Unitary
Unitary builds AI-powered Virtual Agents that automate complex, repetitive operational workflows directly in the tools businesses already use. Its mission is to make automation simple, reliable and low-risk, with human oversight and measurable outcomes.
At a Glance
- Managing general agents (MGAs)
- Insurance brokers and carriers
- Financial-services organisations
- Healthcare organisations
- +3 more
AI Tools by Unitary
(1)
Detoxify
Python Toxic Comment Classifier
Discussions
No discussions yet
Be the first to start a discussion about Unitary
Latest News
The Practical Innovator: Unitary published an interview with Crux Underwriting CDO Sam Worthington on durable AI and automation strategy for insurance.
Unitary joined the Accelerant Marketplace, making its Virtual Agents available to more than 300 Accelerant MGA members.
Unitary announced UK-only and EU-only LLM processing, alongside US-only processing, for customers with regional data-governance requirements.
Unitary joined ILC Property as a corporate partner, extending its engagement with the property-insurance ecosystem.
Products & Services
Custom-built software workers that complete end-to-end operational workflows across existing systems. They combine deterministic software for predictable steps, AI reasoning for interpretation and judgment, and human-in-the-loop escalation for exceptions.
Industry-specific Virtual Agents for insurers, MGAs and brokers, covering workflows such as submissions, claims, renewals, compliance, billing, policy servicing, loss runs and application ingestion without requiring system replacement or complex integrations.
Open-source PyTorch/Transformers models and code for toxic-comment classification, including original, unbiased, multilingual and smaller model variants. It predicts categories such as toxicity, threats, obscenity, insults, identity attack and sexual explicitness; the repository is built by Laura Hanu at Unitary and intended for research, fine-tuning and moderation assistance.
An off-the-shelf classifier announced by Unitary for content classification.
Market Position
Unitary positions Virtual Agents as a fully managed alternative to RPA, generic AI agents, workflow builders, narrow AI tools and traditional BPO. Its differentiation is end-to-end execution across existing applications without system replacement or lengthy integrations, combining deterministic controls and AI reasoning with human escalation, auditability and contractual accuracy/savings SLAs. Its insurance case study reports 99%+ automation, 50% lower costs and a 19% CSAT increase for Attune.
Leadership
Founders
Sasha Haco
Theoretical physicist; completed a PhD at the University of Cambridge, worked with Stephen Hawking on the black-hole information paradox, and was a visiting fellow at Harvard. She joined Entrepreneur First in 2019, where she met James Thewlis, and is Unitary's co-founder and CEO.
James Thewlis
AI and autonomous-systems specialist with an MEng in Computer Science from Imperial College London and a DPhil from the University of Oxford focused on AI for visual recognition. He previously worked with Facebook AI Research and joined Entrepreneur First in 2019, where he met Haco; he is Unitary's co-founder and Chief Scientist.
Executive Team
Sasha Haco
Co-founder and CEO
Former theoretical physicist with Cambridge PhD research involving Stephen Hawking and a Harvard visiting fellowship; met Thewlis at Entrepreneur First in 2019.
James Thewlis
Co-founder and Chief Scientist
AI researcher with Imperial College London and Oxford degrees; previously worked with Facebook AI Research on machine-learning solutions and researched visual recognition.
Board of Directors
Founding Story
Haco and Thewlis met at Entrepreneur First in 2019 and wanted to apply advanced research to a difficult real-world problem. They started Unitary to build multimodal AI that could understand online visual content in context and help platforms moderate harmful material more effectively than human-only review or simple keyword systems; the company later expanded that automation capability into complex business operations.
Business Model
Revenue Model
Fully managed, outcome-based Virtual Agent service sold on a fixed monthly subscription/fee based on workflow complexity, expected volume and completed work. Unitary builds and configures the automation without setup or implementation fees, runs a no-commitment three-month proof of concept, and charges once results are delivered; pricing is based on workflow complexity and volume rather than AI-token usage.
Pricing Tiers
Unitary builds, configures and tests the automation for free, then customers run it in real workflows and can cancel during the POC.
Fee is determined by process complexity and volume; includes hosting, monitoring, maintenance, dedicated support, continuous improvement, security and accuracy/savings SLAs.
Target Markets
- Managing general agents (MGAs)
- Insurance brokers and carriers
- Financial-services organisations
- Healthcare organisations
- Online marketplaces and digital platforms
- Social networks, ad-tech and brand-safety organisations
- Insurance submissions, application ingestion, claims intake, loss runs, renewals, policy servicing, compliance, billing and bordereaux
- Financial-services onboarding, clearing, applications and reporting
- Healthcare patient, referral, rota and prescription administration
- Marketplace seller, buyer, moderation and customer-support operations
- Content moderation and brand safety using multimodal classification
- Replacing or augmenting repetitive BPO and back-office work
- Attune
- Trustpilot
- Substack
- Depop