# Bankstatemently

Aiolum story: https://aiolum.com/story/bankstatemently/ · added 2026-11-12 · updated 2026-11-12

**In short.** Bankstatemently (Bank statement converter (PDF to CSV, Excel, QBO), Singapore): US$375 monthly recurring revenue, August 2026. Evidence grade A: disclosed by a company or a filing, or verified by a payment processor. Team: 1 founder, who calls himself a solopreneur on the About page. Launched August 2025; still running, revenue growing.

Bank statement converter (PDF to CSV, Excel, QBO) · Singapore · AI-native · solo founder · still running, revenue growing · group: Finance, risk and compliance

- Outcome: US$375 monthly recurring revenue, August 2026 (evidence grade of the headline number: A; headline in US dollars: 4500)
- Launch: August 2025; months from launch to the outcome: 12
- Team: 1 founder, who calls himself a solopreneur on the About page
- Outside funding: None found in any source
- Main factor: Niche
- Meters, 1 to 4: evidence strength 3, copyable pattern 3, capital needed 1
- Tags: AI-native, Solo founder, Running, B2C, Fintech

Links:

- Visit bankstatemently.com: https://bankstatemently.com/
- Verified revenue on TrustMRR: https://trustmrr.com/startup/bankstatemently

**The product.** A web service that reads a bank or credit card statement in PDF form and returns the transactions as a spreadsheet file. The outputs are CSV, Excel, QBO (QuickBooks), Xero and JSON. The company also sells an API and a connector for AI assistants.

**Why it worked.** It serves one narrow job, turning a statement PDF into clean rows, and it grew through pages written for single banks. The founder wrote in March 2026 that almost all clicks come through those bank pages. A free tier and a small credit pack let people try it before they subscribe.

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; E = no revenue figure found, so the headline is a non-money fact and the other facts carry their own grades; calc = computed on this page from graded facts.

## Summary

Bankstatemently is a bank statement converter. It reads the table inside a statement PDF. It writes the rows in a form that a spreadsheet or accounting program can import. The founder is Michael Duyvesteijn, who describes himself on the About page as a solopreneur based in Singapore. He wrote on 25 January 2026 that it had been "a good 5 months" since he started working on financial document processing. The product combines a rule-based parser with a large language model (a program that predicts text and can read documents), according to the About page.

The first sale came in January 2026. The payment processor Stripe records US$25 for that month. TrustMRR is a site that reads revenue from payment processors. It reports US$458 for July 2026 and US$1,420 in total. At its last data sync on 4 August 2026 it reports a monthly recurring revenue of US$375 from 14 active subscriptions. The founder wrote on 8 March 2026 that revenue was flat at that time.

The headline figure is grade A because TrustMRR reads it from Stripe through an API key. That key expired on 5 August 2026, so no figure after 4 August is verified. The home page counters, such as "4.1 from 111 user ratings" and "99.9% accuracy", are the company's own statements and are noted under "What to watch".

## Why did Bankstatemently work?

### A plain converter let one founder see many statements

The founder wrote on 8 March 2026 that he first positioned the product simply as a way to convert a statement PDF to CSV. In his words, this was to build muscle: to see many statements and improve the product. He then moved toward jobs such as lending checks, accounting imports and fraud signals.

### Bank pages brought almost all of the clicks

On 8 March 2026 the founder wrote that the PDF-to-Excel space is crowded and that he was building bank-specific pages for the long tail. He wrote that almost all clicks came through them. The same letter reports that Bing traffic fell on 17 February. He names two causes: 308 redirects used instead of 301, and pages for banks not yet processed.

### A free tier and a small pack lower the first purchase

The free plan gives 10 pages a month with no credit card, and a credit pack of 50 pages costs US$7.50. The company says free credits resume after a pack expires. The founder wrote that the first two monthly subscribers seemed to be small businesses with several accounts.

### The founder answered the trust question with a public benchmark

The founder released the first 5 of 15 planned synthetic statements with a free scoring API. The benchmark page gives 15 March 2026 as its date, and his letter about it is dated 18 March. He wrote in that letter that he built the evaluation system before the product features. On 8 March he wrote that he was not adding fake trust signals. The company also offers an API and a Claude Code plugin, which fits the B2B evaluation he described that day.

## How it was built

These details come from the company's About page, privacy policy and the founder's letters. They are grade C.

- A dual engine: rule-based parsing for determinism, and parsing with a large language model for flexibility. The About page says the two reinforce each other.
- The privacy policy lists Anthropic, OpenAI and open-source models for document processing. It names Supabase for the database, Vercel and Fly.io for hosting, Clerk for sign-in and Stripe for payments.
- The founder wrote on 25 January 2026 that he bought a Claude Code Max subscription. He used it with Opus 4.5 to build, including the bank pages for search.
- A new language needed little work. On 8 February 2026 he wrote that adding a banking dictionary let the tool read a Canadian French statement 15 minutes later.
- Outputs: CSV, Excel, QBO, Xero and JSON, an API, an MCP server and a Claude Code plugin. The company has published an open benchmark on GitHub.

Sources: [3, 4, 5, 8, 2, 10, 15]

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

Three companies in our database of failed startups tried to automate financial paperwork or spreadsheets. The values below are from that database, read on 12 November 2026, 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.

### Lendsnap (2016 to 2019; funding US$500K total cash burned; cause: Competition)

- Promise: Collect borrowers' financial documents for a mortgage application automatically, by connecting to financial institutions.
- Why it died: The record says larger mortgage platforms such as Blend and Roostify offered end-to-end products, and a narrow focus on document automation was not enough.
- Record: 

### Sway Finance (2016 to 2019; funding US$2.0M total cash burned; cause: Competition)

- Promise: An AI platform for expense tracking, budgeting and forecasting for small businesses.
- Why it died: The record says it could not stand out against QuickBooks and Xero, relied on paid advertising and did not raise a later round.
- Record: 

### DataNitro (2012 to 2015; funding US$120K total cash burned; cause: Competition)

- Promise: Run Python scripts inside Excel, to move spreadsheet work to code.
- Why it died: The record says it depended on Excel, and Microsoft later built Python into Excel itself.
- Record: 

## What changed

All three built a narrow tool on top of another company's product or against bigger suites. The records say that Blend, Roostify, QuickBooks, Xero and Microsoft covered the same job later or better. Bankstatemently has the same exposure: the founder wrote that any vision-language model can now turn a PDF into a CSV.

The difference is the cost of the work. One person built and runs it, with AI coding tools, and the model providers do the reading. The founder says his answer is accuracy, auditability and repeatability, because a small error in financial data is costly. That is his claim; the sources do not test it.

## Can one person copy Bankstatemently now?

**Verdict: Yes for the method, with small expectations.**

A narrow converter for one document type can reach paying customers within five months of the start, with one person. The verified revenue is small: US$375 a month at the last sync. The steps below are the ones the founder's letters describe.

1. Pick one document type that people struggle to copy out of a PDF, and ship a free version first.
2. Write a page for each bank or form, and publish only for the ones you have processed, as the Bing fall of February 2026 shows.
3. Offer a free tier, a small one-time pack and monthly plans.
4. Publish how you measure accuracy, and add an API for customers who want to automate.

**What is still open.** Other documents that arrive as PDFs and need structured rows: payslips, utility bills, tax forms, shipping documents and insurance statements.

## How much revenue does Bankstatemently make? The numbers, graded

A US$25 first month in January 2026, then US$458 in July 2026, read from Stripe by TrustMRR. The profile stops on 4 August 2026

| Fact | Value | Grade | Note | Sources |
|---|---|---|---|---|
| Monthly recurring revenue, 4 August 2026 | US$375 | A | TrustMRR profile, read from Stripe. Last data sync 4 August 2026. The Stripe key expired on 5 August 2026. | [1] |
| Run rate from that MRR | US$4,500 a year | calc | Calculated: US$375 times 12. | [1] |
| Active subscriptions, 4 August 2026 | 14 | A | TrustMRR profile. The average is about US$26.79 a subscription a month. | [1] |
| Verified revenue in total | US$1,420 | A | TrustMRR profile, all time to 4 August 2026. The monthly rows add up to US$1,421, a rounding difference. | [1] |
| Verified revenue, last 30 days to 4 August 2026 | US$433 | A | TrustMRR profile, shown as 3.4% lower than the 30 days before. | [1] |
| Verified revenue by month, 2026 | January 25, February 70, March 25, April 120, May 245, June 388, July 458, August 90 (to the 4th), all in US dollars | A | TrustMRR monthly timeline, read from Stripe. | [1] |
| First revenue | January 2026 | A | TrustMRR shows US$25 for January. The founder's letter of 25 January 2026 says he made his first revenue that week. | [1, 4] |
| Start of the work | about August 2025 | calc | Calculated: the founder wrote on 25 January 2026 that it had been "a good 5 months" since he started. TrustMRR gives 29 September 2025 as the founded date. | [1, 4] |
| Company | Uncommon Goods Lab Pte. Ltd., UEN 202509282G | C | Footer of the company's pages. No earlier name or earlier product of the same company was found on its pages. | [2, 7] |
| First two paid monthly subscribers | week of 8 February 2026 | C | Founder's letter of 8 February 2026. He wrote that both seemed to be small businesses with several accounts at several banks. | [5] |
| Weekly visitors, 28 February 2026 | close to 200, in more than 10 countries | C | Company update post of 28 February 2026. | [12] |
| Banks supported | 400 banks in 46 countries or regions, 15 languages | C | Home page counters, the company's own statement. | [2] |
| Rating counter on the home page | 4.1 from 111 user ratings | C | The privacy policy says a rating is a 1 to 5 star rating of a conversion or of the service. It does not count customers. | [2, 8] |
| Prices | free 10 pages a month; credit pack US$7.50; Beginner US$30 and Pro US$70 a month | C | Pricing page structured data, November 2026. | [7] |
| Audience on the website | accountants, lawyers, lenders, compliance staff and small business owners | C | Home page use cases. The terms require users to be at least 18 years old. | [2, 11] |
| Time from start to the headline figure | 12 months | calc | Calculated: August 2025 to August 2026. | [1, 4] |

How to read this table. Grade A is a figure that TrustMRR reads from the company's Stripe account. Grade C is a statement by the founder or the company on its own pages. A "calc" value was computed on this page from the rows above it.

## Pricing

Bankstatemently prices in November 2026, read from the structured data on the company's pricing page and its July 2026 update post (grade C)

| Plan | Per month, billed yearly (US$) | Per month, billed monthly (US$) | Limit |
|---|---|---|---|
| Free |  | 0 | 10 pages a month, 5 pages a day |
| Credit pack, one-time | 7.50 one-time |  | 50 pages, valid 30 days |
| Beginner |  | 30 | 400 pages a month, 20 pages per upload |
| Pro |  | 70 | 1,000 pages a month, 50 pages per upload; the July 2026 post says US$50 a month |
| Enterprise | custom |  | from 5,000 pages a month, with API access |

The pricing page draws its price tags by script, so the monthly prices come from its structured data. The yearly price was not readable. The July 2026 update post calls the Pro plan US$50 a month, and TrustMRR lists subscriptions from US$25 a month. Page limits are per month. The credit pack is valid for 30 days.

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

From a first sale in January 2026 to US$458 in a month by July 2026. Each entry carries its grade.

- **About August 2025** (grade C [4]): The founder starts working on financial document processing. He wrote five months later that it had been "a good 5 months".
- **29 September 2025** (grade C [1]): TrustMRR gives this as the founded date of Bankstatemently.
- **January 2026** (grade A [1, 4]): First revenue: US$25 in the month. The founder calls it a sweet moment in his letter of 25 January.
- **8 February 2026** (grade C [5]): The first two paid monthly subscribers arrive. Both push the system to its limits.
- **17 February 2026** (grade C [6]): Traffic from Bing falls overnight after a page restructuring. The founder names two causes in his letter of 8 March.
- **8 March 2026** (grade C [6]): A B2B customer is evaluating the product. An API is shipped on the day it was asked for. The founder writes that revenue is flat.
- **15 March 2026** (grade C [10, 15]): The company publishes an open benchmark with the first 5 of 15 planned synthetic bank statements. The founder's letter about it is dated 18 March.
- **June 2026** (grade C [13]): The company introduces the Bankstatemently Workspace: bulk upload and a search of past conversions.
- **8 July 2026** (grade C [9]): Update 4: several statements in one PDF, scanned checks, files up to 50 pages and a plugin for Claude Code.
- **July 2026** (grade A [1]): US$458 of verified revenue in the month, the highest so far.
- **4 August 2026** (grade A [1]): Last TrustMRR sync: US$375 monthly recurring revenue and 14 active subscriptions. The Stripe key expires the next day.

## What could go wrong for Bankstatemently?

- **Verified figures end on 4 August 2026.** The Stripe key that TrustMRR uses expired on 5 August 2026, so every figure stops on 4 August. The live profile read on 12 November 2026 shows US$0 for the last 24 hours and the last 3 months. No source says whether sales continued, so the trend is from March to July 2026 only.
- **The rating counter counts ratings, not customers.** The home page shows "4.1 from 111 user ratings", and TrustMRR shows 14 active subscriptions. The privacy policy says a rating is a 1 to 5 star rating of a conversion or of the service. It does not say that only subscribers rate. The rating form at /rate is disallowed for AI crawlers in robots.txt and was not read.
- **Home page testimonials are first names only.** Three testimonials carry a first name and a country (Larissa, Canada; Kathleen, United States) or only "A customer", with no photo. They were found only on the home page among the pages read. No web search was possible, so they were not checked elsewhere. The founder wrote on 8 March 2026 that he was not adding fake trust signals.
- **Accuracy figures come from the company's own benchmark.** The home page says "99.9% accuracy, verified across 81 banks". The benchmark page lists the company as the benchmark author. It lists 15 synthetic statements that the company made: 5 can be downloaded and 10 are marked "Coming soon". The company's score of 99.9% there covers those 5 statements. The comparison with Claude Sonnet 4 and ChatGPT 5 on the home page uses one DBS statement of 5 pages.
- **Other home page labels are the company's own statements.** The home page shows a "GDPR compliant" label, "400+ banks supported" and "15 languages". No independent source confirms them.
- **The company markets the tool to lenders.** Lending is a regulated activity, and the company sells to it as one use. Its loan underwriting page says the tool helps to "verify borrower income" and "make faster credit decisions". The same page quotes an unnamed "earned wage access fintech", and no source names that company.

## Sources

1. TrustMRR profile: Bankstatemently (markdown) — https://trustmrr.com/startup/bankstatemently.md (grade A). Read the markdown profile on 12 November 2026. Revenue, MRR and subscriptions are read from Stripe by API key. The founded date and the description are founder-written.
2. Bankstatemently home page — https://bankstatemently.com/ (grade C). Read in full on 12 November 2026, with the structured data. Source for the counters, the testimonials, the footer and the product claims.
3. About Bankstatemently, by Michael Duyvesteijn — https://bankstatemently.com/about (grade C). Read in full on 12 November 2026. The founder's own account of the team and the dual parsing engine.
4. Founder's letter #1, 25 January 2026: This week I made my first revenue — https://bankstatemently.com/blog/founders-letter-2026-01-25-first-revenue (grade C). Read in full on 12 November 2026. Source for the start, the first revenue and the tools used to build.
5. Founder's letter #2, 8 February 2026: Momentum is growing — https://bankstatemently.com/blog/founders-letter-2026-02-08-momentum (grade C). Read in full on 12 November 2026. Source for the first two monthly subscribers and the French language test.
6. Founder's letter #4, 8 March 2026: Revenue is flat, but we're getting better — https://bankstatemently.com/blog/founders-letter-2026-03-08-revenue-is-flat (grade C). Read in full on 12 November 2026. Source for the positioning, the bank pages, the Bing fall and the B2B evaluation.
7. Bankstatemently pricing page — https://bankstatemently.com/pricing (grade C). Read on 12 November 2026. The price tags load by script, so the prices come from the structured data and the plan limits from the page text.
8. Bankstatemently privacy policy — https://bankstatemently.com/privacy (grade C). Read in full on 12 November 2026. Source for the providers, the meaning of a rating and the processing of documents.
9. The Statement #4, 8 July 2026: Longer statements, scanned checks, and a Claude Code plugin — https://bankstatemently.com/blog/statement-upgrades-jul-2026 (grade C). Read in full on 12 November 2026. Source for the 50-page limit and the Pro plan price of US$50 a month.
10. Bank Statement Parsing Benchmark — https://bankstatemently.com/benchmark (grade C). Read the top section and the statement list on 12 November 2026. Source for the publication date, the 5 released and 10 planned statements, and the leaderboard.
11. Bankstatemently terms of service — https://bankstatemently.com/terms (grade C). Read in full on 12 November 2026. Source for the age limit of 18 and the refund terms.
12. The Statement #1, 28 February 2026: Your very first statement from us — https://bankstatemently.com/blog/dashboard-launch-feb-2026 (grade C). Read the opening on 12 November 2026. Source for the weekly visitors.
13. Bankstatemently blog index — https://bankstatemently.com/blog (grade C). Read in full on 12 November 2026. Source for the dates of the letters and of the Workspace update in June 2026.
14. Database of failed startups, local export — /method/#dead (grade C). Record pages read on the website on 12 November 2026. AI-written summaries, not independently verified. Source for the failed predecessors.
15. Founder's letter #5, 18 March 2026: Bankstatemently Open Benchmark for Bank Statement Parsing — https://bankstatemently.com/blog/bankstatemently-open-benchmark-bank-statement-parsing (grade C). Read in full on 12 November 2026. Source for the first 5 released statements and for the evaluation system built first.
16. Bankstatemently use case page: loan underwriting — https://bankstatemently.com/use-cases/loan-underwriting (grade C). Read the top half on 12 November 2026. Source for the marketing to lenders and the unnamed quotations.

Contact: aiolum@livyden.com. Run Bankstatemently? Send corrections or a verified figure to aiolum@livyden.com (subject: Correction: Bankstatemently).