Articul8

Articul8 is an enterprise generative AI software company that builds domain-specific models and agent systems for complex, regulated industries. Its platform is designed to run inside customer environments — cloud, customer VPC, on-premises or air-gapped — so that data, training and inference remain within the enterprise security perimeter. Core components include ModelMesh, described by the company as an autonomous agentic reasoning engine or "agent of agents" that assembles groups of domain-specific agents for a given task, and LLM-IQ, which handles model evaluation and dynamic routing. Additional platform capabilities include an agent factory for building custom agents, multi-agent squads, a knowledge graph linking entities, topics and relationships, no-code MicroApps with industry templates, synthetic and augmented data generation, open APIs and connectors, and support for structured and unstructured text, tables, images, PDFs and streams.

The company maintains a library of domain-specific models (DSMs) and expert agents: A8-Semicon, a Verilog-capable model for semiconductor engineering and chip design workflows; A8-Energy, developed with EPRI and trained on more than 10,000 specialized energy datasets; A8-SupplyChain for manufacturing and supply chain operations and technical documentation reasoning; and A8-Fin for finance tasks such as tabular analysis, portfolio management and compliance. It also markets telco-oriented capabilities, including a Weave Network Topology Agent for network discovery that converts log files and topology images into a queryable graph, and manufacturing design-review workflows using specialized supplier, modeling and process agents. Observability, auditability and traceability of reasoning flows and agent decisions are presented as core features, and the company states SOC 2 Type II compliance.

Articul8 rather than relying on shared general-purpose models emphasizes specialization: press coverage referenced by the company frames its position as arguing that enterprises need better models tuned to their data rather than larger general models, and that ModelMesh differs from frameworks such as LangChain or LlamaIndex by combining Bayesian methods with specialized language models to judge output correctness and maintain consistency across industrial processes.

Founding story

The company describes its origin in the premise that generic AI tools were not built for the complexity, scale or mission-critical nature of enterprise work, with founders and early team members drawn from environments spanning cloud infrastructure to aerospace manufacturing. The business was incubated inside Intel — with an early deployment at Boston Consulting Group beginning roughly two years before the spinout — before being established as an independent company by Intel and DigitalBridge in January 2024 under CEO Arun Subramaniyan, who had previously led aerospace engineering and AI work and served as a vice president and general manager in Intel's Data Center and AI Group.

Business model

Articul8 sells its platform as a core subscription with tiers scaled to team size, data volume and growth stage, deployed either as Articul8-hosted, in the customer's VPC, or on-premises/air-gapped. Add-ons include pre-built customer-specific applications, domain-specific agents and domain-specific models. The platform is also listed on the AWS, Microsoft, Google Cloud and Databricks marketplaces. Rather than licensing standalone models, the company packages its technology as software applications and AI agents tailored to specific business functions.

Recurring subscription revenue from a tiered core platform license, expanded through add-ons such as domain-specific models, domain-specific agents and customer-specific applications, and through cloud marketplace transactions. The company reported more than $90 million in total contract value from 29 paying customers and expected annual recurring revenue of just over $57 million for the year as of January 2026, describing itself as revenue-positive.

Traction

As of January 2026 the company reported more than $90 million in cumulative total contract value from 29 paying customers, including Hitachi Energy, AWS, Franklin Templeton and Intel, and expected annual recurring revenue of just over $57 million for the year, with roughly 45-50% already recognized. It employed 75 people, about 80% in R&D, across the U.S., Brazil and India. The company cites platform metrics such as 4x faster time to ROI, 100% decision traceability, 90% accuracy in domain tasks, 70% less human-in-the-loop time, and use cases scaling to more than 10 billion tokens, along with model benchmark results for its semiconductor, energy, supply chain and finance models.

Latest developments

In January 2026, Articul8's CEO said the company had secured more than half of a planned $70 million Series B at a $500 million pre-money valuation, structured in two installments; the first was led by Madrid-based Adara Ventures, with participation from India's Aditya Birla Ventures, and the round was expected to close in the first quarter of 2026. The CEO said the company was revenue-positive and not cash-strapped, expected to end the year with annual recurring revenue of just over $57 million, and would use proceeds mainly for research and product development plus international expansion in Europe and Asia. The company also markets a growing DSM library — including A8-Semicon, A8-Energy, A8-SupplyChain and A8-Fin — and a Weave Network Topology Agent for telecom and network operations.