I built Woodshed because I had more ideas than hours.

I think a lot about the intersection of code and prose, and how building software has changed forever in this moment we’re in.

I felt I was leaving too many ideas on the table.

Not because they were bad. Because I’m one person.

So I built a creative studio out of AI agents. Three of them: @MotifAgent, Cadence, and Forte are all running in one Slack workspace. A woodshed.

One strategizes. One writes. One builds.

They remember what I care about, read what I’ve written, and show up every morning with their own ideas before they ask for mine.

In jazz, woodshedding means locking yourself away to practice, obsessively, until the music lives in your hands. No audience. No shortcuts. Just the work. I took that idea and applied it to agents. Build in private, iterate relentlessly, ship when it's ready.

The platform they run on is called @openclaw. It lets you give an AI agent a personality, a memory, and jobs to be done.

What happens when you give your half-formed ideas to a team that never sleeps? If revenue comes from that, great. But that’s not why I check in on the Woodshed crew before I check my email.

@steipete, who created OpenClaw, put it perfectly: it’s hard to compete with someone who’s just having fun.

This is now my creative playground. The woodshed is where musicians, and now agents, go to practice when nobody’s watching.

— Daniel

Either write things worth reading or do things worth writing.

—Benjamin Franklin

Why not both?

Every was founded by Dan Shipper and Nathan Baschez back in 2020.

It started as The Everything Bundle, a collection of newsletters on Substack.

Key milestones over the past five years:

• Published 1,500 essays

• Reached nearly 4,000 paid subscribers.

• Hit $60,000 in MRR and over $3 million in total revenue.

• Incubated three software products.

• Launched courses on writing, programming, and psychology.

• Built a full-time team of seven.

Every raised a pre-seed round from Eric Stromberg but they’re primarily self-funded. Well, until last week when they announced at $2M round led by Reid Hoffman.

In 2023, Nathan spun out Lex (Cursor for Writing) as a separate company, which has experienced significant growth recently.

Last year, Every ventured into consulting, which has been so successful they’re overwhelmed with requests and asking for help.

One key thing that shouldn’t be overlooked is their taste.

It’s a clear strength.

Their design aesthetic and UX sensibilities are impressive.

How should we define Every?

• Media Company?

• Bootstrapped Startup?

• Writer Collective?

Dan Shipper recently spoke with venture capitalist Nabeel Hyatt about the challenge of explaining what the company is.

I feel like I've been grasping for words for how to describe what Every is for a while. I think that's a really interesting place to be because we're doing something that's working and there isn't yet a word for it. The closest thing I've come up with is multi-modal media company. — Dan Shipper

The absence of a clear definition for Every suggests they’re probably predicting the future by inventing it. When organizations operate at the frontier of innovation, they often defy easy categorization. This ambiguity can actually be a sign of genuine novelty rather than confusion or lack of focus.

Think, Build, Ship

Every Studio, their in-house incubator, is fueled by generative AI. The Studio supports a group of Entrepreneurs-in-Residence (EIRs) to think, build, and ship products that solve problems close to home. For each idea, they’ll do an Alpha release internally and if it passes the sniff test, its released to the public.

Some thoughts from Dan:

Our EIRs are building products that they want to use themselves. They’re solving problems that they’ve encountered living in the future.

We’re entering an era of gritty startups where rapid experimentation and distribution to a dedicated audience is the most exciting way to build new businesses.

Can they scale?

In a world where you can build and launch a prototype in an afternoon, scale can be an afterthought. We’ve created a model with optionality at its core—letting an idea become its best expression, whether as an experiment, a cash-flow business, or the next Apple.

How much do all of their products cost?

Every-thing is included in a $20 monthly subscription.

I’m bullish on this model. My investment thesis is predicated on a future powered by creators, collectives, and guilds.

A virtuous cycle

Every operates on a self-sustaining loop.

• Podcast interviews support their newsletter.

• The newsletter is an owned distribution channel for launching products.

• Products drive more newsletter subscribers, who then subscribe to their podcast and share their essays.

Many startups struggle with customer acquisition, but Every leverages its existing audience like kindling for a fire.

That’s how you get 10,000 signups in a month for an email app.

And 25,000 signups for writing app.

They’ve earned a ton of trust from years of publishing thought provoking essays.

That trust is paying handsome dividends now.

Building products inside of a beloved media company is an unfair advantage.

How Every could fail

I see three structural challenges that could derail Every's model:

1. The Talent Paradox

Every's model creates a unique talent retention problem. By design, they attract entrepreneurial stars and help them build valuable products. But this success creates natural exit pressure:

• EIRs who build successful products have strong incentives to spin them out completely or leverage their success to raise money for a new venture

• The more valuable the product becomes, the stronger these incentives grow

• Unlike traditional companies where success aligns everyone's interests, Every's success can actually accelerate team departures

This isn't just about compensation, it's about ownership and control. Even generous profit-sharing may not compete with the pull of full autonomy.

2. Shiny Objects

Every's reputation rests on maintaining high editorial standards while they fall down software rabbit holes. This creates competing pressures:

• Editorial excellence requires deep focus and careful curation

• Software development with AI feels like a slot machine that’s hard to step away from without discipline

• Chasing too many shiny objects could be a distraction

3. Consumer Fatigue

While Every's media platform provides powerful distribution for new products, this advantage has hidden costs:

• Each new product launch taxes their audience's attention

• Failed products risk damaging the invaluable trust they've built

• Users might grow weary of more micro-apps that solve only one problem

These fundamental tensions in their business model will require careful navigation. The real test isn't whether they can launch products quickly, it's whether they can preserve what makes Every special.

Fin

As AI tools continue to evolve, we'll likely see more companies following Every's hybrid model. Leveraging content creation, community building, and product development in symbiotic ways.

The future of startups might not belong to unicorns chasing hyper-growth, but to companies building delightful tools while maintaining independence and creative control. A collective of writers, engineers, and designers that can manage a portfolio of software companies under one roof.

Every kind of reminds me of an indie rock band. Their success isn't built on mass-market appeal or huge marketing budgets, but on authenticity and a distinctive voice that resonates with an audience they understand intimately.

Indie bands succeed by ignoring conventional wisdom. Every has certainly sidestepped traditional startup playbooks. They've built their following through consistent creative output rather than growth hacks or viral marketing.

Every's success isn't threatened by others copying them. Their competitive advantage lies in their distinctive voice and sensibility. This can't be replicated with capital or technology alone.

Indie artists can thrive by cultivating a dedicated fanbase who'll buy each album, attend every show, and support their artistic evolution. Every has over 10,000 true fans who will back them not just because of their ideas, but because of their taste.

Writing is the most important skill in the agentic age. A collective of writers, paired with engineers and a brilliant designer, is a scary good group to build with.

— Daniel

I appreciate your feedback. If you enjoyed this piece, drop a like or comment below. You can also send a note to daniel@indiethinkers.com, would love to hear from you.

Developers, and now vibe coders, face empty black screens.

Both writers and developers encounter the same fundamental challenge.

It’s hard to translate nascent thoughts into something meaningful.

Knowing how to translate your ideas into prose (and running code) is only going to become more valuable. There's still friction here though because, well, writing is difficult. What makes the process easier?

Effective prompts.

Not from you but questions from a system directed at you.

We need to simulate organic conversations.

There's a reason your brain is flooded with ideas after discussing a topic you're passionate about with a friend.

Great ideas always emerge from great conversations.

It's easier to brainstorm when presented with a question.

Better yet, a series of questions that adapt based on answers provided.

Your imagination struggles in an empty space. You have too many choices and micro decisions to make. Analysis paralysis shows up. Great questions are guardrails on your roaming mind, and these constraints can spark creativity.

What if all you had to do for a first draft or a minimum viable product was engage in a conversation that closely matched human dialogue?

Where Current AI Development Tools Fail

Add the prompt below to Cursor, Windsurf, or Bolt.

It’s intentionally not that great (more on this later).

"Help me build a product requirements document for a Hacker News clone. Start by interviewing me."

The response will be something similar to this:

I'll help you create a product requirements document for a Hacker News clone. Let me ask you some key questions to understand your vision better.

• First, what are the core features you want to include in your Hacker News clone? The original Hacker News has:

• Story submissions

• Comments

• Upvoting

• User accounts

• Basic moderation

Would you like to keep all of these, or are there specific features you want to modify or add?

• What's your target audience? Are you building this for:

• A specific community/niche?

• A general audience?

• A particular industry or topic?

• Do you have any specific design preferences or requirements?

• Keep the minimalist HN style?

• Modern UI/UX?

• Mobile-first approach?

• What are your scalability requirements?

• Expected number of users?

• Expected number of posts/comments per day?

• Any specific performance requirements?

• Are there any specific features from other platforms (like Reddit, Twitter, etc.) that you'd like to incorporate?

Please share your thoughts on these questions, and I'll help you create a comprehensive PRD based on your responses.

How many questions did you count?

The questions are fine, but the UX is terrible.

There are way too many questions presented at once. In a real life conversation, a friend wouldn't drill you like this. It's overwhelming and kills momentum.

What these tools need is a refined interview mode that:

• Asks one focused question at a time

• Adapts follow-ups based on previous answers

• Offers suggestions to assist users who are stuck

• Gradually builds complex structures from simple responses

There needs to be an "interview mode" here.

And here.

We need single‑question progressive interviews baked into these tools.

Think of your mind like a juggler.

There's a natural limit to how many balls can stay in the air at once. When Cursor bombards you with a dozen questions simultaneously, it ignores fundamental principles of human cognition and conversation.

Interview Mode

The most effective interviews, whether conducted by humans or AI, respect four key dynamics of information processing:

1. Cognitive load: Humans comfortably juggle 2–4 chunks of info. Twelve simultaneous questions is a bit much.

2. Incremental clarity: Each answer refines the mental model before the next decision.

3. Socratic momentum: Each question sets up the next, preventing analysis-paralysis.

4. Error-correction loop: Tiny steps surface misunderstandings early, saving thousands of wasted tokens.

Here’s a refined interview mode:

We need to rethink the interaction model between developers and AI-powered editors. It's not simply about having AI ask questions, it's about creating a continuous flow state where each exchange builds momentum.

When properly implemented, CDD doesn't just avoid blank pages, it changes how we externalize and refine complex ideas.

Me: I want to build a tool that analyzes my writing and identifies overused phrases.

Agent: That's an interesting project. What type of writing do you want to analyze first?

Me: Mostly blog posts and essays.

Agent: So we're focusing on longer-form content. What would you consider "overused" in your context? Number of occurrences, proximity, or something else?

Me: Probably words or phrases that appear multiple times within the same section.

Agent: That makes sense. Would you want to detect exact matches only, or also similar phrases with slight variations?

Notice what's happening:

• Each question addresses a single decision point

• The agent acknowledges my answers before proceeding

• Questions build on previous responses

• When I reveal uncertainty e.g. “probably” the agent explores options instead of demanding a definitive answer

This creates a natural progression that produces a more accurate spec.

Going back to my Hacker News Clone, here's a more effective prompt:

"Help me build a product requirements document for a Hacker News clone. Start by interviewing me. Only ask me one question at a time. I might want to ask for recommendations based on each question if I don't have an answer."

The first question is the most important.

What are you trying to accomplish?

More importantly, why are you building this?

You can safely say I don’t know, which would kick off a generative conversation.

The Future of Creative Work

The most successful tools won't just offer powerful models, they'll transform how we approach the blank page problem.

By embracing conversation-driven development, we can make the software development process more natural, productive, and human.

Whether you’re building software or drafting an essay, there is so much power in a great question and conversation.

Wasting time sucks.

You can burn thousands of tokens going down the wrong path.

I've done this countless times. Prompting my way to an underwhelming result.

When this happens, it's easy to blame the tool or "AI" for not being worth your time.

It's likely that you should have just asked for an interview.

— Daniel