Fenestra: Product, team and how it was built
Fenestra (AI rendering studio that turns architects' sketches, CAD drawings and 3D models into renders and animations, United Arab Emirates): 20000+ architects and designers stated by the company, November 2026. Evidence grade E: no revenue figure was checked. Team: 1 founder (Shaun McCallum); headcount not published. Launched May 2024; still running.Added 6 November 2026
- Outcome
- 20000+ architects and designers stated by the company, November 2026
- Evidence
- E no revenue figure was checked
- Launch
- May 2024
- Status
- Still running
- Team
- 1 founder (Shaun McCallum); headcount not published
- Outside funding
- No outside funding found in the sources; selected for the Web Summit Qatar PITCH 2026 startup competition
- Main channel
- Not stated
What is not verified
No revenue figure was found. We looked for a payment-processor profile, a founder interview with numbers, a company filing and a funding announcement, and found none. The 20,000+ users, 180+ countries and 200,000+ renders are the company's own statements; a sign-up count is not the same as paying customers. A reader could ask the founder for paying-customer numbers or look for a revenue profile on a payment-verified site. This story is grade E: the business is real and its product, price and customers are described from public pages, but no revenue figure was checked. Aiolum shows it as an idea worth studying, not as proof that it makes money.
Fenestra is an AI rendering tool, software that uses image-generation models to turn a drawing or a 3D model view into a finished-looking picture of a building, instead of computing light in a hand-built 3D scene. Its main inputs are sketch to render (a hand or CAD line drawing becomes a photoreal image) and inpainting (the user masks one region, for example a window, and only that region is regenerated). A live mode re-renders as the user draws.
The company's About page names one founder, Shaun McCallum, a Scottish architectural designer who worked at HASSELL and John Wardle Architects and on AR/VR at Meta Reality Labs, and says Fenestra is based in the UAE: "One founder, 20,000+ professionals rendering." An archived copy of the home page from May 2024 already offers browser renders from a prompt or a .GLB model; a redesigned studio opened to everyone on 6 April 2026. In February 2026 it was among the startups in the Web Summit Qatar PITCH competition.
No revenue figure was found. The headline is the company's own count of more than 20,000 architects and designers in 180+ countries and 200,000+ renders, which no outside source confirms, so the story is grade E. Paid plans start at US$10 a month, the company states.

How much revenue does Fenestra 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 |
|---|---|---|---|---|
| Users stated by the company, November 2026 | 20,000+ architects and designers in 180+ countries; 200,000+ renders | C | About page read on 6 November 2026: "More than 20,000 architects and designers in 180+ countries have created over 200,000 designs with it." The home page repeats "20,000+ architects & designers". No outside count was found. | [1, 2] |
| Founder | Shaun McCallum, Scottish architectural designer | C | About page: practised at HASSELL and John Wardle Architects, worked on AR/VR at Meta Reality Labs, co-founded the architecture collective iheartblob, taught at UCLA. UpNext's Web Summit report lists "Fenestra — Shaun McCallum". | [1, 5] |
| Team size | 1 founder; headcount not published | C | About page: "One founder, 20,000+ professionals rendering." Blog posts speak of "we", and one customer quote praises "the team"; no staff count is given. | [1, 2] |
| Base | United Arab Emirates | C | About page: "Fenestra is based in the UAE". | [1] |
| Live product, earliest copy | May 2024 | C | Internet Archive copy of 6 May 2024: "High fidelity renderings, in your browser with AI", with a Render Now button and .GLB model upload. An earlier capture of 4 January 2024 is only a redirect. | [4] |
| Why it started | after the release of Stable Diffusion | C | About page: "When Stable Diffusion was released, AI generation stopped being a curiosity" and the founder "started building the tool the profession would actually use." Stable Diffusion was released in August 2022; the page gives no founding date. | [1] |
| Cheapest paid plan | US$10 a month, or US$8 a month billed yearly | C | Pricing file: "Individual | $10/mo | $8/mo ($96/yr) | 200" credits. | [3] |
| Outside funding | none found | C | No funding round is mentioned on the company's pages or in the Web Summit and UpNext pages. | [1, 5, 6] |
| Event | Web Summit Qatar PITCH 2026 | C | Web Summit's PITCH page lists "Fenestra" under "Proptech & real estate". UpNext (4 February 2026) calls it "a spatial AI canvas built for early-stage architectural design". It is not listed there as a winner. | [5, 6] |
| Time from first archived live page to the user count | 30 months | calc | Calculated: May 2024 to November 2026. | [1, 4] |
How to read this table: a grade C value is a statement by the company on its own pages, by the founder, or by an event organiser about its own list. A "calc" value was computed on this page. No revenue figure exists.
| Plan | Per month, billed yearly (US$) | Per month, billed monthly (US$) | Limit |
|---|---|---|---|
| Free | 0 | 0 | 5 to 10 credits on sign-up, non-commercial use |
| Individual | 8 | 10 | 200 credits a month |
| Pro | 28 | 35 | 1,000 credits a month, image editing, unlimited projects |
| Studio 2k | 48 | 60 | 2,000 credits a month, priority speed and support |
| Studio 8k | 192 | 240 | 8,000 credits a month |
| Team Small | 208.33 | 250 | up to 5 seats, 5,000 shared credits a month |
| Team Growing | 416.67 | 500 | up to 25 seats, 10,000 shared credits a month |
| Enterprise | custom | custom | multi-user licences, admin, onboarding |
Billing is by credits: the company states that an image costs about 1 credit, an enhancement about 2 and a video about 10, and that premium models cost more. Credits reset each billing cycle. Studio has four credit tiers from US$60 to US$240 a month; team plans share one credit pool. One-off credit packs cost US$18.99 (250 credits), US$25.99 (450) and US$44.99 (900). Students get 20% off on request. The pricing file lists the Free plan at 5 credits, while the pricing page and the file's own FAQ say new accounts get 10 credits.
When did Fenestra launch, and how did it grow?
From the first archived home page in May 2024 to the redesigned studio and its connectors for ChatGPT and Claude in 2026. Each entry carries its grade.
- May 2024CThe earliest archived home page offers commercially licensed architectural renders from a text prompt and from uploaded .GLB 3D models, in the browser. [4]
- 27 June 2025CThe founder's newsletter "You made it" describes Fenestra as "an AI-powered spatial rendering tool for architects and interior designers" and thanks early users. [7]
- February 2026CFenestra takes part in the Web Summit Qatar PITCH startup competition; team accounts with shared credit pools follow on 13 February 2026. [5, 6, 8]
- 6 April 2026CA redesigned Fenestra Studio opens to everyone after a month with a group of users; live sketch-to-render had opened to all on 24 March 2026. [7, 8]
- April to September 2026CThe changelog adds GPT Image 2 editing and Seedance 2.0 video (24 April), a credit referral programme (21 May), PDF plan import (17 June) and a model browser with four new video models (1 September). [8]
- November 2026CThe About page states 20,000+ architects and designers in 180+ countries, and the site lists a connector that runs Fenestra from inside ChatGPT or Claude. [1, 3]
Why did Fenestra work?
Four causes, in the order they mattered. Each one is a reusable lesson.
Wrap many models in one tool for one profession
Fenestra does not train its own base model. It adds new image and video models soon after they are released (the changelog calls Nano Banana 2 "available in Fenestra on day one") and puts them behind controls that architects need: masks, keeping the geometry, camera moves and resolution chosen per render.
Credits match a cost that changes per model
Each model costs a different amount to run, so the plans sell credits and the app shows the credit cost before each generation. Fast models are cheap for trying ideas; premium models and video use more credits.
A founder from the customer's side
The About page presents the founder's years in architecture practices as the reason for the product decisions: which inputs to accept, which controls matter and what a client-ready image looks like.
Many search pages for single workflows
The site has pages for single tasks and tools (Revit to render, SketchUp to render, virtual staging, timber or heritage buildings) and a file for AI assistants (llms.txt). Whether these pages bring most users is not stated in any source.
How it was built
These details come from the company's own pages, its changelog and its robots.txt file. They are grade C. [1, 3, 8, 9]
- Image models named by the company: Nano Banana 2 and Nano Banana Pro (Google), GPT Image 2, Qwen Image Edit, Krea 2, and the Flux 2 family (Black Forest Labs); video models include Seedance 2.0 and LTX 2.3.
- The site's robots.txt mentions Vercel and Next.js static assets, so the website is a Next.js app hosted on Vercel; the page code loads PostHog, an analytics service.
- A connector at mcp.fenestra.app lets users render and edit from inside ChatGPT or Claude through MCP (Model Context Protocol, a standard way for AI assistants to call outside tools).
- An Internet Archive copy shows a 3D model viewer for .GLB files already in May 2024; later releases added PDF plan import and nine-angle Multi Shot.
Tools named in the sources
Each one opens its page in the Tools tab, with every story that uses it.
Who tried this before Fenestra, and why did they fail?
Three companies in our database of failed startups tried to sell visualisation of rooms and buildings and failed. The values below are from that database and are not independently verified.
Modsy
The promisePhotorealistic 3D renderings of a customer's own room with furniture and decor ideas.
Why it diedHuman designers in the loop and costly customer acquisition kept it from profit.
Kujiale
The promiseCloud 3D interior design and furniture visualisation for consumers and furniture makers.
Why it diedConsumers did not want to design rooms themselves, and the database says AI made its technology obsolete.
Laurel & Wolf
The promiseOnline interior design for a flat fee, through a marketplace of designers.
Why it diedFlat fees did not cover customer acquisition, and quality fell as the designer network grew.
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
Modsy and Laurel & Wolf paid people, or 3D artists, for every room they showed, so each customer had a high fixed cost. Image models now make a render in seconds from a sketch or a photo, and Fenestra resells that model time with a margin through credits instead of paying for design hours.
The models themselves change every few months. Fenestra's changelog shows it adding new outside models in 2026 within days of release, so its value is in the architecture-specific controls around them, not in a model of its own.
Can one person copy Fenestra now?
Yes, for a founder who knows the profession.
The product is a model wrapper with professional controls, and the About page says one founder built it. The hard parts are knowing what architects need and keeping up with new models; the same approach could serve other visual professions.
Where it is still open: Image tools for one profession built on outside models: landscape plans, kitchens and bathrooms, shopfronts, event layouts.
- 1Pick one profession whose clients judge work by images (architects, landscape designers, kitchen fitters).
- 2Accept the files that profession already makes (CAD exports, PDF plans, 3D models) as inputs.
- 3Call outside image and video models through their APIs, and add masks and controls that keep the original geometry.
- 4Sell credits, show the cost before each generation, and add each new strong model quickly.
What could go wrong for Fenestra?
The risks are the mirror image of the causes above.
What is not verified
No revenue figure was found. We looked for a payment-processor profile, a founder interview with numbers, a company filing and a funding announcement, and found none. The 20,000+ users, 180+ countries and 200,000+ renders are the company's own statements; a sign-up count is not the same as paying customers. A reader could ask the founder for paying-customer numbers or look for a revenue profile on a payment-verified site.
Trust signals and small inconsistencies
The home page shows a "Trusted by global teams" logo row with firms such as Jacobs, SHoP Architects and Manchester Metropolitan University, but only apoprojekt, Urbane Homes and VAPF appear with named quotes; the other logos have no stated source. The About page says "One founder" while a customer quote and the blog speak of a team. The pricing file says commercial use is included on all subscriptions, while the pricing page says "Pro and Team plans are commercial use"; the free credit amount is given as 5 in one place and 10 in another.
Competition
AI rendering for architects is crowded; on this site Visualizee.ai, Visualize AI, ArchiRise and ARQ sell similar sketch-to-render tools, and the company itself publishes a comparison with Lumion, a classic rendering program.
Dependence on model providers
Every render runs on an outside model: the company names Google's Nano Banana models, GPT Image 2, Black Forest Labs' Flux models, Qwen Image Edit, Krea 2 and Seedance 2.0. A price change, a usage limit or the end of one model's API would change Fenestra's credit costs, and the same models are open to every competitor and to general apps such as ChatGPT, for which Fenestra also offers a connector.
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.
- C Fenestra, About page (fenestra.app/about), read 6 November 2026Read in full. Source for the founder, the base in the UAE, the user counts and the reason for starting.
- C Fenestra home page (fenestra.app), read 6 November 2026Read in full, and the image descriptions in the page code for the logo row. Source for the repeated user count, the customer quotes and the logos.
- C Fenestra pricing page and plain-text pricing file (fenestra.app/pricing and fenestra.app/pricing.md), and llms.txt, read 6 November 2026Read in full. Source for the prices, credits, model list and the ChatGPT and Claude connector.
- C Internet Archive copy of the Fenestra home page, 6 May 2024Read in full; the capture list shows a redirect on 4 January 2024 as the first capture. Source for the earliest live product.
- C Web Summit Qatar, PITCH 2026 startupsRead the list as downloaded by the pipeline on 6 November 2026. Source for the PITCH selection and the category.
- C UpNext, Edwin J. Frondozo, 4 February 2026: Web Summit Qatar PITCH 2026Read the section on Fenestra, as downloaded by the pipeline on 6 November 2026. Source for the pitch description and the founder's name.
- C Fenestra blog: You made it (27 June 2025) and Fenestra Studio is now open to everyone (6 April 2026)Read both posts in full (the first at fenestra.app/blog/introduction-to-fenestra). Source for the early description and the studio release.
- C Fenestra changelog (fenestra.app/changelog), read 6 November 2026Read in full. Source for the 2026 release dates and the models added.
- C Fenestra robots.txtRead in full on 6 November 2026; it allows ClaudeBot. Source for the Vercel and Next.js hosting note.
- 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