Granter: Revenue, team and how it was built
Granter (AI agent for public grant applications, Portugal): About US$466K in sales in nine months, November 2025. Evidence grade D: reported, not verified. Team: 2 co-founders at the start in 2023 (about page); 13 people on the about page in November 2026. Launched September 2023; still running.Added 5 November 2026 · updated 6 November 2026
- Outcome
- About US$466K in sales in nine months, November 2025
- Evidence
- D reported, not verified
- Launch
- September 2023
- Status
- Still running
- Team
- 2 co-founders at the start in 2023 (about page); 13 people on the about page in November 2026
- Outside funding
- Techstars accelerator, March 2025 (amount not published); no priced round found
- Main channel
- Not stated
What is not verified
The headline €400K in sales in nine months rests on one source: the Upnext blog's account of the Web Summit PITCH final on 13 November 2025, repeating what the company said on stage. The nine months are not dated, and "sales" may mean contracts signed, not cash received. No payment processor, company filing or second interview gives a revenue figure. A reader could look for the company's annual accounts in the Portuguese business register or a later interview with a dated revenue number. 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.
Granter is a vertical AI agent, an AI program built for one narrow kind of work rather than for general chat, in this case public grants: money that governments and the European Union give to companies for research, hiring or investment projects, against a written application. The agent searches funding databases, checks eligibility (whether the company meets the official conditions of a call), writes the application and scores it against the evaluation criteria before submission.
Bernardo Seixas (CEO) and Bernardo Tavares (CTO), the two co-founders named on the company's about page, started it in September 2023 under the name Lever Tax and renamed it Granter.ai in January 2024. A web archive copy from March 2024 already sold "AI-generated draft applications". The company joined Techstars in March 2025, launched a fully automated version in June 2025 and won the PITCH competition at Web Summit in November 2025. The Upnext blog reported from that pitch that Granter had made €400K in sales in nine months, about US$466,000 at the exchange rate of that day.
That sales figure is a single company claim made on stage, reported by one blog, so it is grade D. No payment processor, filing or second interview confirms it. The about page now lists 13 people, so the small founding team has grown. The page also shows earlier companies that mixed software with human professional work and failed.

How much revenue does Granter 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 |
|---|---|---|---|---|
| Sales in nine months, reported November 2025 | €400,000 | D | Upnext blog, 13 November 2025, in its account of the Web Summit PITCH final: "Traction: €400K in sales in just nine months". A company claim from the pitch; the nine months are not dated, and "sales" may mean contracts signed rather than cash received. | [1] |
| Sales in US dollars | about US$465,960 | calc | Calculated: €400,000 times 1.1649 US dollars per euro, the US Federal Reserve H.10 noon rate for 13 November 2025. | [1, 12] |
| Customer countries, November 2025 | Portugal, Spain, Germany, Belgium and Austria; 20% of inbound demand from the US | D | Same Upnext report of the pitch. | [1] |
| Start | September 2023, as Lever Tax | C | About page timeline: "Sep 2023 Granter is born (originally as Lever Tax)". Renamed Granter.ai in January 2024. | [2] |
| AI drafting in the product, March 2024 | "AI-generated draft applications" and "Join 300+ companies" | C | Internet Archive copy of the home page, 26 March 2024. The same page sold a premium human expert review. | [9] |
| Founding team | 2 co-founders: Bernardo Seixas (CEO) and Bernardo Tavares (CTO) | C | About page; both appear by name in an ECO photo caption of April 2025. | [2, 7] |
| Team, November 2026 | 13 people on the about page | C | Counted on the about page on 5 November 2026. The careers page lists 7 open roles, including freelance funding consultants. | [2, 11] |
| Outside programmes | Techstars (March 2025); Fidelidade Protechting roadshow to China (April 2025) | C | About page; ECO, 13 April 2025. No source gives an investment amount. | [2, 7] |
| Competition wins | Web Summit PITCH 2025 winner out of 2,725 startups; South Summit Madrid 2026, Fintech and Insurtech vertical | C | Upnext, 13 November 2025 and 6 June 2026; Executive Digest, 9 June 2026. | [1, 6, 5] |
| Applications supported, June 2026 | more than 200, worth more than €20 million | C | Executive Digest: "já apoiou a submissão de mais de 200 candidaturas a fundos europeus, num valor superior a 20 milhões de euros" (translation: has already supported the submission of more than 200 applications to European funds, worth more than 20 million euros). This is grant value applied for, not Granter revenue. | [5] |
| Home page counters, November 2026 | "Trusted by 3000+ companies"; "€30M in grants submitted" | C | Company's own counters, not checked. Executive Digest in June 2026 spoke of more than 5,000 organisations on a waiting list. | [3, 5] |
| Prices | free one-week trial; Professional €1,500 plus 5% success fee | C | Pricing page, read 5 November 2026. | [4] |
| Time from start to the November 2025 figure | 26 months | calc | Calculated: September 2023 to November 2025. | [1, 2] |
How to read this table: grade D is a single claim with one date and no second source. Grade C is the company's own page or a press report of the company's words. A "calc" value was computed on this page. No value is grade A or B except the exchange rate.
| Plan | Per month, billed yearly (US$) | Per month, billed monthly (US$) | Limit |
|---|---|---|---|
| Free Trial | 0 | 0 | full platform access for one week |
| Professional | €1,500 plus 5% success fee; 5 projects and 3 applications; human support; period not stated | ||
| Enterprise | custom pricing plus 5% success fee; no project or application limits |
The free trial gives full access to the software for one week. Professional costs €1,500 plus a 5% success fee, meaning 5% of the grant money the customer wins, and adds human experts and post-approval management. Enterprise has custom pricing plus the same 5% success fee. The page does not say whether €1,500 is per month, per year or per application, so the USD columns are empty.
When did Granter launch, and how did it grow?
From a tax-incentive idea in Lisbon to a Web Summit win in 26 months. Each entry carries its grade.
- September 2023CFounded as Lever Tax, with the goal of making public funding easier to reach. [2]
- November 2023CFirst 100 users on the platform, by the company's account. [2]
- January 2024CRenamed Granter.ai. [2]
- March 2024CThe home page sells AI grant discovery, AI-generated drafts and a premium expert review, and claims 300+ companies. The company says €1M+ in funding has been submitted through it. [9, 2]
- March to April 2025CJoins Techstars in Turin and is one of four startups chosen for Fidelidade's Protechting roadshow to China. [2, 7]
- June 2025CGranter 2.0, described as a fully automated grant-writing experience. [2]
- 13 November 2025DWins the Web Summit PITCH final. The Upnext report gives €400K in sales in nine months. [1]
- June 2026CWins the Fintech and Insurtech vertical at South Summit Madrid, the only Portuguese winner that year. [6, 5]
- July 2026CThe CEO advises on common application mistakes in ECO's podcast on European funds. [8]
Why did Granter work?
Four causes, in the order they mattered. Each one is a reusable lesson.
One narrow, paperwork-heavy job
Grant applications follow published rules and evaluation criteria, so an AI agent can be checked against them: the product scores each draft the way an evaluator would, before submission.
Replace the consultant, not the company's team
Executive Digest wrote that the technology reduces to a few hours work that usually takes weeks and needs specialist consultants. The customer keeps its own project and narrative; the agent does the search, the forms and the reporting.
Price tied to the outcome
The paid plans add a 5% success fee on grants won, the usual pricing of human grant consultants. It makes the first purchase easier for a company, and it ties revenue to approval decisions that Granter does not control.
Competitions as marketing
Two large competition wins, Web Summit and South Summit, brought national press in Portugal and a reported share of inbound demand from the US.
How it was built
These details come from the company's own pages, mainly a job advertisement. They are grade C. [10, 11, 3]
- An agent system written in Python with LangChain and LangGraph, according to the AI Agents Engineer job page, with RAG (retrieval-augmented generation: the model reads documents fetched from a database before it writes).
- The job page asks for experience with several LLM providers and model trade-offs; it does not name the model the product uses.
- Django, Celery, Docker and Azure are listed as nice-to-have skills, which suggests parts of the stack.
- The marketing site is built with Framer, visible in the page source.
- Human funding experts, including freelance consultants, support the paid plans.
Tools named in the sources
Each one opens its page in the Tools tab, with every story that uses it.
Who tried this before Granter, and why did they fail?
Three companies in our database of failed startups promised to automate professional paperwork with software and found that the hard cases still needed expensive humans. The values below are from that database and are not independently verified.
Atrium
The promiseLegal work for startups, faster and cheaper than large law firms, through software plus lawyers.
Why it diedThe automation could not handle real legal complexity, so the company carried law-firm costs without law-firm margins.
ScaleFactor
The promiseAI bookkeeping for small businesses, without accountant fees.
Why it diedThe AI did much less than promised, and QuickBooks offered more for less.
jEugene
The promiseA spell-checker for contracts that finds errors automatically.
Why it diedBetter-funded AI rivals moved faster, and law firms adopted new tools slowly.
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 three sold automation of professional documents before large language models existed. Their software handled the easy cases, and people had to do the rest, so costs grew with every customer.
Grant applications are long texts written against public criteria, which is the kind of work current language models do well. Granter still keeps human experts in its paid plans and charges a success fee like a consultant, so the open question is the same as for Atrium and ScaleFactor: how much of each application the software really does as the customer count grows.
Can one person copy Granter now?
Yes, in another country or funding system.
Public funding rules are national and change every programme period, so a local agent with local knowledge has an advantage. A solo founder could build the software side for one country's grant system; the human review and the success-fee model need a funding expert.
Where it is still open: Grant systems outside Portugal and Spain, research funding for universities, and the reporting work after approval, which many companies find harder than the application.
- 1Pick one country's grant programmes and collect the official calls, criteria and evaluator guides.
- 2Build eligibility checks and draft writing against those criteria, and score each draft before submission.
- 3Offer a free trial of the software, and price paid plans with a success fee that buyers already know from consultants.
- 4Enter startup competitions and local press to reach companies that apply for funds.
What could go wrong for Granter?
The risks are the mirror image of the causes above.
What is not verified
The headline €400K in sales in nine months rests on one source: the Upnext blog's account of the Web Summit PITCH final on 13 November 2025, repeating what the company said on stage. The nine months are not dated, and "sales" may mean contracts signed, not cash received. No payment processor, company filing or second interview gives a revenue figure. A reader could look for the company's annual accounts in the Portuguese business register or a later interview with a dated revenue number.
Software with human consultants and a success fee
The product is an AI platform with a self-serve free trial, but both paid plans add human experts and a 5% success fee on grants won, and the careers page recruits freelance funding consultants. Part of the revenue may therefore come from consulting work done by people, not from the software alone. The pricing page does not say how the €1,500 is billed.
The team grew
Two co-founders started the company in 2023. The about page lists 13 people on 5 November 2026, and the careers page has 7 open roles. The company went through Techstars in 2025; no source gives the amount of any investment.
Company counters
The home page claims "Trusted by 3000+ companies" and "€30M in grants submitted"; Executive Digest reported more than 200 applications worth more than €20 million in June 2026. These counters are the company's own and count grant money applied for, not Granter revenue.
Competition and model providers
The main competitors named in the sources are human grant consultants; this is the only grant-application story on the site. The job page says the agent is built with LangChain and LangGraph over outside LLM providers and does not name one, so prices and terms of those providers affect the product's costs.
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 Upnext, 13 November 2025: Web Summit PITCH 2025, the finals that point to what's upnextRead in full. Source for the €400K sales claim, the customer countries and the win out of 2,725 startups.
- C Granter, About Us (timeline and team)Read in full on 5 November 2026. Source for the founding date, the rename, the co-founders and the team count.
- C Granter home pageRead in full on 5 November 2026. Source for the product description and the company's counters.
- C Granter, Plans & PricingRead in full on 5 November 2026. Source for the pricing table.
- C Executive Digest, 9 June 2026: Portuguesa Granter conquista prémio no South Summit com IA que acelera acesso a fundos europeusRead in full (Portuguese). Source for the South Summit win, the application count and the waiting list.
- C Upnext, 6 June 2026: South Summit Madrid 2026, Mitiga Solutions wins, plus the standoutsRead the list of vertical winners.
- C ECO, 13 April 2025: Conheça as quatro startups que vão no roadshow da Fidelidade à ChinaRead in full (Portuguese). Names both co-founders in a photo caption.
- C ECO, 19 July 2026: Dica ECO dos Fundos, erros a não cometer numa candidaturaRead in full (Portuguese). Confirms Bernardo Seixas as CEO in July 2026.
- C Internet Archive copy of granter.ai, 26 March 2024Read in full. Shows the AI drafting product and the human review in early 2024.
- C Granter careers: AI Agents EngineerRead in full. Source for the technical stack.
- C Granter careers pageRead in full. Source for the open roles and the freelance funding consultants.
- A US Federal Reserve, H.10 historical rates: euroRead the row for 13 November 2025: 1.1649 US dollars per euro.
- 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