LatticeFlow

Market position

Positions itself as a pioneer of evidence-based AI governance, contrasted with checklist- or documentation-based governance tools; the company was mentioned by Gartner in the inaugural 2026 Magic Quadrant for AI Governance Platforms and, following the AI Sonar acquisition, describes itself as offering the industry's first end-to-end, evidence-based AI governance solution combining on-prem AI discovery with centralized SaaS governance. Earlier, in 2022, it was ranked the #1 Swiss AI startup among more than 170 companies and included in the CB Insights AI100 list.

Combines pre-deployment, use-case-specific evaluations and adversarial testing with continuous production monitoring, rather than focusing solely on policy management or generic LLM benchmarks. Evaluations are deterministic, reproducible and version-controlled, and every governance decision is traceable to underlying technical evidence with full provenance. The company emphasizes Swiss-based engineering, neutrality and independence, its research links to ETH Zurich and INSAIT, and its authorship of the COMPL-AI EU AI Act framework. In its earlier computer-vision product, the differentiator was detecting model blind spots in addition to standard data issues such as labeling errors and poor-quality samples.

Technology

Deterministic risk-scoring methodology that factors impact, domain, deployment context and industry to produce use-case-specific severity scores; YAML-specified evaluation definitions with caching, rate limiting, parallel execution, robustness and repeatability controls; dataset synthesis, custom scorers, LLM-as-judge and Pass@K metrics; RAG and hallucination testing; agentic probes and automated red-teaming aligned to OWASP and MITRE ATLAS; continuous discovery of AI endpoints and connected components across cloud and on-prem; and a technical evidence engine with exportable, fully traceable logs. Roots in ETH Zurich research on formal methods, symbolic reasoning and machine learning.

Go-to-market

Direct enterprise sales supported by demos, expert consultations, executive briefings and published customer stories; partnerships with technology vendors such as SAP; framework-led adoption through the publicly available AI Atlas registry (atlas.latticeflow.ai) and the open COMPL-AI framework; and analyst-report visibility. Press releases are distributed through Business Wire.

Large enterprises and regulated organizations deploying GenAI applications and autonomous agents, including banking and financial services, energy and critical infrastructure, healthcare, industrial technology and public-sector bodies; internally the platform is aimed at AI and risk leaders needing executive visibility, and AI practitioners performing technical assessments and monitoring, as well as GRC teams.

Geography

Headquartered in Zurich, Switzerland, with press releases issued from San Francisco and Zurich. R&D offices in Zurich, Sofia (opened next to INSAIT following the 2022 Series A) and Dublin (added through the January 2026 AI Sonar acquisition). Customers in Europe and the United States; the team includes an Executive Director for Bulgaria.

History

Founded in 2020 as an ETH Zurich spin-off by researchers working on formal methods, symbolic reasoning and machine learning, with the initial goal of building a platform enabling companies to deliver robust AI models that work reliably in production. The company raised a $2.8M seed round in 2021 and a $12M Series A in October 2022, at which point it had more than 10 customers in Europe and the US and announced an office next to INSAIT in Sofia, Bulgaria. It received a CHF 3M Innosuisse grant in October 2024. In January 2026 it announced the acquisition of Dublin-based AI Sonar, a CloudSphere subsidiary, adding AI asset discovery and making Dublin a third R&D site alongside Zurich and Sofia. In 2026 it launched the AI Atlas registry (April), was mentioned in the first Gartner Magic Quadrant for AI Governance Platforms (June), and launched its unified AI risk control platform (July).

Risks & controversies

The company's own materials note that enterprises struggle to approve GenAI applications and agents beyond restricted internal use, and that AI adoption is outpacing governance capability as organizations scale from dozens to thousands of AI systems; its business depends on evolving regulation such as the EU AI Act and on frameworks including NIST AI RMF, ISO 42001, OWASP and FINMA. The July 2026 platform announcement was distributed as a paid press release.