vøiddo is a fully remote six-person studio. four of us live in israel — tel aviv, jerusalem, haifa — and two of us live in estonia — tallinn and tartu. the company is legally registered in tallinn, estonia, because estonia made it easy to incorporate without flying anywhere and israel made it complicated; the legal entity is estonian, the people are wherever they actually opened a laptop in the morning. we run on one rented vps in a hetzner frankfurt datacentre, three founders worth of patience for finicky tooling, and a postgres database that is older than two of our marriages. we ship ai-flavoured tools, browser extensions, wordpress plugins, dev utilities, and a small library of browser games that are still in active development.
we don’t pitch investors. we don’t run waitlists for products that already exist. we don’t put a credit card in front of the free tier. we don’t do the "book a 30-minute discovery call" thing. if you came here looking for a series-a deck, you came to the wrong studio — we have, instead, a paddle account, a support inbox, a refund policy you can read in thirty seconds, and a public team page with everyone’s actual email.
the studio started in 2024 because the founder, egor michurin, had a small shopify side project and ran out of patience for writing product descriptions by hand. the obvious shortcuts did not work. generic ai writers produced copy that sounded like every other dropshipping store, the dedicated "ai listing tools" were either amazon-only or priced for businesses that pay a marketing agency a five-figure retainer, and the open-source options were either abandoned in 2022 or running on outdated llm apis. there was a real, narrow gap. a useful product description engine for the small seller. nobody had built one that did not feel like a tax to use. so somebody did.
the first version of scrb shipped over a long weekend in tel aviv. one fastapi process, a single postgres table, a brittle wrapper around an llm api, no front-end, no auth, no rate limiting, and a curl one-liner in the readme. it was ugly. it would not pass any kind of code review. but it shaved roughly five hours off the seller’s weekly workflow on the first weekend it shipped, and the second weekend it shaved closer to seven, and by the third weekend other shopify sellers in a small private discord were asking how it worked. that is when we knew there was a product, not a script.
the test we still use is the same one. if a tool saves us time on something we actually do, we ship it and find out whether anybody else needs it. if it does not, it stays a hobby project and we move on. about four out of five things we prototype never become products; the ones that do tend to start as something that solved a real problem for the founder, the qa lead, or a customer who emailed in. nothing in the catalogue exists because we read a market report.
between 2024 and 2026 the studio went from one script to a portfolio that is small but no longer accidental. scrb (web app, chrome / firefox / edge extensions, wordpress plugin) is the first flagship and still the largest revenue line. rankd (the ai culture judge, web app and chrome extension) is the second. tells (text-first analysis for reading what people don’t say out loud) went live on paddle in may 2026 and is the third. on the side there are around 68 free dev utilities at tools.voiddo.com, eleven browser extensions across chrome / firefox / edge, eight browser games in active development at games.voiddo.com, 36 npm packages under the @v0idd0 scope, and four wordpress plugins in the wp.org directory. one person became six. the operating philosophy stayed the same.
we said no to venture capital from the first conversation, and we will say no again. outside money is not free; it buys outside timelines, outside priorities, and an exit clock you cannot stop. bootstrapping is slower, and a bad month genuinely hurts, but it leaves the question "what should we ship next?" entirely up to the six people whose names are on the team page. so far that trade has been worth it. we make less money than we would on a series-a; we sleep through more nights than founders we used to know.
three flagships do most of the talking. scrb writes ai product descriptions for amazon, etsy, shopify, ebay, and woocommerce sellers. paste a product photo or a competitor link, get listing copy in twenty-five languages that does not smell of chatgpt, does not start with "in today’s fast-paced ecommerce landscape," and does not end with "let your customers experience the difference." it is live as a web app at scrb.voiddo.com, as a browser extension on chrome / firefox / edge, and as a free wordpress plugin in the official directory. pricing is human-readable: free 5/month, pro $9.99, business $24.99. cancellation is one click. there is a one-time $19 "listing repair pack" for sellers who have a single stale page that needs surgery rather than a subscription.
rankd is the ai culture judge. name a film, a game, a book, a song, a name, a product, a comma-separated list of arguments — get a letter grade, a numeric score, and one short opinion in roughly four seconds. no fence-sitting, no plot summary, no "it depends on what you like," no rambling preamble about how art is subjective. the page does not pretend the verdict is anything other than the verdict; you are free to disagree, and a lot of people do, which is how we know it is working. it lives at rankd.voiddo.com and as a chrome extension. the creator embed lane lets people drop a rankd verdict card into a blog or substack post.
tells is the third flagship and the most personal. it reads the things people do not say in plain text — read a message, read a person across their writing, read the gaps inside a social profile. v1 went live on paddle in may 2026; free tier is five reads a month, the starter plan is $14.99, pro is $34.99, and there is a $79.99 forensic tier for the cases where somebody wants to argue with the read. twelve languages, native (including hebrew, arabic, and russian, not google-translated). you can open it at tells.voiddo.com if that paragraph made any sense to you; if it did not, the homepage walks through a sample read.
around the three flagships are the supporting tools and games. about 68 free dev utilities at tools.voiddo.com — regex testers, json formatters, jwt decoders, cron explainers, signal-reading tools, the kind of one-purpose things that should always have been one click away and instead are routinely buried behind a sign-up wall on the third-page result of a google search. eleven browser extensions across chrome, firefox, and edge: some free, some paid, none of them request permissions they don’t use. eight browser games at games.voiddo.com — all currently in active development, all with open waitlists, none of them currently playable end-to-end. and 36 npm packages under the @v0idd0 scope, mostly utilities the team needed for internal scripts that we then made public because there was no good reason not to.
nothing in the catalogue exists because someone told us it would "rank well" in search. every product is something at least one of the six of us actively uses, and most of them are something a paying customer asked for. if a product is not pulling its weight, it gets archived; if a product is being used heavily by free users and nobody is upgrading, we either rework the paid tier or sunset the product. there is no graveyard of "we’ll get back to it" pages. the public page count is the same as the internal page count.
the actual rules we use when we disagree about what to do next.
code on a laptop helps nobody. we push to production as soon as a feature is genuinely useful, even if a few rough edges are visible, and we iterate based on what real users actually do rather than what we imagined they would do over an aeropress. that means our first deploy of a feature is usually uglier than a competitor’s announcement post; it also means we know whether the feature works by the end of the week instead of the end of the quarter.
most product debates around here die quietly when somebody actually opens the server logs. we look at error rates, request volumes, conversion funnels, support email patterns, and the support inbox itself. the data is rarely flattering and almost always settles the argument faster than the loudest voice in the slack channel can.
six people and a lot of careful automation is the whole company. that number is on purpose. a small team forces small meetings, sharp scope, aggressive defaults, and a culture where the engineer who would be in three syncs at a series-a startup is, instead, finishing the feature. we will resist growing the headcount for as long as the revenue allows it.
we deploy multiple times a day, in changes you can read in one sitting without scrolling. when something breaks — and it does, regularly, because we are six people shipping daily — rollback is one button on the deploy box and a single line in a slack channel, not a postmortem-shaped meeting that takes a week to schedule. every feature can be reverted in under two minutes; if it cannot, we do not deploy it.
we would rather lose the sale than oversell the tool. if scrb is wrong for your workflow, the support email will tell you so and, where we know one, will recommend a competitor. if rankd cannot judge a thing well, the verdict says so out loud. if tells does not have enough context to read a message, the read says "not enough text," not "based on this single sentence we infer the relationship is in crisis." people remember who did not lie to them.
we use deliberately boring, well-understood tech — postgres, redis, fastapi, react where it pays, vanilla js where it does not, nginx, systemd, paddle. the surprises are reserved for the products themselves. nobody is paged at 3am because we wanted to try a hot new database; nobody is debugging a service mesh on a saturday because we read a conference talk; nobody is rewriting the payments rail because somebody on hacker news said paddle was uncool.
most of what passes for "ai-powered saas" in 2026 is the same growth-stage playbook on a slightly different homepage. a sign-up form that asks for a credit card to see the product. a "free trial" that is fourteen days and converts automatically. a free tier with a 1-of-5 watermark on every output and a "remove watermark" upsell. dark patterns around cancellation that route you through a five-step "are you sure" funnel. a "book a call" button instead of a price. a series-a deck somewhere in the founder’s drive that values the company at sixty times revenue. the people writing this software are optimising for the next funding round and the eventual exit, and you are the input metric.
we do not do any of it. there is no credit card on the free tier of scrb, no credit card on the free tier of rankd, no credit card on the free tier of tells. there is no watermark on free outputs. the pro tiers are flat-priced and one click to cancel; the cancel button is on the account page, where you would expect it, with no "are you sure" modal designed by behavioural economists. the pricing page is a pricing page, not a maze. there is no "book a call to discuss enterprise pricing"; the prices are the prices.
refunds go through paddle, which is our merchant of record. that means a refund request gets processed the same day you ask, not in a support thread that takes nine business days and routes through four levels of escalation. it also means we never see your card number, your full address, or your tax id, which is the way it should be.
none of this is a moral position; it is the product working for the user instead of the user working for the product. it turns out a lot of people prefer it that way, and they tend to come back next time they need something we make.
the team is six people working fully remote across two countries: israel (tel aviv, jerusalem, haifa) and estonia (tallinn, tartu). nobody commutes anywhere; nobody shares a postcode with a colleague. there is no office, there has never been an office, and nobody on the team has expressed the slightest interest in renting one. the four israelis meet in the same room twice a year and the two estonians do the same in tallinn; the whole six lands together once every eighteen months or so, usually around a conference one of us happens to be speaking at. between those moments the work happens in pull requests, logs, slack channels, and short calls that end as soon as the decision is clear.
asynchronous by default. we document everything in markdown, ship pull requests in english, review in english, and let the time zones work for us. when frankfurt is asleep, tel aviv is at lunch and tallinn is finishing breakfast. when tel aviv is asleep, somebody in tallinn is usually awake enough to notice the deploy alert. we run one weekly video standup that nobody is required to attend if they have nothing to say; everyone signs in to a tuesday-morning written status update that lives in a shared notion page, and the same page archives the decision log for the year.
most product decisions are made by whoever is closest to the user. tal does not need a meeting to flip a feature flag; lev does not need a meeting to add a fuzz test. when something is contested, two people get on a call; when something is genuinely strategic, four people get on a call and one of them inevitably mutes themselves to make coffee. founders do not unilaterally override engineers, engineers do not unilaterally override designers, and the design lead has, on more than one occasion, vetoed a launch because the spacing was off.
the team page (/team/) has the full crew with hand-illustrated portraits, long bios, and the occasional anecdote about who shipped a hotfix from the bathroom of a wedding reception. we are not actively hiring full-time, because six is the size we have decided to be. if you build interesting things and want to say hello without a hiring agenda, the contact form works; we read everything and reply to most of it within a working day, unless somebody’s kid has chickenpox.
we pick boring tech on purpose. postgres for everything that has to be true tomorrow, including yesterday’s billing rows. redis for everything that can be wrong for thirty seconds without breaking a paying customer’s day. fastapi for the api surface because it is dull, fast, and refuses to be opinionated about a request body. react for the handful of front-ends that genuinely benefit from it; vanilla js for everything else, because every team that "just adds a framework" ends up maintaining the framework instead of the product. nginx and systemd are in front of all of it because they have been in front of all of it for fifteen years and have outlived three rounds of cool-kid replacement candidates. paddle handles money, vat, and chargebacks. one vps in frankfurt, one operator on call, one rollback button, one postgres database that everything talks to.
things we deliberately do not run: kubernetes, a service mesh, a feature-flag service we did not write ourselves, a vendor observability suite, any data warehouse priced per gigabyte, a separate "ai vector store" beyond what postgres already does well enough, and any messaging queue that is not redis. we have read the conference talks on why we should. we remain unconvinced. the boring stack is what lets six people run sixteen surfaces without anybody being on-call for more than one product at a time.
a more honest breakdown lives at /studio/#stack — including which editor we actually use, which ai model we mostly route through, and which ai model we will not name in public for our own reasons.
vøiddo is a private limited company legally registered in tallinn, estonia. the legal address on every paddle receipt, every dpa, every invoice we sign, and every refund email is the estonian one; that is the address you write to if you need to send a paper letter for any reason, which has happened roughly twice in the lifetime of the studio. we picked estonia because the e-residency programme made it possible to incorporate without flying to a notary, and because the corporate tax setup made it possible to keep the operating overhead in single digits as a percentage of revenue. the legal entity is estonian. the team is wherever the team actually lives.
payments and tax compliance run through paddle (dublin, ireland, merchant of record). this means customers in roughly eighty countries get a local-currency receipt with the correct vat line on it without us ever touching a single tax form. it also means refunds are a one-line email and a same-day reversal, not a six-week support ticket. uptime is monitored from frankfurt, london, and singapore.
we are not registered in israel as a legal entity. we are not registered anywhere else either. the team simply lives in two countries; the company itself lives in one.
there is no support queue with a fifteen-business-day reply target. there is no ticket number and no automated bot replying with "we have received your message." write to support@voiddo.com and a human on the team reads it the same day, usually within a couple of hours during european or israeli working hours. for press, brand assets, factsheets, the founder’s schedule, or a question about the estonian legal entity, the same address works. there is no separate "investor relations" inbox because there are no investors.