Headlime
Headlime is an AI copywriting tool: a web program that writes marketing text, such as advertisements, headlines and blog paragraphs, from a short description of the product. It runs on GPT-3, the text model that OpenAI made available to a first group of developers in 2020 through an API, an interface that lets one program send requests to another. Headlime adds the forms, the prompts that the founder tested against each other, an editor and a landing page builder, and it uses the DeepL translation service for other languages. The user never writes a prompt.
Danny Postma, a Dutch designer and marketer who by his own account could not yet program at the start of 2019, launched it from Bali on 29 June 2020 as a generator of headlines from 200 fixed templates. He sold it first as a lifetime deal, a single payment for permanent access. In the same month he asked OpenAI's chief technology officer for GPT-3 access by email and received it. He rebuilt the product around the model in five months and released that version in December 2020. He says monthly recurring revenue (MRR: the sum of all subscription payments in one month) went from US$1,000 at the start of December 2020 to US$20,000 in February 2021. On 30 March 2021 he announced the sale of Headlime to Conversion.ai, the company now called Jasper.
The price was never published. Postma says "seven figures", the interview site Failory titled his account "+$1M", and a report by TechPP on the day after the announcement says that neither side disclosed the terms. The figure of about US$1 million that many articles give matches a guess by another founder, Pieter Levels, who multiplied US$20,000 by 12 months and by 4. Every money number on this page is therefore grade C: a statement by the seller or an estimate by an outsider. No statement by the buyer about the price was found. The page also shows what happened to earlier small companies that sold marketing and writing software.
The numbers, graded
Each row shows the evidence grade and the source numbers from the list at the end of this page. The grades are explained in How Aiolum verifies.
| Fact | Value | Grade | How we know | Sources |
|---|---|---|---|---|
| Sale price | seven figures; the exact price was never published | C | Founder, in the Failory interview of January 2022 and in the title of his Indie Hackers thread of July 2021. Asked for the price in that thread, he answered only with zipped-mouth emoji. | [2, 1] |
| Lowest price consistent with the published statements | more than US$1,000,000 | C | Failory's headline for the founder's interview reads "+$1M". This page uses US$1,000,000 as its headline figure because it is the lowest value that fits that statement, not because anyone stated it as the price. | [2] |
| Outside estimate of the price, 31 March 2021 | about US$1 million | C | Pieter Levels, another founder, on X, marked by him as a guess: US$20,000 times 12 months times 4. It is not a statement by the buyer or the seller. Later articles give about US$1 million without naming a source. | [7, 6, 17] |
| The same estimate, recomputed | US$960,000 | calc | Calculated: 20,000 times 12 times 4 = 960,000. | [7] |
| Price terms according to the press | not disclosed by either side | C | TechPP, 31 March 2021, and They Got Acquired, 21 April 2022. The second outlet asked the founder for comment and got no reply. | [5, 6] |
| Buyer | Conversion.ai, later called Jarvis and now Jasper | C | Seller's announcement on the Headlime blog and on X. TechPP reports comments by the buyer's founder, Dave Rogenmoser, in Facebook groups. An announcement by the buyer itself was not found. Jasper's chief executive confirms the earlier company name in a 2022 interview that does not mention Headlime. | [3, 4, 5, 2, 16] |
| Monthly recurring revenue, February 2021 | US$20,000 | C | Founder, Failory interview. Levels wrote "$20K+ MRR" on the day after the announcement. | [2, 7] |
| Monthly recurring revenue, start of December 2020 | US$1,000 | C | Founder, Failory interview. December 2020 is the month in which the GPT-3 version came out. | [2] |
| Growth from December 2020 to February 2021 | 20 times | calc | Calculated: 20,000 / 1,000 = 20. The seller's announcement says revenue grew 100% month over month from November 2020. | [2, 3] |
| Yearly run rate at US$20,000 a month | US$240,000 | calc | Calculated: 20,000 times 12. | [2] |
| Price as a multiple of that run rate, if the price was exactly US$1,000,000 | about 4.2 times or more | calc | Calculated: 1,000,000 / 240,000 = 4.17. The real multiple is unknown, because the price and the revenue on the day of signing are unknown. | [2] |
| Sales in the first weeks after launch | more than US$60,000 | C | Founder, Failory interview: most of it from a limited lifetime deal. Not verified. | [2] |
| Lifetime deal tiers at launch | 200 at US$49, 200 at US$89, 200 at US$129 | C | Founder, Indie Hackers thread. | [1] |
| Revenue if all 600 lifetime plans in those tiers sold | US$53,400 | calc | Calculated: 200 times 49 + 200 times 89 + 200 times 129. This is less than the US$60,000 in the row above, so that figure must include other sales. The founder does not explain the difference. | [1, 2] |
| Subscription price steps | US$9, then US$29, then US$59 a month | C | Founder, Indie Hackers thread. He says he kept raising the price to find the highest price that customers accept, and growth did not slow. | [1, 2] |
| Access to GPT-3 | requested by email from OpenAI's chief technology officer in June 2020 | C | Founder. In a 2024 interview he names Greg Brockman and says he was in the first batch of users. | [1, 11] |
| Time to build | about 3 weeks for version 1; 5 months for the GPT-3 version | C | Founder. The Failory interview says one month for version 1. | [1, 2] |
| Acquisition offers | 2, in February 2021 | C | Founder, Failory interview and a 2023 podcast. He names only the buyer. | [2, 12, 1] |
| Length of the sale process | about 1 week of due diligence, about 2 weeks for the contract, 1 to 2 weeks for transfer and payment | C | Founder, Indie Hackers thread. Due diligence is the buyer's check of the seller's information. The contract was an asset purchase agreement, by which the buyer takes over the product and its customers and not the legal company. The money was held in escrow, which means by a third party, until the assets were transferred. | [1] |
| Team | 1 founder, 1 copywriter, 1 customer support person | C | The About page. The founder says he worked on Headlime by himself; They Got Acquired counts "Founder + 2 Members". | [14, 2, 6] |
| Customers | "1,182+ companies" on the home page | C | Company statement. A reader of the founder's thread quoted the same number in July 2021, so the counter has not changed since then. | [13, 1] |
| Hosting cost | US$5 a month for the app server | C | Founder: one Digital Ocean app. He says GPT-3 was a major part of the monthly cost and does not give the amount. | [1] |
| Outside funding | none | C | Founder: he paid his costs from a freelance job of two days a week. Levels and They Got Acquired describe the company as bootstrapped, which means paid for from its own income, without investors. | [1, 7, 6] |
| Legal form | Dutch sole proprietorship | C | Founder, Indie Hackers thread. This is why the country on this page is the Netherlands, although he built the product in Bali, Indonesia. | [1] |
| Time from launch to the announcement of the sale | 9 months | calc | Calculated: 29 June 2020 to 30 March 2021 is 274 days, which is 9 months and 1 day. The founder says 8 months, which is the time from the launch to the offers in February 2021. | [8, 4] |
How to read this table: a grade C value is a statement by the founder in an interview, in his posts on X or on the company's website, a journalist's report of such statements, or an outside estimate that says so. A "calc" value was computed on this page from the rows above it. No value is grade A or B: no filing, payment processor or buyer has published a price or a revenue figure for Headlime.
| Plan | Per month, billed yearly (US$) | Per month, billed monthly (US$) | Limit |
|---|---|---|---|
| Individual | 59 | 1,500 credits a month, 1 user, no white label | |
| Business | 399 | 3 members, unlimited use under a fair-use rule, white label | |
| Extra credits | 10 one-time | 200 credits |
The pricing page shows no yearly plan, so the yearly column is empty. Credits are used only for AI generation. The page says that 1,500 credits are about 25 blog posts or 75 landing pages, and that extra credits cost US$10 for 200. White label means that the customer can show the tool under its own brand. The Business price is for 3 members; a slider on the page goes up to 100 members, and the prices for more members load by script and could not be read. US$59 a month is the price that the founder says he reached before the sale, after US$9 and then US$29. The site has had no new features since March 2021, and this page did not test whether a new subscription can still be bought.
The timeline
From a US$19 ebook of headline templates to a sale to a competitor, nine months after launch. Each entry carries its grade.
- 2015CPostma, then 21 and a freelance conversion specialist (a consultant who raises the share of website visitors that buy), starts Landingfolio, a gallery of landing page designs. He says it made no money for four years. [2]
- April 2020CIn lockdown in Bali he lists products that he could build quickly. One is a tool made from an ebook of headline templates that he and a friend had sold for US$19. [1, 2]
- June 2020CHe emails OpenAI's chief technology officer to ask for GPT-3 access and gets it. [1, 11]
- 29 June 2020CLaunch on Product Hunt and X as a headline generator with 200 templates and a limited number of lifetime plans, on the domain headlime.io. It is second product of the week on Product Hunt. He says the first weeks brought more than US$60,000. [8, 9, 2]
- July to November 2020COther people copy the template tool. He spends five months rebuilding Headlime around GPT-3. [2, 12]
- 24 October 2020CThe archived home page still sells templates for headlines, emails and Facebook ads, does not mention AI, and is signed "Creator of Headlime.io". [15]
- December 2020CVersion 2 with GPT-3 comes out. Monthly recurring revenue is US$1,000 at the start of the month. [2]
- February 2021CUS$20,000 in monthly recurring revenue. Two companies offer to buy Headlime. He says he was close to burnout. [2, 1]
- 17 March 2021CA TechCrunch writer tests Headlime for an article on GPT-3 startups and writes that he is impressed. [10]
- 30 March 2021CPostma announces on X and on the Headlime blog that Conversion.ai has bought Headlime. He takes a strategic role at the buyer. Headlime will get bug fixes but no new features. [4, 3]
- 31 March 2021CTechPP reports that the price is not disclosed and that buyers of the lifetime deal keep Headlime but get no new features and not most of the buyer's product. Pieter Levels guesses a price of about US$1 million. [5, 7]
- 1 July 2021CIn a public question thread on Indie Hackers he describes the sale process and declines to give the price or the revenue. [1]
- 18 January 2022CFailory publishes his interview: "seven figures", and the buyer is now called Jarvis.ai. [2]
- 9 August 2023COn a podcast he says he sold Headlime because he was afraid to build a team, and that he now has the same choice with HeadshotPro, his AI headshot product, which has its own page on this site. [12]
- 13 October 2026Cheadlime.com still loads, with the 2021 acquisition notice, a 2020 copyright line, a sign-up page and prices of US$59 and US$399 a month. The help site does not respond. Whether new customers can still pay was not tested. [13]
Why it worked
Four causes, in the order they mattered. Each one is a reusable lesson.
Timing
Postma had a working copywriting product with paying customers in the month when GPT-3 reached its first developers. He did not wait for general access: he wrote to OpenAI's chief technology officer and was admitted. The GPT-3 version came out in December 2020, when few products of this kind existed. Three months later a competitor had raised close to US$3 million and, in his words, competitors appeared every week.
A lifetime deal first, then rising subscription prices
The first version was a simple tool that filled blanks in 200 headline templates. A limited lifetime deal in three price tiers brought, by his account, more than US$60,000 in the first weeks. During the five months of rebuilding he also had a freelance job of two days a week, which he says paid his costs. After the rebuild he sold subscriptions and raised the price from US$9 to US$29 and then to US$59 a month.
Showing the work in public was the sales channel
He posted progress and previews of the GPT-3 version on X while he built it. He names X, Facebook groups and word of mouth as the main channels, and says paid ads on Google never worked. He says the public posts led to the TechCrunch test. They also led to the buyer: according to They Got Acquired, the founders of Conversion.ai knew Headlime because they followed him on X.
Selling instead of growing the company
In February 2021 he saw three options: raise venture money and work 80 hours a week, stay self-funded and hire, or sell. He was doing support, development and marketing by himself and says he was afraid to hire. He sold to a competitor. The process took about four to five weeks from due diligence to payment.
How it was built
These details come from the founder's Indie Hackers thread and the Failory interview. They are grade C. [1, 2, 13]
- Front end in Nuxt (Vue), server in Node.js with Express, database MongoDB on MongoDB Atlas, hosting on Heroku and Digital Ocean. The app server cost US$5 a month.
- Text generation by GPT-3 through OpenAI's API, and DeepL for other languages. The pricing page lists 11 languages.
- Prompts chosen by A/B tests, which compare two versions on real users: he compared which prompt's results users marked as favourites, then tested the winner again.
- Interface parts mostly from Tailwind CSS components, to release faster.
- Version 1 took about three weeks. Version 2 took five months, and he says the last three were small bug fixes.
- Free use without a trial period. The user pays on reaching a paid feature, and limited-time offers by email move free users to paid plans.
- Code, documentation and help desk were written so that another company could take them over. Due diligence was one Notion document with the buyer's questions and his screenshots as answers.
- The company was a Dutch sole proprietorship. The sale was an asset purchase with payment through escrow.
Tools named in the sources
Each one opens its page in the Tools tab, with every story that uses it.
The dead that came before
Five companies in the Loot Drop database sold software that promised small businesses less marketing or writing work. The values below are from that database and are not independently verified.
Metigy
The promiseAn AI marketing brain for small businesses.
Why it diedSmall businesses wanted the work done for them, not recommendations. The money went into generic advice that few customers used.
Argyle Social
The promiseA social media marketing tool that saves marketers work.
Why it diedLarger companies with bigger marketing budgets won the customers while Argyle released small updates.
Marketfox
The promiseAutomated marketing for online shops.
Why it diedMailchimp and other large marketing automation companies had more money and more features.
Userfox
The promiseSimple automated email marketing for online shops.
Why it diedMailchimp dominated the market, and Userfox did not have the money to compete.
Draft
The promiseA minimal cloud editor for writers with version control.
Why it diedGoogle Docs covered the need for free. A narrow writing tool with a free plan could not pay for itself.
What changed
Each of the five sold a narrow marketing or writing tool, and in four cases a larger company with more money or a free product won the customers. Headlime was in the same position by March 2021. The founder wrote that competitors appeared every week, TechPP reported that Copy.ai had raised close to US$3 million that month, and other people had copied his first version completely.
Two things differ. First, the text was written by a model that none of the earlier companies had. Before GPT-3 a template tool could only fill blanks, which is what Headlime version 1 did. For some months in late 2020 the model was the advantage, because few developers had access. Second, Postma did not try to compete with the larger companies for years. He sold to one of them nine months after launch, at a time when he says revenue doubled every month. The cost of that choice is visible on headlime.com today: a product with no new features since 2021.
Can one person copy it now?
The early access, no. The sequence, yes.
Getting a new model before most developers was one email and some luck, and the founder says he is not sure it would work again. The sequence can be repeated: own a small paid tool with customers in one niche, put a new model into it as soon as an API exists, raise the price while growth continues, and decide early whether to raise money, hire or sell.
Where it is still open: Small paid tools with an existing customer group and one task that a new kind of model can now do. The same sequence applies each time a model for a new type of output, such as voice, video or structured documents, becomes available through an API.
- 1Start with a small tool for which you already have the content, and sell a limited lifetime deal to get the first customers and the first money.
- 2When a new model appears, ask for access directly and rebuild the tool around it, even if that discards work you paid for.
- 3Show the work in public while you build, so that customers and possible buyers know the product on release day.
- 4Raise the price in steps and watch whether growth slows.
- 5Keep code, documentation and support in a state that another company can take over, in case a buyer appears.
What to watch
The risks are the mirror image of the causes above.
The price is not known
"Seven figures" covers everything from US$1,000,000 to US$9,999,999. The seller gives no number, the buyer said nothing in public about the price, and the US$1 million in many articles matches another founder's guess from the day after the announcement. The headline figure on this page is the lowest value of that range.
The revenue figures are the seller's own
US$1,000 and US$20,000 a month come from one interview, ten months after the sale. No dashboard, payment processor or journalist has confirmed them. Asked in July 2021 for the revenue at the time of sale, he declined to answer.
"8 months" is rounded
The launch on Product Hunt was on 29 June 2020 and the announcement on 30 March 2021, which is nine months. The founder's accounts also differ on details: version 1 took three weeks or one month, and one interview places GPT-3 two months after the launch while his own thread has him asking for access in the launch month.
The product stopped
Headlime has had no new features since March 2021. Buyers of the lifetime deal kept access to a product that no longer changes and did not get the buyer's product. A reader who had supported Headlime early complained about that in the founder's thread. The site is still online in October 2026, but this page did not test whether it still generates text or takes payments.
The early advantage is gone
The advantage was access to GPT-3 before most developers had it. By the month of the sale a competitor had raised close to US$3 million. Today any developer can call a text model, and chat assistants write marketing text directly, so a copy of Headlime would start with no advantage.
Profit is unknown
The founder says GPT-3 was a major part of the monthly cost and gives no amount. Revenue is known only as two monthly figures, so neither the total revenue of the nine months nor the profit can be computed.
Sources
The grade next to each source is the best grade that the source can support. "Read in full" means that the text of the page was downloaded and the numbers were checked in it.
- C Indie Hackers, 1 July 2021: Danny Postma, Zero to 7 Figure Exit in 8 Months with Headlime, AMARead in full, with the founder's answers. Source for the price steps, the lifetime deal tiers, the GPT-3 email, the technology, the sale process, the legal form and his refusal to give the price.
- C Failory, 18 January 2022: Building & Selling a SaaS for +$1M in 8 Months, interview with Danny PostmaRead in full. Founder interview. Source for the US$1,000 and US$20,000 monthly figures, the US$60,000 launch, the five-month rebuild, the two offers and the words "seven figures".
- C Headlime blog: Headlime and Conversion.ai Join Forces in Artificial Copywriting, first saved by the Internet Archive on 30 March 2021Read in full on the live site. The seller's announcement. Source for the buyer, the plan to merge the two tools, the end of new features and the 100% monthly growth claim. The date comes from the Internet Archive index.
- C Danny Postma on X, 30 March 2021: thread announcing the acquisitionRead ten posts of the thread through the fxtwitter API. Source for the announcement date, the "8 months" and his new role.
- C TechPP, 31 March 2021: Conversion.ai Acquiring Headlime is Bad News for Headlime UsersRead in full. A journalist's report. Source for the undisclosed terms, the comments of the buyer's founder, the effect on lifetime-deal buyers and the Copy.ai funding.
- C They Got Acquired, 21 April 2022: In less than a year, this founder bootstrapped and sold an AI copywriting platformRead in full. A journalist's report based on public sources; the founder did not answer its request for comment. Source for the undisclosed terms, the team count and how the buyer knew the product.
- C Pieter Levels on X, 31 March 2021: estimate of the sale priceRead in full through the fxtwitter API. An outside guess, marked as such by its author.
- C Product Hunt launch page of Headlime as saved by the Internet Archive, 12 August 2020Read the launch post and the first comments. The live page could not be opened. Source for the launch date of 29 June 2020, the product-of-the-week rank and the lifetime plans.
- C Danny Postma on X, 29 June 2020: launch threadRead four posts of the thread through the fxtwitter API. Source for the launch date and the headlime.io domain.
- C TechCrunch, 17 March 2021: Okay, the GPT-3 hype seems pretty reasonableRead the section on Headlime. A journalist's own test of the product. It gives no revenue or user numbers.
- C Unite.ai, 20 February 2024: interview with Danny PostmaRead the answers about Headlime and GPT-3. Source for the name of OpenAI's chief technology officer and the first batch of users.
- C The Bootstrapped Founder, 9 August 2023: Danny Postma, an indie hacker's business evolutionRead the parts of the transcript about Headlime. Source for the three options, the two interested buyers, the copies of version 1 and the later product.
- C Headlime pricing page and home page, read 13 October 2026Read in full. Source for the pricing table, the languages, the customer counter and the state of the site in 2026. The help site help.headlime.com did not respond.
- C Headlime About page, read 13 October 2026Read in full. Source for the three team members and the origin of the tool as a spreadsheet of templates.
- C Headlime home page as saved by the Internet Archive, 24 October 2020Read in full. Shows the product as a template tool for headlines four months after launch.
- C Sacra, 16 November 2022: Dave Rogenmoser, CEO and co-founder of Jasper, on the generative AI opportunityRead the answer on the company's first product. Source only for the fact that Jasper was called Conversion.ai at launch. It does not mention Headlime.
- C Indie Hackers, 3 November 2023: Danny Postma built and sold AI writing tool for ~$1m in 8 months, by Hazel LimRead in full. A community summary of a longer article on thetinyempires.com, which could not be opened. An example of the US$1 million figure given without a source.
- C Indie Hackers podcast, episode 278, 5 May 2023: Growing a Portfolio of AI Products Past $300k After a 7-Figure Exit with Danny PostmaRead the episode title on the DEV Community mirror. The audio was not transcribed. Shows that the podcast also describes the sale only as seven figures.
- C Loot Drop database export, 1,749 failed startupsLocal export. AI-written summaries, not independently verified. Source for the failed predecessors.