Velvet
YC F25San Francisco, US · Founded 2025 · 4 employees · Hiring · 4 known investors
Velvet creates datasets and evaluations for training multimodal AI models, serving AI labs and enterprises building systems with improved spatial reasoning and latency performance.
Also known as Inline · Lenscape · Velvet Lab
Founders & leadership· Y Combinator alumni (F25)
Velvet was founded in 2025 by Lucas Mantovani and Lucas Tucker.
Investors · 4
Company profile
researched Aug 2026Velvet is a product and research company focused on multimodal models, positioning its work under the heading of world model research and development. It works with AI labs and enterprises to create datasets and evaluations intended to advance the interactive capabilities of AI systems.
The company frames its motivation around two stated gaps in current models: a lack of spatial reasoning capability and high response latency. Its stated objective is to accelerate the development of models that understand the physical world in the way humans do. Velvet's platform accepts dataset samples across text, image, and video media, with submitters providing a description, volume, contact details, and links to image or video samples for review.
Business model
Velvet engages directly with AI labs and enterprises that train or deploy multimodal models, producing datasets and evaluation suites for them. It also operates a contributor-side workflow in which participants submit dataset samples for review and can be paid; the platform supports connecting Stripe or Wise accounts for payouts and displays pending and paid earnings balances.
Traction
The company is hiring across research, engineering, and operations functions.
▸Full profile — market position, technology, go-to-market
Market position
Velvet describes itself as operating in world model research and development, supplying datasets and evaluations to organizations building multimodal AI systems.
Technology
Velvet's work centers on multimodal models and world models, specifically the creation of datasets and evaluations addressing spatial reasoning and response latency. Its data intake supports text, image, and video media types.
Go-to-market
Direct engagement with AI labs and enterprises through meeting requests and outreach forms on its website, alongside an invite-only account system that requires an access code from an invitation. Dataset contributors are onboarded via a sample submission and review process.
AI labs and enterprises that are training or deploying multimodal models.
Compiled by commissioned research from 1 cited public sources — announcements, filings, and press listed under research sources below.
Key figures
latest reportedCompany-reported or press-reported figures, each dated to when it was claimed — not independently audited.
Images
Competitors · 4
by search overlap
Companies competing with Velvet for the same Google search keywords, organic and paid, via search-intersection analysis.
In the news
VELVET stalls at $2—How deep could the current pullback go?cryptonews.net · Jul 2026▸Research sources · 1
primary sources listed
- Velvetvelvetlab.ai · web
1 public sources were cited for this profile; the first-party ones are listed here.
Frequently asked questions
- What does Velvet do?
- Velvet is a product and research company building datasets and evaluations for training multimodal AI models.
- Who founded Velvet?
- Velvet was founded by Lucas Mantovani, Lucas Tucker in 2025.
- Who are Velvet's investors?
- Velvet's investors include Y Combinator, Blockchain Founders Fund, FunFair Ventures, Mindfulness Capital.
- Where is Velvet headquartered?
- Velvet is headquartered in San Francisco, US.