platoseed
Build governed AI agents on your own data, without code
Clarm Atlas 10xes productivity for non-technical operators in high-trust, regulated teams - banks, healthcare, insurance, publicly listed companies - by letting them build their own, fully governed AI agents. Teams build agents on their own approved data: every answer cites its source, all primitives are deterministic and highly governable, every action is logged for audit, and you bring your own AI model.
Clarm provides governed AI agents designed for regulated teams, enabling non-technical users to build agents grounded in approved sources with audit trails and guardrails. It emphasizes compliance (SOC 2 Type II, HIPAA) and a no-code workflow inside pre-approved rails. The product targets operations, compliance, and customer-facing teams in regulated industries.
Atlas lets operators build multiple agents that draft from approved sources, route outputs for human approval, and log an auditable trail. Build blocks are pre-approved by compliance; sources, permissions, and guardrails are configured once. Agents operate inside existing tools (Teams, Slack, email, web) and stay within policy, with optional human review. The lifecycle includes drafting from sources, routing for approval, and an audit trail that records sources cited and building blocks used. Deployments support on-premises or air-gapped options for strict data residency and SOC 2 Type II and HIPAA compliance.
Who it’s for: Regulated enterprises with strict governance needs, including banks, healthcare providers, and public companies requiring source traceability, pre-approved guardrails, and enterprise procurement workflows.
Customers: Legacy
Backed by Y Combinator; actively mentions pilots and waitlist launches, indicating early traction and ongoing customer onboarding
I speak 8 languages. Built Clarm to automate inbound (2025-?). Built the world's most accurate consumer goods forecasts using AI at NielsenIQ (2022-24). $0-7m ARR in 2y. Visiting Research Fellow at the University of Oxford; published AI swarm research with a MIT journal (2021). Led AI at Deutsche Bank SeS before AI was cool (2019-21). $10m ARR in 2y. Banker at Deutsche Bank. Financed bridges, markets, hospitals etc from Peru to Taiwan. MSci Physics, Imperial College London.
Build fully automated workflows for high-trust environments
Clarm builds an AI deep research agent platform for high-trust environments, connecting 40+ enterprise tools to create precise, repeatable workflows that improve information retrieval and decision support for industries like law, asset management, luxury retail, and services firms.
Formerly “Dialin - Dialogue Insights”, “dialin” · why startups rename →

AI agents for private market diligence, monitoring, and reporting

The Control Plane for Enterprise Agents