Engineering teams across real estate, enterprise software, and regulated SaaS are using Culvii to deploy AI agents they can actually put in front of regulators.
JLL manages commercial real estate portfolios for some of the world's largest corporations. They needed AI agents for lease analysis and portfolio valuation — but legal and compliance blocked every previous attempt.
JLL's platform team wanted to automate lease analysis, portfolio valuation, and market research — work consuming thousands of analyst hours per year.
The problem: lease analysis involves legally consequential decisions. A single misread clause can cost a client millions. Legal and compliance teams weren't opposed to AI — they were opposed to AI they couldn't audit.
Their existing options gave them no answer. LangChain could chain prompts together, but it had no concept of agent identity, no access controls, and no audit trail that would hold up in a compliance review. Building those things from scratch was estimated at 4–6 months of platform engineering work.
JLL integrated Culvii Kit into their portfolio management platform in three weeks.
Every AI agent involved in lease analysis was registered with a cryptographic DID identity. Capabilities were scoped precisely: a lease-extractor agent could read lease documents but not write to the valuation system. The valuation-writer agent could write valuations but couldn't access raw client documents.
For high-value leases — anything above $1M annually — the platform team used Culvii Kit's HITL primitive to route execution through a legal team approval checkpoint before the agent's output was committed. Approvers received a Slack notification with full context and approved or flagged with one click.
Every step — every document read, every model call, every HITL approval — was logged in an immutable Merkle Tree audit trail.
Quote pending approval.
Visot builds workflow automation software for enterprise clients. Per-tenant agent isolation was becoming a bottleneck, consuming platform engineering time that should have gone to product.
As Visot expanded their agent capabilities, they ran into a consistent problem: each new enterprise client needed customised agent access controls, isolated audit trails, and assurance that their agents couldn't access another client's data or capabilities.
Building per-tenant agent isolation from scratch was becoming a bottleneck. Every new client engagement required platform engineering time to customise the identity and access layer. The team was spending more time on infrastructure than on product.
Visot replaced their custom agent execution layer with Culvii Kit's Actor and Workflow primitives, backed by Culvii Kit's multi-tenant registry.
Each enterprise client became a Culvii Kit tenant. Agent identities, capabilities, and audit logs were automatically isolated per tenant — no custom code required. New client onboarding went from a multi-week engineering engagement to a configuration step.
The Workflow primitive allowed Visot to recreate their complex, multi-step automation flows with built-in HITL checkpoints — letting enterprise clients configure approval gates without Visot's engineering team being involved.
Quote pending approval.
AI agents for regulatory filing review with HITL approval chains and tamper-evident audit trails.
Multi-step contract analysis workflows with cryptographic agent identity and capability-scoped tool access.
We're working directly with a small number of engineering teams to deploy Culvii in production. If you're building AI agents in a regulated industry, we'd like to talk.