Localhero.ai: Product, team and how it was built
Localhero.ai (Translation management for software teams: a GitHub Action that adds AI translations of new strings to every pull request, Sweden): Paid plans live since May 2025, November 2026. Evidence grade E: no revenue figure was checked. Team: Founder Arvid Andersson, who calls it "a small bootstrapped company" where customers "talk directly to me" (about page, November 2026); headcount not published. Launched December 2024; still running.Added 6 November 2026
- Outcome
- Paid plans live since May 2025, November 2026
- Evidence
- E no revenue figure was checked
- Launch
- December 2024
- Status
- Still running
- Team
- Founder Arvid Andersson, who calls it "a small bootstrapped company" where customers "talk directly to me" (about page, November 2026); headcount not published
- Outside funding
- Bootstrapped, as the founder states on the about page; no outside funding found
- Main channel
- Search and AI answers
What is not verified
No revenue figure was found. We looked for a payment-processor profile (TrustMRR, Stripe), a founder post or interview with a dated amount, and a company filing, and found none; Swedish company-data sites block AI agents, so the holding company's accounts were not read. The founder's Hacker News posts mention paying customers without numbers. A reader could look for a revenue statement from the founder or the annual report of Andersson-Larsson Holding AB. This story is grade E: the business is real and its product, price and customers are described from public pages, but no revenue figure was checked. Aiolum shows it as an idea worth studying, not as proof that it makes money.
Localhero.ai is a translation management system (TMS), software that stores an app's text in every language and organizes who translates and reviews it. It works through a GitHub Action, a small program that GitHub runs automatically on each change to a repository. On every pull request (a proposed code change that colleagues review before it is merged) it translates the new text keys in the app's locale files (YAML for Rails, JSON for React, .po files for Django and Phoenix) and commits the result. Arvid Andersson in Stockholm builds it; the company behind it is Andersson-Larsson Holding AB.
The domain showed the product, "AI-powered translations that capture your brand's unique voice", in December 2024, and the command-line tool was first published to npm on 5 December 2024. By May 2025 the pricing page listed paid plans, paid through Polar and Stripe. The first customer was Kundo, a Swedish customer-service platform where the founder also worked as a consultant; it translates into nine languages with Localhero.ai. In his monthly Hacker News posts the founder reported a first fully self-serve customer in spring 2026, a "pretty big Swedish company" in June 2026 and "a few new paying customers" in July 2026.
No revenue figure was found, so this story has grade E. The prices are public: a free plan with 250 credits a month, then US$29, US$79 and US$350 a month, as the pricing page states. The page shows that the product is live and sold, not how much it earns.

How much revenue does Localhero.ai make?
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 |
|---|---|---|---|---|
| Founder | Arvid Andersson, Stockholm | C | The about page is a letter signed "Arvid Andersson, Founder". The case study calls him a person with "Over 20 years in tech, as an engineer and in leadership roles". | [3, 7] |
| Company | Andersson-Larsson Holding AB, registration no. 559254-7078 | C | Terms of service: "Localhero.ai (the "Service") is a product by Andersson-Larsson Holding AB". The registration date of this holding company was not checked. | [6] |
| Team and funding | "a small bootstrapped company" | C | About page: "We're a small bootstrapped company, so if you need help or have feedback, you'll talk directly to me." No headcount is published. | [3] |
| First public version | December 2024 | B | Internet Archive copy, 27 December 2024: "AI-powered translations that capture your brand's unique voice" with a contact button. The npm registry dates the first release of @localheroai/cli to 5 December 2024. | [14, 13] |
| Paid plans first seen | May 2025: Starter US$29 a month, Professional US$149 a month | B | Internet Archive copy of the pricing page, 20 May 2025, with payment "through Polar.sh / Stripe". | [15] |
| Prices, November 2026 | free plan, then US$29, US$79 and US$350 a month | C | Pricing page, read 6 November 2026. | [2] |
| Paying customers, July 2026 | "A few new paying customers this month" | C | Founder's comment in the Hacker News thread "What Are You Working On? (July 2026)", 16 July 2026. No count given. | [11] |
| First self-serve customer | spring 2026 | C | Founder, 10 May 2026: "Got our first fully selfserve customer a few weeks back (found us through the docs)." | [9] |
| Reference customer | Kundo, nine languages, "rolled out across all our services" | C | Case study on the company blog, 7 June 2026. The founder writes that he was "also doing some consulting at Kundo" while he built Localhero. | [7] |
| GitHub Marketplace app | 10 installs | B | GitHub Marketplace page, read 6 November 2026. The app is free there; paid plans are bought on localhero.ai. | [16] |
| AI models | Azure OpenAI (EU data zone) for translations; Anthropic's Claude through AWS Bedrock for checks | C | Security page: "Translations run on Microsoft Azure OpenAI with Data Zone EU deployments" and "Some analysis (glossary, style guide and quality checks) uses Anthropic's Claude models through AWS Bedrock." | [5] |
| Time from the first release to this page | 23 months | calc | Calculated: December 2024 to November 2026. | [13, 2] |
How to read this table: a grade C value is the company's or the founder's own statement. A grade B value was read on a page that a third party controls (the Internet Archive's dated copies, the npm registry, GitHub Marketplace). A "calc" value was computed on this page. No value is a revenue figure.
| Plan | Per month, billed yearly (US$) | Per month, billed monthly (US$) | Limit |
|---|---|---|---|
| Free (open source or side project) | 0 | 0 | 1 project, 250 credits a month |
| Starter | 24.17 | 29 | 4 team members, 2 projects, 1,250 credits a month |
| Growth | 65.83 | 79 | 8 team members, 5 projects, 3,000 credits a month |
| Professional | 291.67 | 350 | 30 team members, 20 projects, 12,500 credits a month |
One credit translates one key of up to about 100 characters into one language. Every paid plan starts with a 30-day free trial without a card. Yearly billing gives two months free; the monthly equivalents of the yearly prices are calculated. A 500-credit top-up costs US$19 on any paid plan. Enterprise prices are on request.
When did Localhero.ai launch, and how did it grow?
From a one-page site in December 2024 to three paid plans and a GitHub Marketplace app, in 23 months. Each entry carries its grade.
- December 2024BThe command-line tool @localheroai/cli is first published to npm (5 December), and localhero.ai shows "AI-powered translations that capture your brand's unique voice". [13, 14]
- May 2025BThe pricing page lists a Starter plan at US$29 a month and a Professional plan at US$149 a month, with payments through Polar and Stripe. [15]
- November 2025CThe founder describes the product on Hacker News: "Automated i18n translations that run in your CI pipeline." [8]
- May 2026CThe founder reports the first fully self-serve customer, who found the product through the documentation. [9]
- June 2026CA new paying customer, "a pretty big Swedish company", and a customer in Japan who asked for Easy Japanese; the Kundo case study is published on 7 June. [10, 7]
- July to November 2026C"A few new paying customers" in July; a GitHub App on the Marketplace, an MCP server and three paid plans at US$29, US$79 and US$350 a month by November. [11, 2, 16]
Why did Localhero.ai work?
Four causes, in the order they mattered. Each one is a reusable lesson.
Translations inside the pull request
The translations are committed to the same pull request that added the strings, so the feature and its translations merge together and the main branch is never half translated. There is no export step and no second pull request.
Glossary, style guide and memory
Each project has a glossary (terms that must be translated one way, or not at all), a tone setting and a translation memory built from past edits. The founder wrote in March 2026 that edits made by a product manager in the web page now feed back into future translations.
A surface for people who do not use git
Every pull request gets a page where reviewers see the new strings in every language, change them and approve them; the edits are synced back to the branch as a commit.
How it was built
These details come from the company's own pages, its llms.txt file and the founder's posts. They are grade C. [4, 5, 2, 8]
- A GitHub Action (localheroai/action), a command-line tool run with npx @localheroai/cli, a GitHub App and a REST API.
- An agent skill for Claude Code and Cursor that gives the coding assistant the project's glossary and tone, and a remote MCP server (Model Context Protocol, a standard that lets AI assistants call outside tools) for Claude.
- Translations run on Microsoft Azure OpenAI inside the EU; glossary, style and quality checks use Anthropic's Claude models through AWS Bedrock in Frankfurt.
- The application and database run on Render in Frankfurt; card payments go through Polar and Stripe.
- Support for Rails YAML, i18next JSON and gettext .po files from Django, Phoenix and LinguiJS, with working example apps on GitHub.
Tools named in the sources
Each one opens its page in the Tools tab, with every story that uses it.
Who tried this before Localhero.ai, and why did they fail?
Three companies in our database of failed startups promised faster translation or automation inside the developer workflow and failed. The values below are from that database and are not independently verified.
TurboTranslations
The promiseFaster and cheaper translations through a marketplace of human translators.
Why it diedIt bet on human translators just as machine translation became good enough for most texts.
Sider
The promiseAutomated code review and static analysis for development teams, run in the CI/CD pipeline.
Why it diedFree open-source tools and large platforms that bundled code analysis took its market.
Pagedraw
The promiseTurn Sketch designs into React code to save developers time.
Why it diedIt bet on Sketch when designers moved to Figma, and developers preferred to write that code themselves.
The failed startups come from our database of 1,749 dead startups, retold here in our own words. Its entries are AI-assisted summaries of public sources, so these cards carry no evidence grade.
What changed
TurboTranslations organized human translators, and machine translation took away its advantage. Localhero.ai starts from the machine translation and sells the work around it: the pull-request step, the glossary, the review page and the memory of past edits.
Sider sold one step in the CI pipeline and was squeezed between free tools and large suites. Localhero.ai faces the same risk, since a team can call a model API from its own script; the founder answers this in his blog post on using the OpenAI API directly and in his March 2026 post: "there is a lot of stuff around it to make it work well over time in a product org."
Can one person copy Localhero.ai now?
Yes, for a developer who knows one workflow well.
One founder builds and sells it on rented AI models, with small hosting costs and no outside money. The hard part is not the translation but fitting into the tools a team already uses and earning trust with reference customers.
Where it is still open: The same pattern for other review steps that wait on people: legal copy checks, accessibility text, app-store descriptions in many languages.
- 1Pick a step that makes finished code wait (translations, screenshots, release notes).
- 2Run it where developers already work: a GitHub Action and a command-line tool.
- 3Give non-developers a web page to review and edit, synced back to the code.
- 4Start with one customer you know, write the case study, then add a free plan for open-source projects.
What could go wrong for Localhero.ai?
The risks are the mirror image of the causes above.
What is not verified
No revenue figure was found. We looked for a payment-processor profile (TrustMRR, Stripe), a founder post or interview with a dated amount, and a company filing, and found none; Swedish company-data sites block AI agents, so the holding company's accounts were not read. The founder's Hacker News posts mention paying customers without numbers. A reader could look for a revenue statement from the founder or the annual report of Andersson-Larsson Holding AB.
The first customer and the founder's consulting
The Kundo case study says the founder was "also doing some consulting at Kundo" while he built the product, and it is the first customer and the main home-page quote. The home page also shows a second quote and a row of customer logos (Kundo, Willful, Bespoke, Mynewsdesk, OutdoorMap, Qasa); only Kundo has a published case study.
Company start date and the founder's other site
The product launched in December 2024, but the legal seller is a holding company, Andersson-Larsson Holding AB, whose registration date was not checked. The founder also runs infrabase.ai, a directory of AI infrastructure tools, as a side project, as his Hacker News posts say.
Competition and model providers
The site already has Linguana, which translates websites built with Webflow and Framer, and TranslateMom for video; Localhero.ai translates the text inside software and competes more directly with TMS products such as Lokalise, Phrase and Crowdin, which its blog compares. Translations depend on Microsoft Azure OpenAI and checks on Anthropic's Claude through AWS Bedrock, as the security page states; their prices and rules set Localhero.ai's costs.
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 Localhero.ai home page, read 6 November 2026Read in full. Source for the workflow, the quotes and the customer logos.
- C Localhero.ai pricing, read 6 November 2026Read in full. Source for the plans, credits, top-ups and payment processors.
- C About Us - Localhero.ai, read 6 November 2026Read in full. The founder's letter, the first customer and "small bootstrapped company".
- C Localhero.ai llms.txt, read 6 November 2026Read in full. The company's description for AI assistants, the CLI, the agent skill, the MCP server and the blog list.
- C Localhero.ai security and data privacy, read 6 November 2026Read the sections on hosting, AI providers and sub-processors.
- C Localhero.ai terms of service, read 6 November 2026Read the sections naming the company, the free plan rule and the data processing agreement.
- C How Kundo automated its localization workflow, Localhero.ai blog, 7 June 2026Read in full. The first customer, the founder's consulting and the migration steps.
- C Arvid Andersson (arvida), comment in "Ask HN: What Are You Working On? (Nov 2025)", 11 November 2025Read in full through the Hacker News Algolia API (search by author).
- C arvida, comment in "Ask HN: What are you working on? (May 2026)", 10 May 2026Read in full through the Hacker News Algolia API (search by author). The first self-serve customer.
- C arvida, comment in "Ask HN: What are you working on? (June 2026)", 15 June 2026Read in full through the Hacker News Algolia API (search by author). The Swedish customer and Easy Japanese.
- C arvida, comment in "Ask HN: What Are You Working On? (July 2026)", 16 July 2026Read in full through the Hacker News Algolia API (search by author). "A few new paying customers this month".
- C arvida, comment in "Ask HN: What Are You Working On? (March 2026)", 9 March 2026Read in full through the Hacker News Algolia API (search by author). Outreach to Swedish companies and the learning loop.
- B npm registry: @localheroai/cliRead the registry record (registry.npmjs.org): first release 5 December 2024, maintainers, licence.
- B Internet Archive copy of localhero.ai, 27 December 2024Read in full. First copy that shows the product and the company name.
- B Internet Archive copy of localhero.ai/pricing, 20 May 2025Read in full. First copy of the pricing page with paid plans.
- B Localhero.ai on GitHub Marketplace, read 6 November 2026Read the listing, install count and plan.
- C Database of failed startups, local exportLocal export. AI-written summaries, not independently verified. Source for the failed predecessors.
Run this company? Send corrections or a verified figure to aiolum@livyden.com