MausBot: Revenue, team and how it was built
MausBot (Open-source chat app for a team of AI agents, with paid cloud machines, India): US$5.5K in the last 30 days, October 2026. Evidence grade A: disclosed by a company or a filing, or verified by a payment processor. Team: 2 founders, plus outside open-source contributors. Launched 11 August 2026; still running, revenue growing.Added 25 October 2026
- Outcome
- US$5.5K in the last 30 days, October 2026Revenue growing
- Evidence
- A disclosed by a company or a filing, or verified by a payment processor
- Launch
- 11 August 2026
- Status
- Still running
- Team
- 2 founders, plus outside open-source contributors
- Outside funding
- None
- Main channel
- Not stated
MausBot is a chat app in which each contact is an AI agent: a program that uses a language model to plan steps and then act, for example open a browser, fill in a form, edit a file or draft an email, instead of only answering in text. The app does not contain its own model. It starts the CLI (command-line interface, a program that is used by typing commands in a terminal) of Claude, Codex or Grok on the user's computer, with the user's existing login, and a small local harness, a server that owns every agent process and passes its messages to the chat window, shows the work. Risky actions such as shell commands appear as approval cards that the user allows or denies. The app is open source under the Apache 2.0 licence: anyone may read, change and reuse the code. The money comes from Cloud plans that keep the agents running on a rented Linux machine when the user's laptop is closed.
Omkar Satpute and Milind Soni make it, through Supamaus Software Private Limited, a company incorporated in India. The public repository was created on 11 August 2026, the day xAI released Grok Bot in beta, and the founders wrote on Product Hunt that they open-sourced their work about two hours after that launch. By 25 October 2026 the repository had 5,595 stars and 941 forks. The payment provider records read by TrustMRR show the first payment on 27 September 2026, US$753 in September, US$4,751 in October to the 25th, and US$5,503 in the last 30 days. Monthly recurring revenue was US$7,020 from 121 active subscriptions.
The revenue figures are grade A: TrustMRR reads them from the company's Dodo Payments account through an API key. But they cover one month. All revenue except about US$1 arrived in the 30 days before 25 October, and one day, 16 October, brought US$1,549, about 28% of it. This page does not treat one month as a trend. The profile text says the agents write "posts and replies"; the company's pages show that they do this only in apps that the user connects, after the user approves, and its terms forbid spam. The page also shows earlier companies that sold a personal assistant and failed.
How much revenue does MausBot 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 |
|---|---|---|---|---|
| Revenue in the last 30 days, at 25 October 2026 | US$5,503 | A | TrustMRR, read from the Dodo Payments account by API key, synced 25 October 2026 at 06:29 UTC. One 30-day period. The discovery endpoint of TrustMRR, read the same day from a different snapshot, gave US$5,520.81. | [1, 2] |
| Revenue since the start, at 25 October 2026 | US$5,504 | A | Same source and sync. The same value appears for the last 3, 6 and 12 months, so the business had almost no revenue before 26 September 2026. | [1] |
| Revenue by month, 2026 | September US$753; October to the 25th US$4,751 | A | TrustMRR monthly table. The first payment, US$200, is on 27 September 2026, so September covers four days. | [1] |
| Share of all revenue that arrived in October | about 86% | calc | Calculated: 4,751 / 5,504 = 86.3%. | [1] |
| Largest single day | US$1,549 on 16 October 2026, about 28% of the 30 days | A | TrustMRR daily table; the share is calculated: 1,549 / 5,503 = 28.1%. Ten of the 30 days in the table show US$0. The profile does not say what was bought on 16 October. | [1] |
| Revenue in the last 7 days, at 25 October 2026 | US$1,711 | A | TrustMRR period table. | [1] |
| Monthly recurring revenue and active subscriptions | US$7,020 a month; 121 subscriptions | A | TrustMRR snapshot. MRR is the monthly value of the subscriptions that are active now. It is higher than the 30-day revenue because many subscriptions began recently, some of them after a 3-day free trial. The discovery endpoint showed US$7,219 in its snapshot. | [1, 2, 4] |
| Average MRR per active subscription | about US$58 a month | calc | Calculated: 7,020 / 121 = 58.0. Between the Pro price, US$49, and the Max price, US$99. | [1, 4] |
| 30-day revenue at a yearly rate | US$66,036 | calc | Calculated: US$5,503 x 12. This is the money figure that this site stores for the story, not a forecast. MRR x 12 is US$84,240. | [1] |
| Visitors in the last 30 days | 35,787; US$0.15 of revenue per visitor | A | TrustMRR, from the connected Google Analytics account. The company's partners page says 90,000 unique visitors a month, which this figure does not confirm. | [1, 12] |
| People using the free app | 10,000 or more | C | Company partners page (10k+) and TrustMRR profile text (estimated user count 10,000). Not verified. | [12, 1] |
| GitHub repository, 25 October 2026 | 5,595 stars, 941 forks | B | GitHub's public data for milind-soni/OpenMausBot. A star is a bookmark by a GitHub user; a fork is a copy of the code in another account. The repository was created on 11 August 2026 at 18:58 UTC. | [8] |
| Release pace | the newest 100 releases run from version 0.1.21 on 16 August to 0.1.127 on 24 October 2026 | B | GitHub release data of the public releases repository. | [9] |
| Team | 2 people | C | The company's post-about-us page says that only "the two of us who run MausBot" can read the submissions. TrustMRR profile text: team size 2 to 5. Outside contributors also send code: the repository owner's account has 3,704 commits, and the next three accounts 544, 345 and 334. | [5, 1, 8] |
| Outside funding | none reported | C | TrustMRR profile text says Bootstrapped. The home page lists the Google for Startups Cloud Program and Claude for Startups, which are programs of those vendors; no source read here mentions an equity investment. | [1, 3] |
| Company | Supamaus Software Private Limited, incorporated in India | C | Terms of service. The terms name Indian law and the courts at Pune. | [11] |
| Payment provider | Dodo Payments | A | TrustMRR verification line. The company's terms say that Dodo Payments acts as merchant of record and handles taxes; the TrustMRR profile says Merchant of record: No. The two sources disagree on this point. | [1, 11] |
| Product Hunt launch | 13 October 2026, number 7 of the day, 158 points | C | Product Hunt product page, read 25 October 2026. | [10] |
| Time from the public repository to the October 2026 figure | 2 months | calc | Calculated: 11 August to 25 October 2026 is 75 days, about 2.5 months; the first payment came 47 days after the repository. | [8, 1] |
How to read this table: a grade A value was read by TrustMRR from MausBot's Dodo Payments account through an API key, last synced on 25 October 2026 at 06:29 UTC, or from a connected Google Analytics account. A grade B value is visible in public code or public repository data on GitHub. A grade C value is a statement by the founders or the company. A "calc" value was computed on this page from the rows above it. Grade A covers money received and traffic. It does not cover costs, margin or profit.
| Plan | Per month, billed yearly (US$) | Per month, billed monthly (US$) | Limit |
|---|---|---|---|
| Free, desktop app | 0 | 0 | your own computer and AI account; unlimited bots |
| Starter, Cloud | 15 | 2 shared CPUs, 2 GB memory, 100 computer-hours a month | |
| Pro, Cloud | 49 | 4 shared CPUs, 4 GB memory, 300 computer-hours a month | |
| Max, Cloud | 99 | 8 shared CPUs, 8 GB memory, 800 computer-hours a month | |
| Team | 135 | 169 | up to 5 people; 1,000 computer-hours a month, shared |
| Business | 479 | 599 | up to 20 people; US$25 a month for each extra person |
| Business self-hosted | 299 | billed yearly; up to 20 people on your own servers |
The desktop app is free. Starter, Pro and Max are monthly Cloud plans for one person, each with a 3-day free trial that needs a card; the first payment is on day 4. Pro includes US$10 and Max US$20 of Claude credit every month. Team and Business plans start with a call; the yearly prices are per month, billed yearly. In every plan the customer pays the model provider separately, through an own Claude, ChatGPT or Grok plan or an API key. The TrustMRR profile text lists older launch prices: Personal US$29, Pro US$49 and Max US$89 a month.
When did MausBot launch, and how did it grow?
From a public repository on the day a large competitor launched to the first payment in seven weeks, and US$5,503 in the next 30 days. Each entry carries its grade.
- 11 August 2026BxAI releases Grok Bot in beta, a messaging app of agents in its cloud, on Grok only. The same day the founders create the public repository, first named OpenGrokBot. They wrote that they open-sourced it about two hours after the launch, from work they had already started. [6, 8, 10]
- 16 August 2026BVersion 0.1.21, the oldest of the newest 100 releases. The pace is more than one release a day until late October. [9]
- 17 August 2026CFirst blog post: the open-source alternative to Grok Bot, on the Claude, Codex and Grok tools the user already has. [6]
- 22 September 2026CEleven blog posts in one day. By 8 October the blog has 39 posts, most of them comparisons with other agent products. [13]
- 27 September 2026AFirst payment: US$200. September ends at US$753. [1]
- 6 October 2026CBlog post Meet MausBot, signed by Omkar Satpute and Milind Soni, under the new name. [7]
- 13 October 2026CProduct Hunt launch, number 7 of the day with 158 points. [10]
- 16 October 2026AUS$1,549 in one day, the largest day in the table. [1]
- 25 October 2026AUS$5,503 in the last 30 days, MRR US$7,020, 121 active subscriptions, 35,787 visitors in 30 days. Version 0.1.127 was released the day before. [1, 9]
Why did MausBot work?
Four causes, in the order they mattered. Each one is a reusable lesson.
Speed
The founders published an open version of a new product category on the day a large company launched it. People who searched for an open-source Grok Bot, or who did not want to move to Grok, found a working repository the same day. The company's blog says that people search for "open source grok bot", and its pages use those words.
The customer pays for the model
The agents run on the customer's own Claude, Codex or Grok subscription, through the official command-line tools. MausBot pays for no model use except the small Claude credit in its higher plans, so a Cloud plan at US$15 a month is mostly the price of a small rented server.
Free app, paid always-on machine
The desktop app is free and complete, but agents on a laptop stop when the laptop sleeps. The Cloud plans sell the part that a free app cannot give: a machine that keeps scheduled jobs running, cloud desktops for the agents and phone alerts. Checkout for the 3-day trial asks for a card, and the first payment is on day 4 unless the user cancels.
Content that answers comparison searches
Most of the 39 blog posts compare MausBot with a named product of xAI, Meta, OpenAI, Manus or another open project. A rewards page offers up to US$50 in Claude credits for honest public posts on the poster's own account, with the reward disclosed where the platform asks for it.
How it was built
These details come from the public repository and from the founders' profile text on TrustMRR. The repository details are grade B, the profile text grade C. [8, 1, 9]
- Desktop shells for macOS, Windows and Ubuntu in Electron; the chat interface in React, Vite and Tailwind CSS. The profile also lists React Native, Swift, SwiftUI and Kotlin for the phone apps.
- A harness server on 127.0.0.1 starts the claude, codex and grok command-line tools and turns each provider's protocol into one stream of events for the chat window.
- A permission broker turns each risky action, such as a shell command or a file edit, into an approval card in the chat.
- Cloud computers for the agents come from Boat (boat.dev), which the README calls a sponsor. Connected apps such as Gmail, Slack and GitHub come through Composio. Hosted voices come from ElevenLabs or Fish Audio.
- Payments through Dodo Payments. The profile lists Node.js, Docker, Cloudflare, Vercel and Fly.io for the back end.
- More than one release a day, with outside contributors sending code to the public repository.
Tools named in the sources
Each one opens its page in the Tools tab, with every story that uses it.
Who tried this before MausBot, and why did they fail?
Five companies in the Loot Drop database sold a personal or work assistant that would do tasks for its users, and failed. The values below are from that database and are not independently verified.
Zirtual
The promiseVirtual assistants for busy business owners: email, scheduling and routine tasks.
Why it diedThe assistants were employees, so costs grew with every customer. It expanded faster than its cash allowed and stopped paying.
Exec
The promiseAn on-demand concierge for any task, from cleaning to personal errands.
Why it diedIt tried to cover every task, while a competitor chose one kind of task and made it work.
Operator
The promiseA personal shopper in an app, with people and software finding hard-to-find products.
Why it diedHuman work in every request made each order expensive, and large online stores were cheaper and faster.
Niles
The promiseA bot inside Slack that answered staff questions from the company's knowledge.
Why it diedSlack added its own features, and cheaper competitors arrived.
Petalite
The promiseA privacy-first voice assistant that ran on the device instead of in the cloud.
Why it diedIt could not compete with the hardware, data and ecosystems of Amazon and Google, even with better privacy.
The failed startups come from Loot Drop, a free database of dead startups, retold here in our own words. Loot Drop describes its entries as AI-assisted summaries of public sources, so these cards carry no evidence grade.
What changed
Zirtual, Exec and Operator sold an assistant that was partly or fully a person, so each new task added labour cost, and the price that customers would pay did not cover it. Niles and Petalite built software assistants, but a platform owner, Slack or Amazon and Google, offered the same thing inside a product that people already used.
MausBot removes most of the cost side. The agent is a language model, and the customer pays the model provider through a subscription that the customer already has, so the business sells only a rented machine and the app around it. The platform risk is still there and is larger: xAI, Meta, OpenAI and Manus each sell their own agent product, and MausBot depends on their command-line tools and on their subscription rules. Its answer so far is to support several providers at once and to keep the code open, so that a user is never tied to one of them.
Can one person copy MausBot now?
The timing, rarely. The structure, yes.
An open copy published on the launch day of a large company's product is an opening that lasts hours. The structure can be repeated: a free open-source app that runs on the customer's own AI subscription, and a paid hosted machine for the part that must run all day.
Where it is still open: Open, multi-provider versions of each new agent product that the large AI companies release on one model only.
- 1Let customers bring their own AI subscription or API key, so the model is never your cost.
- 2Give the full app away as open source, and charge for the always-on hosted version.
- 3Write a comparison page for each product that people search alternatives for.
- 4Connect revenue to a verifier such as TrustMRR from the first payment, so outsiders can check the numbers.
What could go wrong for MausBot?
The risks are the mirror image of the causes above.
One month of revenue
All revenue except about US$1 arrived in the 30 days before 25 October 2026. A business that is two months old, with one month of sales, has not yet shown whether customers renew. The next months of the same TrustMRR profile will show it.
One large day
US$1,549 of the US$5,503 came on 16 October, three days after the Product Hunt launch, and ten of the 30 days show no revenue. The profile does not say what was bought that day, for example a yearly or team plan.
It runs on other companies' tools
The agents use the Claude, Codex and Grok command-line tools with the customer's own subscription. A change in those tools or in the providers' subscription rules could stop the product or raise its price.
Large competitors
xAI, Meta, OpenAI and Manus sell their own agent products, which the blog compares with MausBot. Each predecessor on this page lost to a company with more money or a platform of its own.
Open code can be copied
The code is under the Apache 2.0 licence and has 941 forks. Anyone may host it and sell a competing Cloud plan.
Company claims and verified data differ
The partners page says 90,000 unique visitors a month; the connected Google Analytics account shows 35,787 in 30 days. The terms call Dodo Payments the merchant of record; the TrustMRR profile says it is not.
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.
- A TrustMRR: MausBot, markdown profile (data as of 25 October 2026)Read the markdown profile in full on 25 October 2026. Revenue, MRR, subscriptions and traffic come from its Verification Sources (Dodo Payments API key, Google Analytics); the description, team size, funding, prices and tech stack are founder-written profile text, grade C.
- A TrustMRR public discovery endpoint, fastest-growing startups (read 25 October 2026)Read in full. A second snapshot of the same verified data: 30-day revenue US$5,520.81 and MRR US$7,219.
- C MausBot home pageRead in full on 25 October 2026. Source for the ready-made bots, the approval rule and the startup programs.
- C MausBot pricing pageRead in full on 25 October 2026. Source for every price, the trial terms and the included Claude credit.
- C MausBot: Post about MausBot, get Claude creditsRead in full. Source for the post-rewards rules and the team of two.
- C MausBot blog, 17 August 2026, updated 3 October 2026: Open source Grok Bot: MausBot, the free alternativeRead in full. Source for the Grok Bot launch date and the first name, OpenGrokBot.
- C MausBot blog, 6 October 2026: Meet MausBot, your own team of AI bots, in a chat appRead in full. Names both founders.
- B GitHub: milind-soni/OpenMausBot, README and repository dataRead the README in full. Creation date, stars, forks and commit counts read from GitHub's public API for the repository and its contributors list on 25 October 2026.
- B GitHub: milind-soni/openmausbot-releases, releasesRead the releases page and the release list from GitHub's public API (newest 100 releases).
- C Product Hunt: MausBot, Open source Grok Bot: Free, LocalRead in full, including the maker's launch comment. Source for the launch date, rank and points.
- C MausBot terms of serviceRead the sections on the company, payments, acceptable use and governing law.
- C MausBot: Become a Featured PartnerRead in full. Source for the company's user, visitor and Discord figures.
- C MausBot blog indexRead in full. Source for the dates and topics of the 39 posts.
- C Loot Drop database export, 1,749 failed startupsLocal export. AI-written summaries, not independently verified. Source for the failed predecessors.