Henry AI: Revenue, team and how it was built
Henry AI (AI that builds pitch decks, offering memoranda and underwriting models for commercial real estate brokers, United States): Just under US$1M ARR, July 2025. Evidence grade D: reported, not verified. Team: 3 people in the Y Combinator launch post of August 2024 (the 2 co-founders and a remote team member); 8 full-time employees in February 2025 (Commercial Observer); 20 people on the Y Combinator company page read on 5 November 2026; headcount at the July 2025 figure not published. Launched 16 August 2024; still running, revenue growing.Added 5 November 2026
- Outcome
- Just under US$1M ARR, July 2025Revenue growing
- Evidence
- D reported, not verified
- Launch
- 16 August 2024
- Status
- Still running
- Team
- 3 people in the Y Combinator launch post of August 2024 (the 2 co-founders and a remote team member); 8 full-time employees in February 2025 (Commercial Observer); 20 people on the Y Combinator company page read on 5 November 2026; headcount at the July 2025 figure not published
- Outside funding
- US$4.3M seed (announced 20 February 2025) and US$16.5M Series A led by FirstMark Capital (raised in autumn 2025, announced 29 July 2026), plus Y Combinator
- Main channel
- Founder's posts and audience
What is not verified
The headline, just under US$1 million in annual recurring revenue at the end of July 2025, rests on one source: Sammy Greenwall's own words in a blog post that Rippling, a vendor to Henry, published on 7 August 2025. The post's text says "surpassed $1 million", while the founder's quoted words say "just under"; this page uses the lower wording. Upstarts wrote in July 2026 that Henry does not disclose its revenue and estimated that it "runs in the millions" from its prices, which is an inference, not a figure. No payment-processor profile, investor statement or filing confirms any revenue number. A reader could look for a revenue figure in a later funding announcement or an interview in which an investor states it. 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.
Henry AI, a New York company, sells AI software to commercial real estate brokers. Commercial real estate means buildings that earn rent or are used by businesses: offices, shops, warehouses, apartment blocks and hotels. When a broker wants to sell such a building, or wants to win the right to sell it, the firm prepares an offering memorandum (OM), a long marketing document with the building's rents, costs, market data and photos, or a shorter pitch deck and a broker opinion of value (BOV), an estimate of the sale price. Henry uses large language models (LLMs), AI programs trained on very large amounts of text that can read documents and write new text, to turn the deal data into these documents in minutes instead of days.
Sammy Greenwall, a former real estate finance analyst and co-founder of the financing marketplace Lev, and Adam Pratt, an engineer who had sold a software company for fire departments, started Henry in 2024 and went through Y Combinator's summer 2024 batch. Their launch post of 16 August 2024 names three people. Greenwall said in a Rippling blog post published 7 August 2025 that Henry had 62 brokerages as customers and "just under $1 million in ARR at the end of July", with no money spent on marketing and no sales team. The same post's own text says Henry passed US$1 million in annual recurring revenue (ARR), the yearly value of the subscriptions running at one moment, eight months after launch.
The company did not stay small. It had eight full-time employees when it announced a US$4.3 million seed round in February 2025, and the Y Combinator page lists 20 people in November 2026, after a US$16.5 million Series A led by FirstMark Capital. In July 2026 Upstarts reported more than 150 customer firms and prices from US$2,000 a month; it wrote that Henry does not disclose its revenue and estimated that it "runs in the millions". The headline figure rests on one founder statement on one date, so this page grades it D: reported, not verified.
How much revenue does Henry 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 |
|---|---|---|---|---|
| Annual recurring revenue, end of July 2025 | just under US$1 million | D | Greenwall quoted in the Rippling blog, published 7 August 2025: "We have over 62 brokerages across the country, both small, medium, and enterprise size. We just under $1 million in ARR at the end of July." The blog's own text says Henry "has surpassed $1 million in ARR" eight months after launch; this page uses the founder's own, lower words. Rippling sells HR software to Henry, and the post is part of its marketing. | [1] |
| The ARR figure per month | about US$83,000 a month | calc | Calculated: US$1,000,000 divided by 12. Because the founder said "just under", the real value is a little lower. | [1] |
| Revenue in July 2026 | not disclosed; "runs in the millions" (estimate) | D | Upstarts, Alex Konrad, 29 July 2026: "Henry doesn't disclose exact revenue, but with pricing starting at $2,000 per month and scaling to seven figures based usage, it runs in the millions." A journalist's inference from prices, not a stated figure. | [5] |
| Customers | 62 brokerages (July 2025); more than 150 firms (July 2026) | C | Founder in the Rippling blog for 2025. Upstarts for 2026: "used by more than 150 firms including Newmark and Berkadia, and spanning four of the five largest brokerages in the U.S." The home page says "150+ firms" and "Over $150B in underlying deal value". | [1, 5, 6] |
| Customer growth | about 2.4 times in one year | calc | Calculated: 150 firms divided by 62 brokerages, July 2025 to July 2026. The 2026 count is "more than 150", so the ratio is a lower bound. | [1, 5] |
| Marketing and sales at the headline date | no marketing spend, no sales team | D | Founder in the Rippling blog: "We've never spent a dollar on marketing, and we don't have a sales team. It's been all product-led growth and me posting on LinkedIn." Upstarts adds that Greenwall renamed himself "The deck guy" on LinkedIn and reached prospects through his content and personal outreach. | [1, 5] |
| Team at the start | 3 people named in the launch post; 2 listed founders | C | Y Combinator launch post, 16 August 2024: "We are Sourav, Sammy and Adam", with the note that the third member works remotely. The Y Combinator company page lists two founders: Greenwall (Founder/CEO) and Pratt (Founder/CTO). | [3, 2] |
| Team, February 2025 | 8 full-time employees | C | Commercial Observer, 20 February 2025: "Henry AI has eight full-time employees". The same article calls Henry "The 8-month-old Manhattan-based company". | [4] |
| Team, November 2026 | 20 people | C | Y Combinator company page, read 5 November 2026: team size 20, location New York City. The careers page lists eight open roles, most of them in sales (Head of Sales, Head of Growth, Account Executive, three Director roles). | [2, 7] |
| Team growth | 2.5 times from February 2025 to November 2026 | calc | Calculated: 20 people divided by 8 full-time employees. No source gives the headcount in July 2025, when the headline figure was stated. | [2, 4] |
| Seed round | US$4.3 million, announced 20 February 2025 | C | Commercial Observer: co-led by Susa Ventures and 1Sharpe Ventures, with other funds and angel investors; Y Combinator also backs the company. | [4] |
| Series A | US$16.5 million, led by FirstMark Capital | C | Upstarts, 29 July 2026: raised "quietly" in autumn 2025 and announced with the launch of Henry Deal; Thomson Reuters Ventures also invested. The home page shows the same round. | [5, 6] |
| Total outside funding | US$20.8 million | calc | Calculated: US$4.3 million plus US$16.5 million. Y Combinator's own investment amount is not stated in these sources. | [4, 5] |
| Y Combinator batch | Summer 2024 | C | The Y Combinator company page says S24 (Summer 2024), and Upstarts says the founders were "midway through Y Combinator's summer batch" in 2024. The Rippling blog says W24; this page follows Y Combinator's own page. | [2, 5, 1] |
| Founded | 2024 | C | Y Combinator page: year founded 2024. Commercial Observer's "8-month-old" in February 2025 points to about June 2024. The launch date on this page is the Y Combinator launch post, 16 August 2024. | [2, 4, 3] |
| Time from launch to the headline figure | 11 months | calc | Calculated: August 2024 to July 2025. The Rippling blog says "eight months after launch", which would put the launch near the end of 2024; no source gives that later date. | [1, 3] |
| AI model named | a Google Gemini research model | C | Commercial Observer, February 2025: Henry "connects with a Google-made Gemini research model to search everything publicly available on the internet", said Greenwall. Upstarts, July 2026: the new product is designed to swap in whichever state-of-the-art model a task needs. | [4, 5] |
| Human work in the decks product | yes, for now | C | Upstarts: the decks product "still involves some human intervention on the back-end for now (to be phased out over the next year or so)". Home page FAQ: "Henry's team curates every deck before it reaches your client." | [5, 6] |
| Security | SOC 2 Type II certified | C | Home page: "SOC 2 Type II certified. Encrypted at rest and in transit." The audit report itself was not seen. | [6] |
How to read this table: a grade D value is one founder statement on one date, or a journalist's estimate, with no outside check. A grade C value is what the company, its founder or its Y Combinator page says about the product, the team or the funding, or what a journalist reports from them. A "calc" value (grade X) was computed on this page. No revenue value is grade A, B or C.
| Plan | Per month, billed yearly (US$) | Per month, billed monthly (US$) | Limit |
|---|---|---|---|
| Henry (deal platform) | from 2,000 | Upstarts, July 2026; billing period not stated; scales with usage | |
| Henry Decks, standalone | price not published | ||
| Large firm contracts | Upstarts: up to seven figures a year, depending on usage |
The company's site has no pricing page; every button leads to "Book a Call". Upstarts wrote on 29 July 2026 that pricing starts "at $2,000 per month" and scales "to seven figures based usage". The source does not say whether the price is billed monthly or yearly. Henry Decks is also sold alone, without the underwriting and database parts; its price is not published.
When did Henry AI launch, and how did it grow?
From a launch post by three people in August 2024 to a stated US$1 million in annual recurring revenue within a year, and then to a Series A and a team of 20. Each entry carries its grade.
- 2019 to 2024CGreenwall co-founds Lev, a commercial real estate financing marketplace, and helps take it to a Series B. Pratt had co-founded Halligan, software for fire departments, acquired by Vector Solutions in 2019. [2, 5]
- 2024CGreenwall leaves Lev and asks Pratt, a friend from high school, to build Henry with him. They pitch brokers with mockups before writing code; most say no, a few early adopters say yes. [5, 1]
- 16 August 2024CLaunch post during Y Combinator's summer 2024 batch: an AI copilot that builds deal decks and financial analysis for brokers, by a team of three. [3, 2]
- 20 February 2025CUS$4.3 million seed round announced. Eight full-time employees; Greenwall says brokerages from five of the top 10 in the U.S. use the product, without naming them. [4]
- End of July 2025DGreenwall says Henry has more than 62 brokerages and just under US$1 million in annual recurring revenue, with no marketing spend and no sales team. [1]
- 7 August 2025DRippling publishes its founder spotlight with the figure; the post's own text says Henry passed US$1 million in ARR eight months after launch. [1]
- Autumn 2025CHenry raises a US$16.5 million Series A led by FirstMark Capital, without announcing it. [5]
- Early 2026CAfter new Anthropic Claude models, the founders see weaker demand from prospects and fewer upsells, Upstarts reports; the company is still growing. [5]
- May to 15 June 2026CA 35-day product sprint builds Henry Deal, which runs underwriting, comparable sales and the deck from one prompt, designed to work without human help. [5]
- 29 July 2026CUpstarts announces the Series A and Henry Deal: more than 150 customer firms, prices from US$2,000 a month, revenue not disclosed. [5]
- 5 November 2026CThe Y Combinator page lists 20 people; the careers page lists eight open roles, most in sales. [2, 7]
Why did Henry AI work?
Four causes, in the order they mattered. Each one is a reusable lesson.
A founder who did the work himself
Greenwall built dozens of these decks as a young analyst and grew up around his family's commercial real estate business, Upstarts reports. He knew which document hurt most, the offering memorandum and the broker opinion of value, and he knew the brokers' language when he sold to them.
Sell before building
The founders showed mockups to brokers first. Most rejected the idea, the founder told Rippling, but a few early adopters agreed to try it, and that was the signal to build the first version. One early user later told Greenwall that a Henry deck had won him a listing against 11 other brokers.
Founder posts instead of a sales team
The founder said in July 2025 that growth had come from the product and from his own LinkedIn posts, with no marketing spend. Upstarts reports that both founders flew to pitches, and that Pratt changed the code live during meetings.
AI output checked by people
The deck product is not fully automatic: Henry's team curates every deck before it reaches the client, according to the home page, and Upstarts says this human step is to be phased out. That made the output good enough for brokers whose reputation depends on the document, while the software did most of the work.
Moving from decks to the whole deal
When stronger general models arrived in early 2026, the founders feared that a decks tool could be copied with a general chatbot. Their answer was Henry Deal: underwriting in Excel, comparable sales and a buyer list from the firm's own data, and the deck, all from one prompt, plus a property database that grows with each deal.
How it was built
These details come from the company's own pages, its Y Combinator pages and press interviews with the founders. They are grade C. [4, 5, 6, 1]
- The first product read a broker's underwriting or capital model and deal files, and used a Google Gemini research model to collect public market data, according to Commercial Observer.
- The output is a deck or offering memorandum on the broker's own brand, with tables, maps, market overviews and a sponsor profile; people at Henry review each deck before delivery.
- Henry Deal, launched for test users on 15 June 2026, builds an underwriting model that lives in Excel through an Excel add-in and syncs changes back to the deck and data room.
- Brokers can start a deal by sending an email or a chat message with a rent roll; no login is needed for the email flow.
- The new product is designed to swap in whichever AI model is best for a task, Pratt told Upstarts, rather than depend on one model.
- The company says it is SOC 2 Type II certified and keeps each firm's data separate. It uses Rippling for onboarding, payroll, expenses and contractor compliance, the founder told Rippling.
Tools named in the sources
Each one opens its page in the Tools tab, with every story that uses it.
Who tried this before Henry AI, and why did they fail?
Four companies in our database of failed startups promised to modernize commercial real estate with software, or to make presentations automatically, and failed. The values below are from that database and are not independently verified.
Dadao
The promiseA digital marketplace that connected commercial landlords with tenants in China and removed the broker.
Why it diedOffline relationships between brokers and clients stayed stronger than the platform; it spent heavily on customer acquisition, and demand for flexible offices fell after Covid.
Enzo
The promiseAn operating system for property management: maintenance, vendors and operational workflows for real estate operators.
Why it diedProperty managers resisted changing their work, and each customer needed costly onboarding, so the unit economics never worked at scale.
LiveLoop
The promisePowerPoint files turned into live documents that a team could edit together.
Why it diedMicrosoft bought it in 2015 and then built the same features into Office.
Qwiki
The promiseNarrated multimedia presentations generated automatically for any search query.
Why it diedGeneration was expensive, users kept reading text, and it ended inside Yahoo, in the shadow of Google and mobile video.
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
Dadao and Enzo tried to change how real estate people work, either by removing the broker or by replacing the operator's tools, and both met an industry that would not change its habits. LiveLoop and Qwiki tried to automate or improve presentations, and lost to the large office suites or to high costs. Henry sells into the same conservative industry and makes the same kind of document.
Two things differ. First, Henry does not remove the broker or ask him to learn a new system: it writes the document that the broker already has to produce, on his brand, and it can be used from email and Excel. Second, large language models can now read a rent roll and write the market text and tables, work that earlier tools could not do and that firms paid analysts to do. The risk that LiveLoop met is still there: a general AI model from a large company may learn to make the same decks, which is why Henry moved to the whole deal and the firm's own data in 2026.
Can one person copy Henry AI now?
Not as a solo founder from zero. The pattern, yes.
Henry's start rests on things a new founder may not have: years as an analyst and as a co-founder in the same industry, a technical co-founder who had already sold a company, Y Combinator, and later US$20.8 million from investors, according to Commercial Observer and Upstarts. The pattern can be copied on a smaller scale: choose one document that a whole industry must produce by hand for every deal, and make an AI tool that writes it in that industry's format.
Where it is still open: The same approach for other fields with one heavy, repeated document per deal: residential appraisals, equipment leasing proposals, construction bids, grant applications and insurance submissions.
- 1Pick a document that every transaction in a field requires and that you have written yourself many times: an offering memorandum, an appraisal, a tender response, an insurance submission.
- 2Show mockups to the people who write it before you build, and start with the few who say yes.
- 3Keep a person checking the output at first, so customers can trust it with their name on it.
- 4Post about the work under your own name where the buyers already read, and let each won deal bring the next customer.
What could go wrong for Henry AI?
The risks are the mirror image of the causes above.
What is not verified
The headline, just under US$1 million in annual recurring revenue at the end of July 2025, rests on one source: Sammy Greenwall's own words in a blog post that Rippling, a vendor to Henry, published on 7 August 2025. The post's text says "surpassed $1 million", while the founder's quoted words say "just under"; this page uses the lower wording. Upstarts wrote in July 2026 that Henry does not disclose its revenue and estimated that it "runs in the millions" from its prices, which is an inference, not a figure. No payment-processor profile, investor statement or filing confirms any revenue number. A reader could look for a revenue figure in a later funding announcement or an interview in which an investor states it.
Team size and growth
The launch post of August 2024 names three people, and Commercial Observer counted eight full-time employees in February 2025. The Y Combinator page lists 20 people in November 2026, after the US$16.5 million Series A, and the careers page lists eight open roles, most in sales. No source gives the headcount in July 2025, so this page cannot say how many people produced the stated revenue.
From no sales team to a sales team
The founder said in July 2025 that Henry had no sales team and no marketing spend. In November 2026 the careers page seeks a Head of Sales, a Head of Growth, an Account Executive and several enterprise directors, so the low-cost way of finding customers described here belongs to the first year.
People in the loop
The home page says "Henry's team curates every deck before it reaches your client", and Upstarts says the decks product still involves some human work in the back end, to be phased out over the next year or so. Part of the product's quality in the period of the headline figure came from staff, not only from the software.
General AI models as competitors
Upstarts reports that new Anthropic Claude models in early 2026 brought weaker demand from prospects and fewer upsells, and a customer told Upstarts that he could probably do the work on Claude himself. Henry's answer is the firm's own data and a deeper product; whether that holds is not yet shown by any figure.
Claims on the home page
The home page says Henry has teams "from all of the largest CRE brokerages"; Upstarts wrote "four of the five largest brokerages in the U.S." in July 2026. The Rippling blog says Y Combinator W24, while Y Combinator's own page says Summer 2024. The customer quotes on the home page carry full names, titles and firms, and the company publishes named case studies.
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 Rippling blog, The Rippling Team, 7 August 2025: How Henry AI hit $1M ARR in 8 months, with zero marketing spendRead in full, downloaded 5 November 2026. A vendor's marketing post with quotes from Sammy Greenwall. Source for the ARR statement, the 62 brokerages, no marketing spend and no sales team, selling before building, the W24 claim and Henry's use of Rippling.
- C Y Combinator company page: Henry, Automating Deal Decks for Commercial Real Estate BrokersRead in full (the page data embedded in the HTML), downloaded 5 November 2026. Source for the batch (S24), year founded, team size 20, location, the two founders and their backgrounds.
- C Y Combinator launch post, 16 August 2024: Henry, AI copilot for commercial real estate brokersRead in full as embedded in the Y Combinator company page download of 5 November 2026 (created_at 2024-08-16). Source for the founding team of three, the first product and Greenwall's background at Lev.
- C Commercial Observer, Philip Russo, 20 February 2025: Henry AI Raises $4.3M in Seed RoundRead in full. Source for the seed round and investors, the company's age, eight full-time employees, the Gemini research model and the customer claim.
- C Upstarts, Alex Konrad, 29 July 2026: Exclusive: Henry AI Raises $16.5M, Launches Henry Deal For Real Estate BrokersRead in full (the free part of the article). Source for the Series A, the 150+ firms, the starting price, the revenue estimate, the founders' background, the LinkedIn outreach, the human step in the decks product, the Claude competition and the Henry Deal sprint.
- C Henry AI home pageRead in full, downloaded 5 November 2026. Source for the product description, the 150+ firms and deal value claims, the email and Excel workflow, the deck curation by staff, SOC 2 and Henry Decks. The site has no pricing page.
- C Henry AI careers page on Ashby, read 5 November 2026Read the list of open roles in the page data. Source for the eight open roles, most of them in sales.
- 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