Problem

Relari addresses the difficulty of building AI agents that behave reliably in real business processes. Today, teams must manually encode agent behavior in code, visual workflow blocks, or prompts, while also managing state and memory, coordinating multiple agents, making tool calls dependable, and handling complex errors. That creates a costly mix of engineering effort, operational oversight, and risk: an agent that works in a demo may fail unpredictably in production.

The clearest economic opportunity is automating high-volume, error-sensitive work that people currently review or “babysit.” Nuvi’s highlighted use cases include mortgage-document verification, legal prospecting, invoice validation, contract compliance, and customer onboarding. The mortgage example reports roughly an 80% reduction in processing time, while other published examples claim a tenfold increase in sales-pipeline growth and a 90% reduction in invoice-processing errors.

Product / Service

Relari’s public-facing product is Nuvi, a no-code AI-agent builder built around natural-language specifications. A user describes a task or uploads existing documents; Nuvi translates that specification into an agent, handling the underlying architecture, integrations, prompt optimization, and deployment. The resulting agent can run as an API, embedded chat interface, or background automation, connecting to systems such as HubSpot, Gmail, Slack, Zendesk, Google Drive, Outlook, and Xero.

The differentiator is reliability by construction rather than prompt experimentation alone. User stories become behavioral tests, agents can be simulated before launch and continuously validated afterward, and specifications are readable, editable, and versioned for collaboration. Nuvi is delivered as a product-led SaaS with free and $20-per-month individual plans, a $500-per-month Team plan, and custom enterprise plans offering security controls, deployment options, SLAs, and optional white-glove implementation. Relari also publishes the Agent Contracts and Continuous Eval open-source frameworks that reinforce its testing and evaluation positioning.

Market

Relari competes in the emerging AI-agent builder and agentic-automation market, with an adjacent position in AI-agent testing and evaluation. Its own comparison places Nuvi against n8n, RelevanceAI, OpenAI GPTs, LindyAI, and Lovable: n8n is the more technical visual workflow alternative, RelevanceAI and OpenAI GPTs emphasize simpler prompt-based agents, LindyAI focuses on personal assistants, and Lovable focuses primarily on front-end generation. In the evaluation layer, the broader competitive set includes MLflow, DeepEval, Ragas, Arize Phoenix, and LangSmith.

The company is an active Y Combinator Winter 2024 startup founded in 2024, and Relari says it is backed by Y Combinator, Soma Capital, General Catalyst, and other investors. Its public traction includes multiple named business case studies, paid self-serve and enterprise pricing, and a growing agent library; third-party databases report approximately $500,000 of disclosed funding, while one estimates $330,000 in 2025 revenue. Relari has not publicly confirmed that revenue figure, so the evidence supports an early commercial startup with customer traction rather than a clearly established scale business or a definitively pre-revenue company.