Founder Playbook · The Bootstrapped Founder
12 tactics from Pieter Levels
Pieter Levels — Indie Hacking is Dead. Now what?
Watch the full episode“Indie hacking is dead and then the I try to explain... if remote work was called remote hacking you know and then Co happened and remote work became normal it was it would also be dead because it's not hacking anymore it's just normal now... it's become more competitive more saturated and more difficult because you're competing not just with a lot of India hackers now... you're also competing with big tech companies now”
Indie hacking isn't hacking anymore — it's mainstream and competitive
Like remote work after COVID, indie hacking has stopped being a subculture and become a default path. Big-tech teams now monitor indie founders' Twitter to copy features within weeks. VC-funded founders are quietly switching to indie for their next venture. The hack has become the standard — which means the easy advantages are gone. Expect more competition, less novelty, and adjust strategy accordingly.
“everybody has say problems retention which is the turn is very high um and defensibility like you make something and immediately you have a lot of clones because everybody's working with the same stuff everybody's using gp4 uh for Tech stuff almost or llama... within I would say if you're smart within three months with if you're less smart maybe six months to a year so it's very difficult to make a startup AI startup now”
AI startups have structural moat problems — assume profit margins collapse
All AI startups share the same underlying tech (GPT, Llama, Stable Diffusion), so clones arrive within 3-6 months and price competition pushes margins toward zero. Model the AI business as a high-revenue, low-margin, short-window play — not a long-term moat. Pieter explicitly tracks AI startup profit margin separately and accepts it will keep falling; the strategy is harvest then sell, not compound.
“the cost of fine tuning was $3 and then when they saw my tweets where I was sharing my Revenue uh that I was making so much money with this they said sorry we need to increase the price to $20 per training... so you have to go one level higher so you have to get your own uh virtual private server with gpus you know but then you need to learn the code”
Suppliers raise prices when they see your public revenue — abstract dependencies from day one
When you publicly share revenue and your supplier sees the numbers, the price you pay almost always goes up. Pieter's fine-tuning costs jumped 7x once his Avatar AI revenue tweets went viral. Build with an abstraction layer (config-swappable provider) from day one so when a supplier raises prices you can migrate within a week instead of being trapped. Also: weigh the cost of being publicly transparent against the price discovery you're giving suppliers.
“I talked to one influencer he posted about photo.com my AI photo startup and the um Mr went from 12K to like 40 or 50k it's insane... and it stayed that way uh and of course it was sure so I also worked in it but this shows you have a really big effect of these influence and it's way bigger effect than press like I also got a lot of press for these AI startups and this press does almost nothing”
One trusted influencer can dwarf all the press combined
Pieter's Photo AI jumped from $12K to $40-50K MRR — sustained — after one creator post. Comparable press coverage produced almost no measurable lift. The lesson: redirect PR budget and effort toward identifying 5-10 trusted creators in your niche and earning their genuine endorsement. Logos on a Featured-In strip don't convert; one creator's personal recommendation can be a sustained MRR step-change.
“we always do like Tech swap cuz he sees my photos that suddenly look much better on Twitter my AI photos and he ask me man what are you doing now and I and we give each other hints so we say like man maybe look at this feature and try this... but we're not competing because he's doing head shots and I don't want to do head shots”
Befriend your direct competitor — swap tactical notes and both grow faster
Pieter and Danny Postma run overlapping AI photo products and DM tactical notes regularly. The trick: niche differently enough (Danny = headshots, Pieter = general photo studio) that you're not zero-sum, then share what's working. Both grow faster than either would alone. Identify 1-2 competitor-peers, separate the niche cleanly, and exchange honest specifics. The market is bigger than either of you, and the relationship is its own moat.
“it's not about PHP jig it's more about the point that it doesn't matter that you have these developers who work in Enterprise in agencies and there's this agency MLM I feel like where um you have a you have a company that doesn't know anything about tech they come to a web agency... these web agency need to sell like the best stuff so they say we use the newest technology”
Ignore the tech-stack cult — ship with what you already know
Pieter runs $40K+/mo businesses on PHP and jQuery. The pressure to adopt the newest framework comes from agency sales narratives and VC-funded framework companies needing evangelist growth — not from your customer. Ship with the stack you already know fast and validate. If your business doesn't survive long enough to need rewriting, the framework choice never mattered. If it does, rewrites are a great problem to have.
“Avatar ID and this was man it wasn't even code it was just a index of HTML actually was just a page with have examples of the avatars... link to type form no stripe checkout stripe payment link... I would manually so I immediately I had like 100 orders so manually I did I think these 100 or 200 orders myself so I would download the photos and then I would go to this this uh platform to do this finetuning”
Validate with a landing page + Stripe link + manual fulfillment — code comes after the first 100 orders
Pieter launched Avatar AI as a single HTML page with examples + a Stripe Payment Link + a Typeform for uploads, then manually fulfilled the first 100-200 orders by hand. No backend, no auth, no automation. Earned ~$3-6K in one night doing the work himself, then automated only the steps that proved demanded. Code is the most expensive form of validation; do the work manually until the volume forces the automation.
“I repeat myself all the time and then if I repeat myself 10 times like you know don't repeat yourself as the Mantra then I write a function but I don't people like people try immediately write a function for something you repeat twice it's like man just tweet repeat it TW you know like this kind of stuff”
Don't extract functions on the second repeat — let it repeat 10 times first
Standard DRY advice tells you to extract a function the second time you repeat code. For a solo indie hacker pre-PMF, that's premature optimization. Pieter's rule: let it repeat 10 times before you build the abstraction. By repetition 10, you know what the shape of the abstraction actually needs to be. Earlier than that, your function design will be wrong and you'll refactor it twice anyway. Validate the pattern first, abstract once.
“beautiful design is like a red flag for me you know these beautiful gradients and these now you have these borders that move like it's fancy it's so fancy if it's too fancy it means you spend too much time on design or you it's some VC starter that spent too much money on designers if it's the beginning if it's not validated yet... I prefer a very ugly web page in the beginning that just uh cuz man look at Google look at the beginning it was very ugly look at Facebook the first page was very ugly”
Beautiful pre-launch design is a red flag — ugly + working signals validation
When Pieter scrolls Product Hunt and sees a launching site with animated gradients and pixel-perfect everything, he assumes the founder hasn't validated yet — the time went to design instead of customers. Google's and Facebook's original pages were ugly because the founders were optimizing for whether the thing worked. Until you have paying customers, beautiful design is leverage you spent on the wrong axis. Ship ugly; polish after $10K MRR.
“I got them valued recently and the multiples for iars are very good like they're very they're quite high like um normally for like the any stars you get like two three X or something right for AI it can be like five or six or even eight because it's kind of hype now... you need to make you need to cut these cost rapidly and then you need to go to a broker and then you need to sell for like good good amount of money”
Sell AI businesses at 5-8x while the hype premium lasts
AI startups currently sell at 5-8x ARR multiples vs ~2-3x for traditional SaaS — a hype premium that will compress as the market normalizes. If you build an AI startup, plan the exit explicitly: get profit margins as high as possible by cutting infrastructure cost, then list with a broker (Acquire, FE) while AI is still hot. Treating it as a 3-5 year compounding business risks the multiple disappearing under you.
“I like to say opinions and then like about Frameworks or something and I say something and then I like to hear what people like when people reply I learn from that and I change my opinions like I have strong opinions weakly held I do change my opinions all the time but it's a stage it's like a Podium it's it's it's a show kind of it's inevitable that it becomes a show”
Hold strong opinions weakly — say what you think, update from honest replies
On social you're on a stage — 300K people aren't having a conversation with you, they're receiving a broadcast. Stage-honest is different from coffee-honest: stake out controversial positions in plain language because that's what cuts through, then update genuinely when the smart replies push back. "Strong opinions weakly held" is the operating mode that generates both reach and real learning. Hiding behind safe takes produces neither.
“I think they changed it more to like where if something goes viral in the beginning it becomes pumped maximum so it goes like before I would get like 100 retweets now if something viral gets a thousand retweets it goes very far but on the other side uh often many tweets get like two likes... what I love one thing interesting he also increased these long post like now you can write blog posts yes these blog post by definition get a lot of view time because of seconds right these do work really well”
Twitter's algorithm now pumps extreme + long-form — lean into blog-style posts
Twitter/X now measures view-time per tweet — extreme tweets get pumped to viral, normal posts get buried regardless of follower count. Adapt by writing long-form "blog" posts inside Twitter (which by definition rack up view-seconds) and clip them into shorter follow-up tweets. Product updates and ambient content underperform now; longer narrative pieces with a clear hook outperform. Treat Twitter as a publishing platform, not a status feed.