at 393% annualized, and valuation up at least 10× since McCabe returned.

Key Takeaway

If you can see where costs are heading, price for the future. Outcome-based pricing removes customer risk and accelerates adoption. The founders who move fast and bet on trajectory will own their markets while competitors optimize for day-one margins.

— Eoghan McCabe, Intercom CEO

Weekly Challenge

One Experiment. One Week. One Win.

The Goal

Enable one way for users to share something they’ve created inside your product.

How It Works

Start by identifying the most valuable output your power users create, whether that’s a template, workflow, prompt, preset, or configuration.

1. Build the simplest path for them to share it (public link, gallery submission, community post, or export function).

2. Reach out to five power users and ask them to share something they’ve made.

3. Track how many new users discover your product through those shared assets.

Why It Works

Every shared template is a landing page you didn’t have to build. Users who contribute feel ownership and stick around longer. The content they create often covers use cases you never thought of, expanding your product’s reach without expanding your roadmap.

Spotlight

Share your user contribution system and early results in the TAAFT community by end of week. We'll showcase the most creative implementation in an upcoming newsletter.

AI Market Watch

Deals, Discoveries, and Demand

Megadeals

• xAI – $20B Series E (generative AI)

• Etched.ai – $500M round (AI chips)

• Skild AI – $1.4B Series C (robotics AI)

• Cyera – $400M Series F (AI data security)

• Merge Labs – $252M seed (brain-computer interfaces)

• Onebrief – $200M Series D (AI military mission planning)

Top Research

• Mind Your Tone (how being rude affects LLM accuracy)

• Argos (multimodal RL verification for grounded reasoning)

• CooperBench (why coding agents can’t be your teammates yet)

• Context Engineering 2.0 (framework for human-AI context management)

• LLM Inference Hardware (Google research on architecture opportunities)

• SciSciGPT (multi-agent AI research collaborator, published in Nature Computational Science)

Search Trends

Top “best AI for…” searches:

1. Customer service

2. Coding

3. Research

4. Image generation

5. Video editing

Thanks for reading!

Got a project in motion? Hit reply and share what you’re working on with us.

Keep an eye on your inbox. We’ll be back soon with more founder stories, fresh tools, and data worth your time.

Signing off,
— AI Empires