Text2SQL.ai: Revenue, team and how it was built
Text2SQL.ai (Text-to-SQL web app, API and desktop app, Not stated in the sources): US$8.3K a month, July 2025. Evidence grade D: reported, not verified. Team: 1 founder, who calls himself a bootstrapped solopreneur; current headcount not published. Launched November 2022; still running.Added 7 November 2026
- Outcome
- US$8.3K a month, July 2025
- Evidence
- D reported, not verified
- Launch
- November 2022
- Status
- Still running
- Team
- 1 founder, who calls himself a bootstrapped solopreneur; current headcount not published
- Outside funding
- None reported
- Main channel
- Not stated
What is not verified
The headline of US$8,333 a month rests on one founder directory. ARRFounder listed it in a record extracted by software on 18 July 2025 and says its data comes primarily from founders' posts on X; the original post was not found, so the figure may be older than July 2025. IdeaWave shows the same number, labelled self-reported, and its flat daily line from August to November 2026 only repeats the ARRFounder record. No TrustMRR profile exists at trustmrr.com/startup/text2sql (not found on 7 November 2026). A reader could look for a payment-processor-verified profile, or a dated post by the founder with a revenue screenshot. The headline number is grade D: one claim or one estimate, with no second source. Aiolum shows it because the story is useful, and marks it so that it is not mistaken for a verified figure.
Text2SQL.ai turns a question such as "how many users signed up in the last 12 months, by month?" into SQL, Structured Query Language, the language used to read and change data in relational databases. The user can add a database schema, the list of tables and their columns, or connect a database, so the generated query uses the real table names. It also sells an API, an interface that lets other programs call the same function, and a desktop app.
Gary Miklos, who describes himself on Product Hunt as a "bootstrapped solopreneur" and a software engineer, had the first version online by 6 November 2022 as a free beta built on OpenAI's GPT-3 Codex model, with a "buy me a coffee" link for support. Today the app is a subscription after a 7-day free trial, and the company's documentation lists a team seat at US$29 a month. The founder directory ARRFounder listed Text2SQL on 18 July 2025 at US$8,333 a month, or US$100,000 a year, and IdeaWave still shows that figure.
That headline is grade D: one directory record, which ARRFounder says it extracted with software from public founder posts, with no payment-processor check and no original founder post found. The same record labels the product "sold", which no other source confirms. The page also shows earlier companies that tried to let people ask data questions without learning a query language.

How much revenue does Text2SQL.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 |
|---|---|---|---|---|
| Revenue per month, July 2025 | US$8,333 (US$100,000 a year) | D | ARRFounder product record for Text2sql, founder Gary (@gary_miklos). The record says it was extracted on 18 July 2025 with the source type "ai"; ARRFounder says its data is collected primarily from founders' posts on X. The founder's own post was not found. | [1] |
| Run rate from the July 2025 figure | US$99,996 a year | calc | Calculated: US$8,333 times 12. ARRFounder rounds it to US$100,000. | [1] |
| Status in the ARRFounder record | "sold" | D | The page shows a "Sold" label. No buyer, price or date is given, and no other source mentions a sale. | [1] |
| Same figure on IdeaWave, November 2026 | US$8.3K a month, "self-reported by the founder" | D | IdeaWave names ARRFounder as its source and says it re-checks it every morning; its daily line is flat from 1 August 2026, when tracking began, to 7 November 2026. That is a copy of the July 2025 record, not a new measurement. | [2] |
| First public version | online by 6 November 2022, free | C | Earliest Internet Archive copy. The FAQ said: "We are using the OpenAI GPT-3 Codex model" and that the app stays free while that model's API is free, and asked for support through a "buy me a coffee" link. | [3] |
| Founder | Gary Miklos, "Bootstrapped solopreneur" | C | His Product Hunt profile, opened on 8 November 2022, links to Text2SQL.AI and says he is a software engineer and an indie maker. ARRFounder and IdeaWave give the founder as Gary, @gary_miklos. | [4, 1, 2] |
| Goal stated in the founder's profile | "build 5+ SaaS products in 2023 to reach 10k MRR" | C | Product Hunt profile text. It is a goal, not a result. | [4] |
| Price of a Pro team seat | US$29 a month or US$239 a year | C | Teams documentation. Added or removed seats are prorated; a Stripe invoice shows each added seat. | [6] |
| API allowance and price | 100 requests a month on Pro, then US$0.05 per request | C | Home page FAQ and the Get started documentation. The API is only for Pro plan users and above. | [5, 7] |
| Free trial | 7 days, all features | C | Home page FAQ: after the trial a subscription is needed. | [5] |
| AI model provider | OpenAI | C | Home page FAQ, in the answer on data safety. The first version used OpenAI's GPT-3 Codex. | [5, 3] |
| Hosting, database and login | Vercel; Supabase | C | The terms of service say the site is hosted on Vercel. The FAQ points to Supabase's security policy for the database and authentication providers. | [8, 5] |
| Releases in late 2025 and 2026 | API chat demo and comparison guide (27 October 2025), Safe Mode (6 November 2025), Insights (20 November 2025), charts in the API (17 March 2026) | C | Company blog list. Safe Mode lets generated queries only read the database, never change or delete data. | [9] |
| Logo wall on the home page | 18 company and university logos under "Trusted by Leading Companies" | C | Among them Google, Airbnb, PayPal, Spotify, Bloomberg, Harvard, MIT and Stanford. No case study or link supports them. | [5] |
| Named customer case study | SimplyFleet | C | Its founder is quoted: "Most requests generate the right SQL on the first try." | [11] |
| Outside funding | none reported | C | The founder calls himself bootstrapped. No funding round was found. | [4] |
| Time from launch to the July 2025 figure | 32 months | calc | Calculated: November 2022 to July 2025. | [3, 1] |
How to read this table: a grade D value comes from one founder directory, ARRFounder, or from IdeaWave, which copies it; nobody has checked it. A grade C value is the company's own page or the founder's own profile. A "calc" value was computed on this page from the rows above it. No value is grade A or B.
| Plan | Per month, billed yearly (US$) | Per month, billed monthly (US$) | Limit |
|---|---|---|---|
| Free trial | 0 | 0 | 7 days, all features |
| Pro, per seat | 19.92 | 29 | unlimited messages; 239 a year per seat; 100 API requests a month shared by the team |
| Extra API requests | 0.05 per request | 0.05 per request | above the 100 included each month |
| Enterprise | custom | custom | custom solution and limits |
The Teams documentation gives the price of each seat on the Pro plan, US$29 a month or US$239 a year; it does not state the owner's own Pro price separately and says the owner's seat is already counted. The documentation also names a Free or Basic plan without a price. API access starts with Pro and includes 100 requests a month, with more at US$0.05 each.
When did Text2SQL.ai launch, and how did it grow?
From a free beta on OpenAI's GPT-3 Codex in November 2022 to a subscription with seats, an API and charts in 2026. Each entry carries its grade.
- 6 November 2022CEarliest archived copy of text2sql.ai: a free beta that translates plain English to SQL with OpenAI's GPT-3 Codex model, with a box for the user's own tables and a donation link. [3]
- 8 November 2022CGary Miklos opens his Product Hunt profile with a link to Text2SQL.AI and the goal of reaching 10k MRR with 5 or more SaaS products in 2023. [4]
- 15 February 2025CLast update of the terms of service; the site is operated as "Text2SQL.ai" and hosted on Vercel. [8]
- 18 July 2025DARRFounder lists Text2sql at US$8,333 a month and US$100,000 a year, with the status "sold". [1]
- 27 October 2025CThe company publishes an open-source chat demo for its API and a comparison of seven text-to-SQL tools. [9, 10]
- 6 November 2025CSafe Mode: generated queries can only read the database. [9]
- 20 November 2025CInsights: the query, the result table, a chart and an explanation in one view. [9]
- 17 March 2026CChart generation added to the API. [9]
- 1 August to 7 November 2026DIdeaWave tracks the ARRFounder figure every day; it does not change. [2]
Why did Text2SQL.ai work?
Four causes, in the order they mattered. Each one is a reusable lesson.
Early on a new model
The first version was online by 6 November 2022, built on OpenAI's GPT-3 Codex while that model's API was in a free beta. It was free to use and asked only for donations.
A name that is the search term
The domain text2sql.ai is the phrase people type. The site footer lists one page per database (MySQL, PostgreSQL, Oracle PL/SQL, Microsoft SQL Server) and per task (fix, explain and optimize SQL, spreadsheet formulas, regular expressions). No source says how many customers arrived through search, so the channel on this page is unknown.
The schema, not the data
Only table and column names go to the model, the company's guide says, and a desktop app runs against a local database so the credentials never leave the laptop. Safe Mode blocks queries that would change or delete data. These answer the first question a company's security reviewer asks.
Seats and an API
After a 7-day trial the app is a subscription, and the documentation lists each added team seat at US$29 a month. The API, included from the Pro plan, lets a customer build SQL generation into its own product, which the SimplyFleet case study describes.
How it was built
These details come from the company's own pages and an archived copy of the first version. They are grade C. [3, 5, 8, 6, 9, 7]
- The first version sent the user's sentence, and optionally the user's table list, to OpenAI's GPT-3 Codex model, the archived FAQ says.
- Today the FAQ says the system sends requests to OpenAI, and it points to Supabase for the database and login.
- The website is hosted on Vercel, the terms of service say.
- Payments: the Teams documentation says each added seat appears on a Stripe invoice and is prorated.
- An API with the same functions, plus an open-source chat demo released in October 2025 that shows how to call it.
- A desktop app runs the same AI against a local database, so database credentials stay on the user's laptop.
- Separate landing pages for each database and each task, listed in the site footer.
Tools named in the sources
Each one opens its page in the Tools tab, with every story that uses it.
Who tried this before Text2SQL.ai, and why did they fail?
Four companies in our database of failed startups promised that people could work with data without learning a query language or a programming tool. The values below are from that database and are not independently verified.
Ask Jeeves
The promiseAsk the search engine a question in plain language instead of typing keywords and operators.
Why it diedHuman editors wrote the answers and could not keep up with the web, while Google's ranking by algorithm cost almost nothing per query.
Fieldbook
The promiseSpreadsheets with the power of a database, without needing to know SQL.
Why it diedExcel and Google Sheets added database-like features, and users would not switch away from them.
Skysheet
The promiseA cloud spreadsheet with real-time collaboration and better charts.
Why it diedGoogle and Microsoft moved their own spreadsheets to the cloud, and Skysheet had no clear difference left.
DataNitro
The promiseRun Python scripts inside Excel for data analysis.
Why it diedIt was limited by Excel's design, and Microsoft later added Python to Excel itself.
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
All four tried to put a simpler layer between a person and data, and all four lost to a larger product that added the same feature or did the job more cheaply. Ask Jeeves is the closest: it answered plain-language questions, but people had to write the answers by hand. Text2SQL.ai does the translation with a language model, so the cost of one more question is one model call, not one more editor.
The risk of the predecessors is still there. ChatGPT itself writes SQL, and the company's own 2025 guide says it is many users' first stop. What Text2SQL.ai adds is the context that a general chat window lacks, the customer's own schema and database connection, plus the controls that companies ask for: read-only Safe Mode, metadata-only processing, team seats and a local desktop app.
Can one person copy Text2SQL.ai now?
The product, yes. The head start, no.
A text-to-SQL app is a small, well-defined product that one engineer can build on a model API, and this one started as a free beta with a donation link. The early position of November 2022 is gone: the company's own guide compares it with six other tools, and general chat assistants write SQL too.
Where it is still open: The same translation for other query languages that people find hard: spreadsheet formulas and regular expressions are already on its own menu; log search, analytics event queries and GraphQL are similar.
- 1Pick one task that a general chat assistant does badly without context, and supply the context: here, the customer's database schema.
- 2Send only metadata to the model and offer a local version, so a company's security review can approve it.
- 3Sell seats and an API, so that one user can grow into a team account or an integration.
- 4Make one landing page per variant of the task (each database, fix, explain, optimize).
What could go wrong for Text2SQL.ai?
The risks are the mirror image of the causes above.
What is not verified
The headline of US$8,333 a month rests on one founder directory. ARRFounder listed it in a record extracted by software on 18 July 2025 and says its data comes primarily from founders' posts on X; the original post was not found, so the figure may be older than July 2025. IdeaWave shows the same number, labelled self-reported, and its flat daily line from August to November 2026 only repeats the ARRFounder record. No TrustMRR profile exists at trustmrr.com/startup/text2sql (not found on 7 November 2026). A reader could look for a payment-processor-verified profile, or a dated post by the founder with a revenue screenshot.
A directory marks it as sold
The ARRFounder record has the status "sold" and the page shows a "Sold" label. No buyer, price or date was found, the terms of service name only "Text2SQL.ai" as the operator, and the site kept releasing features until March 2026. This page treats the business as running; if it was sold, the founder named here may no longer run it.
Logo wall and testimonials without sources
The home page shows 18 logos under "Trusted by Leading Companies", among them Google, Airbnb, PayPal, Spotify, Harvard, MIT and Stanford, with no case study or link for any of them. The testimonials name people with job titles but link to nothing. The one case study read for this page is with SimplyFleet, whose founder is quoted by name.
Competition
AI2SQL, a solo text-to-SQL tool already on this site, reported US$10,000 a month in January 2025 at grade C. The company's own October 2025 guide compares Text2SQL.ai with ChatGPT, SQLAI.ai, AI2SQL, MindsDB, Vanna AI and Wren AI, and says SQLAI.ai starts at US$5 a month.
Dependence on OpenAI
The FAQ says the system uses OpenAI for processing, and the first version ran on OpenAI's GPT-3 Codex beta, which the old FAQ tied to the product being free. A price change, a model retirement or a policy change at OpenAI reaches the product directly, and ChatGPT, from the same provider, is also a competitor for simple SQL questions.
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.
- D ARRFounder: Text2sql | $8K/mo (product record, extracted 18 July 2025)Read in full, including the page's embedded record (mrr 8333, arr 100000, extracted_at 2025-07-18, status sold, source ai) and the directory's disclaimer that founders' data comes primarily from X.
- D IdeaWave: Text2sql revenue: $8.3K MRR (self-reported by the founder)Read in full on 7 November 2026. It names ARRFounder as its source and shows the daily checks since 1 August 2026.
- C Internet Archive copy of text2sql.ai, 6 November 2022Read in full. The earliest snapshot in the archive's index; source for the free beta, the GPT-3 Codex model and the donation link.
- C Product Hunt: Gary Miklos' profileRead the profile text and its embedded data: the bio, the link to Text2SQL.AI and the date the profile was opened.
- C Text2SQL.ai home page, read 7 November 2026Read in full, including the FAQ answers in the page data. Source for the trial, the API allowance, OpenAI and Supabase, the logo wall and the testimonials.
- C Text2SQL.ai documentation: TeamsRead in full. Source for the seat price and the Stripe invoice line.
- C Text2SQL.ai documentation: Get startedRead in full. Source for the Pro and Enterprise plan features and the API price.
- C Text2SQL.ai Terms of Service, last updated 15 February 2025Read in full. Source for the operator name and Vercel hosting.
- C Text2SQL.ai blog: All articlesRead in full. Source for the release dates from October 2025 to March 2026.
- C Text2SQL.ai blog, 27 October 2025: Best SQL AI Tools in 2025: Complete Comparison GuideRead in full. The company's own comparison; source for the competitors, metadata-only processing and the remark that ChatGPT is many users' first stop.
- C Text2SQL.ai case study: SimplyFleetRead in full. A customer quote published by the company.
- C Text2SQL.ai /pricing, read 7 November 2026Read in full. It shows only "Under Construction"; prices on this page come from the documentation instead.
- 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