# Userdesk

Aiolum story: https://aiolum.com/story/userdesk/ · added 2026-10-31 · updated 2026-10-31

**In short.** Userdesk (AI support chatbot trained on a company's website, Italy): Sold at an undisclosed price after about US$1.1K MRR, March 2024. Evidence grade D: reported, not verified. Team: 1 founder, solo. Launched June 2023; sold 9 months after launch.

AI support chatbot trained on a company's website · Italy · AI-native · solo founder · sold · group: Customer support agents

- Outcome: Sold at an undisclosed price after about US$1.1K MRR, March 2024 (evidence grade of the headline number: D; headline in US dollars: 13200)
- Launch: June 2023; months from launch to the outcome: 9
- Team: 1 founder, solo
- Outside funding: None reported
- Main factor: Speed
- Meters, 1 to 4: evidence strength 1, copyable pattern 3, capital needed 1
- Tags: AI-native, Solo founder, Bootstrapped, Exit, B2B, Customer service

Links:

- Visit userdesk.io: https://userdesk.io/

**The product.** A no-code tool to build a chatbot that answers visitors' questions from a company's own content: web pages, sitemaps, Notion pages, Google Drive files and PDFs. The owner pastes a link, the bot reads the content, and a code snippet puts the chat window on the website or in Slack. The answers came from OpenAI models.

**Why it worked.** In early 2023 every company wanted a ChatGPT that knew its own website, and a solo engineer could build one in about three months on a stack he already used. The founder said most customers found it through AI directories. He sold it because his audience of developers on X did not overlap with the people who buy support chatbots.

Evidence grades: A = disclosed by a public company, a filing or a payment processor, or verified by one; B = two or more independent reports agree, or a journalist confirmed it; C = a founder claim that no one else verified; D = reported, not verified: one founder claim or a third-party estimate with no second source; calc = computed on this page from graded facts.

## Summary

Userdesk was a retrieval chatbot service: a chatbot that first searches a company's own documents for the passages that match a question and then asks a language model to write the answer from those passages. The customer added a website link, a sitemap, Notion pages or PDF files, and Userdesk turned them into a chat window that could be embedded on any site or used in Slack. It used OpenAI's GPT-3.5 by default and GPT-4 on the higher plans, and it had a free plan with 30 AI messages a month.

Luca Restagno, an Italian software engineer who had already sold three products, started it in early 2023 and launched it in June 2023. He wrote in April 2024 that he had pre-sold 20 lifetime licenses at US$69 in 24 hours, that it reached about US$1,100 in MRR (monthly recurring revenue, the subscription money that comes in every month) in under twelve months, and that he then sold it, his fourth exit. A newsletter, No Code Exits, repeated these figures in July 2026 and said the sale went through Acquire.com, a marketplace for small software businesses, in March 2024.

The revenue figure is the founder's own statement, made once, and the newsletter repeats it, so it is grade D: reported, not verified. The buyer and the price were not disclosed. The founder also reported that about US$500 a month of that recurring revenue came from one enterprise customer. The page shows earlier companies that sold support chat and FAQ bots and failed.

## Pricing

Userdesk prices in December 2023, before the sale, from an Internet Archive copy of its pricing page (grade C)

| Plan | Per month, billed yearly (US$) | Per month, billed monthly (US$) | Limit |
|---|---|---|---|
| Free | 0 | 0 | 1 chatbot, 30 AI messages a month, 50,000 training words |
| Hobby | 16 (190 a year) | 19 | 1 chatbot, 2,000 AI messages a month; monthly price from the founder's post |
| Growth | 41 (490 a year) |  | 2 chatbots, 4,000 AI messages a month, GPT-4, daily sync |
| Pro | 83 (990 a year) |  | 5 chatbots, 15,000 AI messages a month |
| Enterprise | 416 (4,990 a year) |  | 50 chatbots, 100,000 AI messages a month, no branding |
| Remove branding add-on |  | 39 | removes "Powered by Userdesk" |

The archived page showed each plan's price per month on the annual plan and the yearly total, with two months free on annual billing. The founder wrote that subscription plans started at US$19 a month; the other monthly prices were not shown in the archived copy. The model was GPT-3.5 by default, with GPT-4 on Growth and higher. A US$39 a month add-on removed the "Powered by Userdesk" mark. The current site belongs to a later owner and may have other prices.

## How much did Userdesk make, and what was it sold for? The numbers, graded

A one-person AI support chatbot that the founder said reached about US$1,100 a month in recurring revenue, sold nine months after launch at a price nobody disclosed

| Fact | Value | Grade | Note | Sources |
|---|---|---|---|---|
| Monthly recurring revenue before the sale | about US$1,100 | D | Founder's post of 4 April 2024: "$1.1k MRR it's difficult to achieve in less than 12 months." No Code Exits repeated it on 31 July 2026: "grew to about $1,100 MRR". One founder statement; no payment processor or buyer confirmed it. | [1, 2] |
| Run rate from that figure | about US$13,200 a year | calc | Calculated: US$1,100 times 12. | [1] |
| Sale | March 2024, through Acquire.com, buyer and price not disclosed | C | The founder announced the sale in April 2024 and wrote that he listed it on "the platform Acquired", got about 20 interested people and a couple of offers. No Code Exits and Starter Story give March 2024; No Code Exits names Acquire.com. Starter Story: "the buyer and sale price, were not disclosed." | [1, 2, 3] |
| Largest customer | an enterprise customer at US$500 a month | D | Founder: "it's an enterprise customer ($500/month) who subscribed", the spike in his September chart. | [1, 2] |
| Share of MRR from the enterprise customer | about 45.5% | calc | Calculated: US$500 / US$1,100. | [1] |
| Pre-sale before launch | 20 lifetime licenses at US$69 in 24 hours | C | Founder's post. Starter Story writes about 30 licenses; the founder's own number is used here. The founder says few of these buyers used the product. | [1, 3] |
| Revenue from the pre-sale | about US$1,380 | calc | Calculated: 20 licenses times US$69. | [1] |
| Launch | June 2023 | C | Founder: "I launched in June 2023 with a pre-launch and a subscription-based model a few weeks later." | [1, 2] |
| Main customer source | AI directories | C | Founder: "I found my customers mainly from AI directories." He adds that they also helped the domain's search ranking. | [1] |
| Founder's audience on X | about 20,000 followers, mostly developers | C | Founder's post and No Code Exits. Starter Story gives about 13,000 at the time of the pre-sale. | [1, 2, 3] |
| Time from launch to the sale | 9 months | calc | Calculated: June 2023 to March 2024. | [1, 2] |
| Founder's earlier exits | Iterspace (2021), Hivoe and Inboxs (2023) | C | Founder's post. Hivoe and Inboxs were sold to Hypefury after X raised its API prices. | [1] |

How to read this table: a grade D value rests on one statement, here the founder's own post, repeated by a newsletter. A grade C value is a statement by the founder or the company's own pages. A "calc" value was computed on this page from the rows above it. No value is grade A or B.

## When did Userdesk launch, and how did it grow? Timeline

From the chatbot wave of early 2023 to a sale through a marketplace, in nine months. Each entry carries its grade.

- **2021** (grade C [1]): Restagno sells Iterspace, a product built with a former colleague, before it has revenue.
- **2023** (grade C [1]): He sells Hivoe and Inboxs, two tools for X users, after X's new API prices make them hard to run.
- **February to March 2023** (grade C [1]): He starts Userdesk after the wave of interest in AI chatbots.
- **Before launch** (grade C [1]): A pre-sale of a US$69 lifetime deal sells 20 licenses in 24 hours, from posts to his X followers.
- **June 2023** (grade C [1, 3]): Launch, with subscriptions a few weeks later. Built in about three months with React and Next.js on Vercel and AWS.
- **September 2023** (grade D [1]): An enterprise customer subscribes at US$500 a month, the founder wrote.
- **December 2023** (grade C [4, 1]): The pricing page shows a free plan and four paid plans, GPT-3.5 by default and GPT-4 on higher plans. The founder launches Shipped.club, a Next.js boilerplate for developers.
- **March 2024** (grade D [1, 2, 3]): Sold through Acquire.com after a letter of intent, due diligence and an asset purchase agreement, with MRR at about US$1,100 by the founder's account. Price not disclosed.
- **4 April 2024** (grade C [1]): The founder publishes the story of the sale.
- **31 July 2026** (grade C [2]): No Code Exits retells the sale as a lesson in founder-product fit.

## Why did Userdesk work?

### Speed

The founder reused the stack of his earlier products (React, Next.js, Vercel, AWS) and shipped in about three months, while interest in website chatbots was at its peak in 2023.

### AI directories

He said most customers came from AI directories, the lists of new AI tools that many people browsed in 2023. The listings also brought links that helped the site rank in search with few blog posts.

### A wide price ladder

The archived pricing page listed plans from a free tier to an Enterprise plan of US$4,990 a year. The founder explained that one enterprise customer at US$500 a month moved the total quickly, because the entry plans were small.

### Selling instead of switching audiences

The founder's distribution was his audience of developers on X, and they were not the buyers of support chatbots. He said founder-product fit mattered more than he expected, sold the product through a marketplace, and moved to a product for developers.

## How it was built

These details come from the founder's post, the archived pricing page and Starter Story's summary. They are grade C.

- React and Next.js, hosted on Vercel, with Chakra UI components and AWS for background jobs, the same stack as his earlier products.
- OpenAI models: GPT-3.5 (the ChatGPT API) by default, GPT-4 on the Growth plan and higher.
- Training sources: websites, sitemaps, Zendesk help centers, GitBook, Notion, Google Drive and PDF files, with a daily automatic resync on Growth and higher.
- Delivery: an embeddable chat widget, a direct link and a Slack app.
- A Discord community for users, used for feedback and announcements.
- Separate accounts with a dedicated email for every service, a lesson from his first sale, so that the hand-off to the buyer was quick.

Sources: [1, 3, 4]

## Who tried this before Userdesk, and why did they fail?

Four companies in our database of failed startups sold chat windows or answer bots for support before large language models made them cheap to build. The values below are from that database and are not independently verified.

The failed startups come from our database of dead startups, retold here in our own words; Its entries are AI-assisted summaries of public sources, so they carry no evidence grade.

### Answerly (2013 to 2016; funding US$550K; cause: Product/Tech Failure)

- Promise: FAQ chatbots that would deflect customer support tickets so humans only handled the complex cases.
- Why it died: Language understanding was too weak at the time, and Zendesk and Intercom were already in place.
- Record: 

### Niles (2017 to 2020; funding US$2.8M; cause: Competition)

- Promise: A Slack bot that answered staff questions from the company's own knowledge.
- Why it died: Slack's own features and many cheaper competitors left it no room.
- Record: 

### Hipmob (2012 to 2015; funding US$120K; cause: Competition)

- Promise: Customer chat built into mobile apps and websites.
- Why it died: Intercom and Zendesk added the same features faster, and Hipmob had almost no funding.
- Record: 

### RightHello (2014 to 2022; funding US$600K; cause: Ran Out of Cash)

- Promise: Live chat for online shops that turns visitors into buyers with well-timed conversations.
- Why it died: Intercom raised far more money and won the chat tool market; RightHello could not afford to compete.
- Record: 

## What changed

Answerly and Niles tried to answer questions from a company's own knowledge when software could not read free text well, so building the bot was expensive and the answers were poor. In 2023 the OpenAI API made good answers cheap, and one engineer could build the whole product in three months without funding.

The same change let hundreds of competitors build the same product, which is the risk that ended Hipmob and RightHello. The founder named the crowded market and his unclear target customer as his main problems. He did not try to outlast the competition; he sold the product while it was still growing.

## Can one person copy Userdesk now?

**Verdict: Yes, as a small product to build and sell, not as a large business.**

A website chatbot is now a common product with many competitors, but the method is easy to repeat: reuse a stack you know, ship in weeks, list the product where AI buyers look, and sell it through a marketplace when you are not the right owner.

1. Reuse a stack you already know so that the first version ships in weeks.
2. List the product in AI directories, where buyers of new AI tools search.
3. Keep each service on a dedicated account so that a future sale is a quick hand-off.
4. If your audience does not match the product's buyers, sell it through a marketplace while it is growing.

**What is still open.** Answer bots for one niche with its own content sources, for example a bot trained on a software product's documentation or a shop's product catalog, sold to an audience the founder already has.

## What could go wrong for Userdesk?

- **What is not verified.** The figure of about US$1,100 in monthly recurring revenue rests on one statement by the founder, Luca Restagno, in his post of 4 April 2024; No Code Exits repeated it in July 2026 without new evidence. No payment processor, marketplace listing or buyer has confirmed it, and the sale price was not disclosed. A reader could look for the Acquire.com listing or a statement by the buyer.
- **One customer was almost half of the revenue.** By the founder's account, one enterprise customer paid US$500 a month, about 45.5% of the US$1,100. Losing that one contract would have cut recurring revenue almost in half.
- **Sources disagree on small details.** Starter Story's summary says about 30 pre-sale licenses and 13,000 followers; the founder's own post says 20 licenses and 20,000 followers. The founder says he used "the platform Acquired"; No Code Exits names Acquire.com.
- **Founder's marketing tactics.** Starter Story writes that the founder explored LinkedIn Ads and cold outreach for leads. The founder's own post names only his X audience and AI directories.
- **Dependence on one model provider.** The archived pricing page said the chatbots used OpenAI models. The founder had already sold two products because X raised its API prices, and he wrote that platform risk is real.
- **The current site belongs to the buyer.** userdesk.io is still online in October 2026 under the later owner. Its pages and prices are not used for the facts on this page, except to show that the product still exists.

## Sources

1. Luca Restagno, DEV Community, 4 April 2024: I've sold my 4th SaaS as a solopreneur. The backstory — https://dev.to/ikoichi/ive-sold-my-4th-saas-as-a-solopreneur-the-backstory-jcb (grade D). Read in full. The founder's own account, first published on his newsletter. Source for the MRR, the enterprise customer, the pre-sale, the channel and the sale process.
2. No Code Exits, 31 July 2026: His SaaS Was Growing. He Sold It Anyway, Because He Was the Wrong Founder for It — https://nocodeexits.substack.com/p/his-saas-was-growing-he-sold-it-anyway (grade D). Read in full. A newsletter retelling of the founder's account. Source for the March 2024 date and the name Acquire.com.
3. Starter Story: How Luca Restagno Grew Userdesk SaaS to $1K MRR in 12 Months — https://www.starterstory.com/userdesk-breakdown (grade C). Read the public part. A summary of the founder's statements. Source for the stack, the Discord community, the marketing channels and the undisclosed buyer and price.
4. Userdesk pricing page, Internet Archive copy of 10 December 2023 — https://web.archive.org/web/20231210110621/https://userdesk.io/pricing (grade C). Read in full. Source for the plans, limits, models and training sources before the sale.
5. Userdesk home page, Internet Archive copy of 30 January 2024 — https://web.archive.org/web/20240130203633/https://userdesk.io/ (grade C). Read in full. Source for what the product did before the sale.
6. Userdesk home page, read 31 October 2026 — https://userdesk.io/ (grade C). Read in full. The site under the later owner; used only to show that the product is still online.
7. Database of failed startups, local export — /method/#dead (grade C). Local export. AI-written summaries, not independently verified. Source for the failed predecessors.