Key Intelligence Insight
V7 is not primarily a data labeling company. It started as one, used that position to learn how documents work at an industrial level, and then built an agent platform on top of that foundation that is now structurally better positioned than most enterprise AI competitors. The Darwin-to-Go pipeline was not accidental -- it was a deliberate migration from teaching AI to deploying AI, timed to the moment LLMs made that transition viable. The result is a company that combines proprietary document intelligence infrastructure with a rapidly expanding go-to-market motion in the most compliance-sensitive corners of the enterprise. That combination is harder to replicate than it looks.
Founding Story
V7 was founded in London in 2018 by Alberto Rizzoli and Simon Edwardsson. The founding insight was specific: AI models need structured, high-quality training data to perform at production-grade accuracy, and no tool existed to generate that data efficiently at scale.
Rizzoli arrived at this problem through an unusual path. His prior company, iPoly, was a computer vision app that identified real-world objects in real time for people with visual impairments. It scanned over two billion objects across 26 languages and earned recognition from the President of Italy. The jump from assistive consumer technology to enterprise data infrastructure was not a pivot -- it was a compression of the same insight. Vision models could transform industries. The bottleneck was training data.
V7 Darwin launched publicly in December 2019 after a beta at CVPR that summer. The product built a reputation in healthcare and medical imaging -- teaching AI to detect cancer in MRI scans, perform microscopy analysis for drug discovery, and handle pathology workflows at companies including Roche and Paige. The mechanism was workflow-driven annotation: instead of labeling data in isolation, Darwin embedded annotation directly into the model development process, with auto-annotation, active learning, and human-in-the-loop review stages built in.
A $3M seed round followed in October 2020. A $33M Series A arrived in November 2022. Total funding now sits at $43M, backed by Radical Ventures, Temasek, Air Street Capital, and a set of AI research heavyweights: Ashish Vaswani (inventor of the Transformer architecture), Oriol Vinyals (Director of Research at DeepMind), and François Chollet (creator of Keras). The investor roster is not decorative. These are people who understand the technical foundation of what V7 is building.
The Go pivot began in earnest around 2023 and launched formally in April 2024. The thesis: the same teams who had been using V7 to label documents for model training were now in a position to use AI models to process those documents operationally. The V7 platform already understood document structure at a level that out-of-the-box LLM calls did not. That infrastructure became the foundation for a new product line.
Product
V7 operates a dual-platform structure. The two products share infrastructure but serve fundamentally different stages of the AI value chain.
V7 Darwin remains the company's foundational data annotation platform, now with a focused application in computer vision and medical imaging. Darwin handles image, video, and specialized medical formats with auto-annotation, SAM2 integration, multi-planar annotation, and model-in-the-loop active learning. Its primary market is AI research teams building domain-specific models -- particularly in healthcare and life sciences. Darwin is available on the AWS Marketplace. It is not the growth story, but it is the credibility and cash-flow foundation that makes Go possible.
V7 Go is the active expansion surface. Launched in April 2024 and expanded aggressively through 2025 and into 2026, Go is an AI agent platform for document-intensive knowledge workflows in finance, legal, insurance, and real estate. The product has three integrated layers:
AI Agents: 300+ pre-built, configurable agents for specific roles and workflows -- financial statement analysis, contract review, lease abstraction, claims processing, due diligence, underwriting review. Each agent can be customized through a no-code workflow builder with conditional logic, branching, and configurable human review stages.
Knowledge Hubs: Launched in August 2025, Knowledge Hubs are permissioned internal data repositories that use V7's proprietary Index Knowledge technology. The system supports up to 10 million data points per hub and 80 proprietary indexing dimensions, delivering accuracy 25% better than standard RAG approaches. This is the company's primary moat -- a corpus architecture purpose-built for regulated enterprise data.
AI Skills: Introduced in February 2026, Skills are composable workflow instructions that standardize repeatable tasks across agents -- covering single-app functions, multi-step workflows, financial model parsing, and deep research tasks.
The platform supports 1M+ token context windows, 400 app integrations, 6,000 pre-understood actions, and connections to major LLM providers including OpenAI, Anthropic, and Google with no model lock-in. Security certifications include ISO 27001, SOC 2, HIPAA, and GDPR.
V7 claims 95-99% extraction accuracy on document processing benchmarks -- above LLMs (80-95%), RPA/IDP tools (80-99%), and hyperscalers (70-80%). The company holds the #1 ranking on Legalbench and FinanceBench.
The implementation motion is compressed deliberately: introductory call to commercial discussion in 11 days, with POC delivery in between.
Market, Competition & Business Performance
Market
V7 targets what it calls the "operations" layer of enterprise AI -- the workflows that answers and task execution cannot yet reach. The framing is precise and intentional: AI tools have commoditized question-answering and single-step task completion. The remaining white space is operational process automation, specifically in document-heavy, compliance-sensitive industries.
The addressable population is large. The business process outsourcing industry -- the current human-labor infrastructure handling these workflows -- represents approximately $270 billion annually. Finance, insurance, legal, and real estate are among its densest segments. These are industries where document volume is high, accuracy requirements are non-negotiable, and regulatory audit trails are mandatory. V7 has positioned itself precisely at that intersection.
The company's go-to-market motion concentrates on enterprise buyers in those four verticals. Customer evidence includes Centerline (35% productivity increase in diligence), Star Mountain (financial statements processed 21x faster, 54% accuracy improvement), and Pinsent Masons (complex document processing at scale). The company also counts GE Healthcare, Roche, Merck, Sony, and Mercedes-Benz among its named customers, though the primary current growth vector is financial services, insurance, and legal.
Healthcare remains a Darwin-era presence rather than a Go target, though the infrastructure overlap is significant.
Competition
V7 competes across three distinct competitive surfaces simultaneously.
The first is intelligent document processing (IDP): Rossum, Hyperscience, ABBYY FlexiCapture, Docparser, and Parseur. These are legacy or semi-legacy tools built for extraction, not for agent-driven workflow orchestration. V7's accuracy advantage and visual source grounding (AI citations that link every extracted value to its exact document location) are structurally difficult for IDP tools to match without rebuilding their architectures.
The second is horizontal automation platforms: Zapier, n8n, and Make. These tools have significant integration breadth -- Zapier covers 8,000+ apps -- but they were not built for document intelligence. The mechanism: connecting apps is categorically different from understanding documents. V7's advantage in this quadrant is not feature count; it is extraction accuracy and audit trail capability that horizontal platforms cannot provide natively.
The third is vertical or enterprise AI competitors: Scale AI, Nanonets, Super.AI, UiPath, and Automation Anywhere. Scale AI's primary business is data annotation for model training -- adjacent to Darwin, not Go. UiPath and Automation Anywhere operate in RPA, which executes structured workflows on structured data; unstructured document processing remains a weakness. Nanonets and Super.AI are closer competitors in document extraction, but neither has V7's 300+ pre-built agent library or its proprietary Knowledge Hub infrastructure.
The more interesting competitive question is the hyperscaler ceiling. Google Document AI, AWS Textract, and Azure Form Recognizer all compete in document extraction. V7's benchmark accuracy claim (95-99% vs. 70-80% for hyperscalers) is the central differentiator, combined with workflow orchestration capability that cloud-native extraction services do not provide. Whether that accuracy gap persists as hyperscalers invest in their document AI layers is the defining competitive question for V7's next phase.
Business Model
V7 uses a three-component enterprise pricing structure: a base platform fee for access to the Go agent library, per-seat user licensing, and volume-based data processing charges. Pricing scales with document throughput, not feature access. The model is usage-based at the margin, which aligns vendor incentives with customer adoption -- expanding usage generates expanding revenue without additional sales motion.
No list pricing is published. All contracts go through a consultative sales process with white-glove implementation: dedicated solution engineers build custom integrations, configure industry-specific agents, and deliver POCs before commercial commitment. The 11-day first-call-to-commercial-terms timeline reflects a deliberate motion to compress sales cycles and reduce procurement friction.
V7 Darwin operates on a separate pricing track -- also enterprise and usage-based, structured around annotation volume and platform access.
Traction
V7 has 104 LinkedIn-reported employees as of early 2026, reflecting 22% annual headcount growth. Engineering grew 37% year-over-year. Sales grew 27%. Customer Success grew 33%. Arts and Design doubled. The hiring pattern signals a company moving from product-building to go-to-market scaling -- with the engineering growth indicating continued platform investment in parallel.
The company reached named enterprise customers across financial services, insurance, legal, construction, real estate, healthcare, and pharma. Cited ROI metrics from customers include $2M+ in annual operational savings, 10+ hours saved per investment review, and 100+ properties analyzed per hour in real estate. These are self-reported and should be read as directional, not audited.
Partnership activity accelerated through 2025: a strategic integration with DMS in February, a partnership with Maai Services Group for M&A automation in August, and a partnership with Alchemy for insurance document processing in August. The March 2026 "Finally, Free to Think" campaign marks V7's first major brand-level marketing push -- typically a signal of a company transitioning from product-market validation to category-building.
V7 was ranked #1 on Sifted's B2B SaaS Rising list in June 2024. The company operates from London HQ with offices in San Francisco and New York.
The core question for V7 is whether its accuracy advantage and Knowledge Hub infrastructure compound into a defensible position before hyperscalers and better-funded horizontal platforms close the gap. The evidence to date suggests the answer is yes -- but the next 24 months will determine whether that holds at enterprise scale.