Dynadok: Revenue, team and how it was built
Dynadok (AI validation of company documents, Brazil): R$1M (US$186K) revenue in 2024 as reported by the company, October 2025. Evidence grade D: reported, not verified. Team: Founder and CEO Willian Valadão with partner and COO Rodrigo Grossi at the launch in September 2024; the company site lists six partners in November 2026; total headcount not published. Launched May 2024; still running.Added 5 November 2026
- Outcome
- R$1M (US$186K) revenue in 2024 as reported by the company, October 2025
- Evidence
- D reported, not verified
- Launch
- May 2024
- Status
- Still running
- Team
- Founder and CEO Willian Valadão with partner and COO Rodrigo Grossi at the launch in September 2024; the company site lists six partners in November 2026; total headcount not published
- Outside funding
- None as of October 2025 (company statement); the founder said he planned to seek investment in 2026 for expansion to the European Union
- Main channel
- Not stated
What is not verified
The headline, R$1 million of revenue in 2024, rests on statements by the company. Willian Valadão gave it to Diário do Comércio on 28 October 2025, and IT Forum's article of 29 July 2025, which uses a company photo and quotes, said the company had passed R$1 million of annual revenue. The two do not agree in detail: IT Forum's "700%" growth to R$3 million implies about R$375,000 in 2024, while Diário do Comércio's R$1 million implies tripling. A company that the founder said was born in May 2024 would have reached R$1 million in about eight months. No payment processor, filing or auditor confirms either figure. A reader could look for the company's actual 2025 result in later press or a funding announcement. 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.
Dynadok is a Brazilian company in intelligent document processing (IDP), software that combines optical character recognition (reading text from images and scans) with machine learning and language models to read a document, understand its fields and decide whether it meets a rule. Its customers are large companies in construction, pulp and mining, education and health. A construction company, for example, must confirm that each subcontractor has paid its workers' social security before it releases a payment; Dynadok reads those documents and lists the problems in seconds.
Willian Valadão, a software entrepreneur from Belo Horizonte, says on the company site that the idea came during a visit to the UK technology scene in 2023. The startup site San Pedro Valley announced the launch in September 2024, with Rodrigo Grossi as partner and COO and two construction companies as named customers. In October 2025 Valadão told the newspaper Diário do Comércio that the company was born in May 2024, had 20 clients and had billed R$1 million (about US$186,000) the year before, with no outside investment. In July 2025 the trade site IT Forum had reported that the company projected R$3 million of revenue for 2025.
The revenue figures are statements by the company to two outlets, and they do not agree with each other in their details, so this page grades the headline D, reported but not verified. The R$3 million for 2025 was a projection; no source found states the result. The page also shows earlier companies that tried to replace paperwork and contract checks with software and failed.

How much revenue does Dynadok 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 2024, as stated in October 2025 | R$1 million (about US$186,466) | D | Diário do Comércio, 28 October 2025, interview with Willian Valadão: "No ano passado, a Dynadok registrou faturamento de R$ 1 milhão, e a expectativa é encerrar 2025 com R$ 3 milhões" (translation: last year Dynadok billed R$1 million, and it expects to close 2025 with R$3 million). Company statement, not checked by the newspaper. | [1] |
| The figure in US dollars | about US$186,466 a year, or about US$15,539 a month on average | calc | Calculated: R$1,000,000 divided by 5.3629 reais per US dollar, the Federal Reserve H.10 rate for 28 October 2025, the date of the article; divided by 12 for the monthly average. | [1, 8] |
| Revenue claim, July 2025 | above R$1 million of annual revenue | D | IT Forum, 29 July 2025, subheading: "Sem investimento externo, empresa faz parte dos 5% das startups com até três anos de vida [que] consegue superar a marca de R$ 1 milhão em receita anual" (translation: with no outside investment, the company is among the 5% of startups up to three years old that pass R$1 million of annual revenue). The article does not say over which twelve months. | [2] |
| Projection for 2025 | R$3 million, a projection, not a result | D | IT Forum, 29 July 2025: "projeta encerrar 2025 com um faturamento de R$ 3 milhões, um crescimento de 700% em relação ao ano anterior" (translation: projects closing 2025 with revenue of R$3 million, growth of 700% over the year before). Diário do Comércio, 28 October 2025, gives the same R$3 million and calls it tripling. About US$559,399 at the 28 October 2025 H.10 rate. | [2, 1, 8] |
| 2024 revenue implied by the IT Forum growth rate | about R$375,000 | calc | Calculated: R$3,000,000 divided by 8 (700% growth means eight times). This does not agree with the R$1 million for 2024 in Diário do Comércio, which implies 200% growth. See the watch list. | [1, 2] |
| Founded | May 2024, after an idea from 2023 | C | Diário do Comércio: "A empresa, que nasceu em maio de 2024" (translation: the company, born in May 2024). San Pedro Valley's company note of September 2024: "Fundada em 2024". The company site dates the origin to the founder's UK visit in 2023 and says it was an AI company from the first day: "Nascemos como uma empresa de Inteligência Artificial" (translation: we were born as an artificial intelligence company). | [1, 3, 4, 5] |
| Clients, October 2025 | 20, with a target of 30 more by the end of 2025 | C | Diário do Comércio, from the founder. Medium and large companies, mainly in pulp, construction and education. | [1] |
| Named customers on the company site | Afya, Cogna, Unicesumar, Vitru Educação, Cenibra, Novelis, Emccamp and others | C | Company FAQ page, read 5 November 2026. Emccamp and Gel Engenharia were already named as customers at the launch in September 2024. | [7, 3] |
| Outside funding | none as of October 2025 | C | Diário do Comércio: "não houve captação de recursos no mercado para investimentos, devido à rápida geração de caixa" (translation: no funds were raised from the market because of fast cash generation). IT Forum, July 2025, also says no outside investment. | [1, 2] |
| Team | headcount not published; six partners listed in November 2026 | C | At the launch in September 2024 two people were named: Valadão as CEO and Grossi as partner and COO. The company's About page lists a founder and CEO and five partners (CTO, COO, commercial and marketing, projects, services). No source gives the number of employees. | [3, 4] |
| Customer results claimed | one university cut its grant-document team from 25 to 10 people | C | Customer quotation on the company site (Unicesumar, ProUni documents). The founder also told Diário do Comércio of a client that went from about 70 people to eight for the same checks. | [5, 7, 1] |
| Accuracy claimed | above 95% on average | C | Founder to Diário do Comércio, October 2025; Vitru Educação is quoted on the site with 90% of enrolments validated by the AI alone. | [1, 5] |
| Ranking | 4th AI company in Brazil in the 100 Open Startups ranking 2025 | C | Company site. Not checked against the ranking itself. | [4] |
| Time from founding to the October 2025 statement | 17 months | calc | Calculated: May 2024 to October 2025. | [1] |
How to read this table: grade D is a figure that only the company stated, without an outside check. Grade C is a statement by the company or its customers about facts other than money. Grade X was calculated on this page. Quotations in Portuguese are followed by an English translation marked (translation). No value is grade A or B.
| Plan | Per month, billed yearly (US$) | Per month, billed monthly (US$) | Limit |
|---|---|---|---|
| Setup and implementation | quote only | one-time; 1 to 4 months of project work | |
| Page allowance subscription | quote only | quote only | price per page falls with volume; no per-user fee |
| Calculator example, 20,000 pages a month | about 2,359 (R$12,800) | company's example scenario, not a list price | |
| Add-on modules (anti-fraud, certificate issuing and others) | quote only | quote only | may be charged separately |
Dynadok sells by quote. A one-time setup fee pays for its project team to design the document flows and rules and to go live, which takes one to four months. After that, a monthly subscription buys an allowance of pages processed by the AI, which expires monthly or yearly; the price per page falls as the volume grows. There is no charge per user or licence. Modules such as anti-fraud and automatic issuing of certificates can cost extra. The one figure on the site is an example from the company's savings calculator: R$12,800 a month for 20,000 pages a month, about US$2,359 at the Federal Reserve H.10 rate of 30 October 2026 (5.4252 reais per US dollar). It is a scenario, not a list price.
When did Dynadok launch, and how did it grow?
From a visit to the UK technology scene in 2023 to 20 corporate clients in October 2025. Each entry carries its grade.
- 2023CWillian Valadão, a software entrepreneur with more than 20 years of experience, visits the UK technology scene, including TechX London, and decides to build AI document validation in Brazil. [4, 3]
- May 2024CDynadok is founded in Belo Horizonte. [1]
- 23 September 2024CSan Pedro Valley announces the launch, with Rodrigo Grossi as partner and COO and the construction companies Emccamp and Gel Engenharia as customers. [3]
- 2024DThe company later states revenue of R$1 million for this year. [1]
- 29 July 2025DIT Forum reports a projection of R$3 million of revenue for 2025, with no outside investment. [2]
- 2025CNamed the 4th AI company in Brazil in the 100 Open Startups ranking, according to the company. [4]
- 28 October 2025DValadão tells Diário do Comércio of 20 clients, R$1 million of revenue in 2024, and plans to seek investment for the European Union in 2026. [1]
- August 2026CThe company publishes its charging model: setup fee plus a monthly page allowance, no per-user fee. [6]
Why did Dynadok work?
Four causes, in the order they mattered. Each one is a reusable lesson.
A narrow, painful check
Brazilian law makes the company that hires a subcontractor share the risk of unpaid labour and tax obligations, so construction, mining and pulp companies collect monthly certificates from every supplier. Universities check papers for the ProUni and FIES student grants. Both are high-volume, rule-based checks done by large manual teams, and a wrong or expired document has a cost.
A few large customers
The founder said the company served 20 medium and large clients in October 2025. Large customers pay a setup fee and a monthly subscription by volume, so a small number of contracts can carry the company.
Price by page, not by seat
The customer pays for pages processed, with no charge per user. The price follows the work the AI does and grows with the customer's volume, and every department of the customer can use the platform without a new licence.
Suppliers do the uploading
Third parties upload their own documents and get told in seconds what is wrong, and the platform asks them to renew before a document expires. The customer's staff only handle the exceptions, a model the company calls human in the loop.
How it was built
These details come from the company's site and the launch article. They are grade C. No source names the AI model providers. [7, 3, 5, 6]
- A combination of large language models, advanced OCR, computer vision and retrieval-augmented generation (RAG, where the model looks up stored reference text before it answers), according to the company's FAQ.
- At launch the company described IDP with machine learning, generative AI and natural language processing, and contrasted it with robotic process automation (RPA), scripts that copy what a person clicks.
- Configurable checklists and rules per document type, and cross-checks between documents and with the customer's systems.
- A portal where suppliers, students and candidates upload documents, with expiry tracking and renewal reminders.
- Automatic issuing and lookup of 27 official certificates and registers, such as federal tax, labour debt and FGTS certificates, directly at the government source.
- A REST API for integration with the customer's ERP, and a project team that sets up each customer over one to four months.
Tools named in the sources
Each one opens its page in the Tools tab, with every story that uses it.
Who tried this before Dynadok, and why did they fail?
Three companies in our database of failed startups tried to turn paperwork or contract checks into software and failed. The values below are from that database and are not independently verified.
Canvas
The promiseTurn paper forms into digital workflows on mobile devices.
Why it diedMicrosoft and Adobe offered the same, and low-code platforms took the market while it struggled with the move to mobile.
Glider
The promiseOne place to manage and negotiate contracts.
Why it diedLarge players such as DocuSign covered the need, and its price was too high for small businesses.
jEugene
The promiseAutomatic error checking for legal contracts, like a spellchecker for lawyers.
Why it diedBetter-funded AI rivals moved faster, law firms bought slowly, and it ran out of money.
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
The three predecessors sold document software before language models existed. Their tools could fill in forms or flag known patterns, but a person still had to read each document and decide. They also sold to many small buyers, and large platforms with the same features took those buyers away.
Dynadok differs in two ways. Language models and modern OCR can now read a scanned certificate or a 100-page safety programme and apply a written rule to it, so the software does the reading itself and people review only the exceptions. And it sells a narrow, regulated check, Brazilian labour, tax and grant documents, to large companies that pay by volume, a niche that the general platforms do not cover in detail.
Can one person copy Dynadok now?
Partly. The niche needs local rules and enterprise sales.
A solo founder can build an AI checker for one document type with current models. What made Dynadok sell is knowledge of local regulations, a list of documents that every large company in a sector must collect, and a team that implements each customer. That is a small-team business with long sales cycles, not a self-serve product.
Where it is still open: Supplier and subcontractor compliance in other countries with their own certificates, grant and benefit applications, and customs documents for importers and exporters.
- 1Find a document check that a law or regulator makes every large company in one sector repeat each month.
- 2Build the reading and the rules for those few document types first, with a person reviewing the exceptions.
- 3Let the third parties upload their own documents and get the errors back at once.
- 4Charge a setup fee and a price per page, so revenue grows with the customer's volume.
What could go wrong for Dynadok?
The risks are the mirror image of the causes above.
What is not verified
The headline, R$1 million of revenue in 2024, rests on statements by the company. Willian Valadão gave it to Diário do Comércio on 28 October 2025, and IT Forum's article of 29 July 2025, which uses a company photo and quotes, said the company had passed R$1 million of annual revenue. The two do not agree in detail: IT Forum's "700%" growth to R$3 million implies about R$375,000 in 2024, while Diário do Comércio's R$1 million implies tripling. A company that the founder said was born in May 2024 would have reached R$1 million in about eight months. No payment processor, filing or auditor confirms either figure. A reader could look for the company's actual 2025 result in later press or a funding announcement.
A projection, not a result
The R$3 million for 2025 was a projection in July and October 2025. No source found in November 2026 states what the company actually billed in 2025.
Team size not published
Two people were named at the launch in September 2024. The company site lists a founder and five partners in November 2026, and each customer gets a project team for setup, so the company may be larger than six people. No source gives the headcount.
Part software, part implementation
The product is a software platform with an API, but each contract starts with a paid setup of one to four months by Dynadok's project team, and new rules or documents can cost extra. Revenue figures may include this implementation work.
Founding date differs between sources
The company site dates its origin to 2023, the founder's UK visit. Diário do Comércio says the company was born in May 2024, and the launch note says it was founded in 2024. This page uses May 2024.
Personal data
The platform reads identity documents, health and safety records of workers and student papers, and checks politically exposed person and sanctions lists. The company says its COO is responsible for compliance with Brazil's data protection law (LGPD).
Sources
The grade next to each source is the best grade that the source can support. "Read in full" means that the text of the page was downloaded and the numbers were checked in it.
- D Diário do Comércio, 28 October 2025, Michelle Valverde: Dynadok lança IA para validação automatizada de documentosRead in full, in Portuguese. Interview with the founder. Source for the 2024 revenue, the founding month, the client count, the no-funding statement and the plans for 2026.
- D IT Forum, 29 July 2025, Bella Winckler Matrone: Dynadok projeta faturamento anual superior a R$ 3 milhõesRead in full, in Portuguese. Based on company statements and quotes from both partners. Source for the 2025 projection, the 700% growth figure and the claim of more than R$1 million of annual revenue.
- C San Pedro Valley, 23 September 2024: Dynadok: nova startup lança IA de validação documentalRead in full, in Portuguese. Launch announcement with the company note. Source for the launch date, the founding year, the COO's role, the early customers and the technology description.
- C Dynadok: Quem Somos: Conheça nosso Propósito, Equipe e ValoresRead in full, 5 November 2026. Source for the 2023 origin, the leadership list and the ranking claim.
- C Dynadok home pageRead in full, 5 November 2026. Source for the features, the document types, the customer quotations, the AI-from-day-one statement and the savings calculator example.
- C Dynadok: Preços e Modelo de Cobrança (updated August 2026)Read in full, 5 November 2026. Source for the charging model; it publishes no list prices.
- C Dynadok: Perguntas e respostas sobre a Dynadok, and the AI Info pageRead in full, 5 November 2026, with https://dynadok.com/ai-info/. Source for the technology stack, the named customers and the 27 official certificates.
- A Federal Reserve, H.10 release: historical rates for the Brazilian real, reais per US dollarRead the daily table (last update 2 November 2026). Used 5.3629 for 28 October 2025 for the revenue figures and 5.4252 for 30 October 2026 for the calculator example.
- 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