Knowlee vs MarvelX (2026): Horizontal Agentic OS vs Vertical Insurance OS

Quick verdict. MarvelX is an agentic AI platform for regulated industries that started with insurance, shipping its ClaimOS product — live with Companjon for embedded insurance — that automates up to 90% of claims processing. It is built by a team with deep FSI and AI pedigree (ex-bunq Head of Data & AI) and is designed to plug into existing insurance systems without a rip-and-replace. Knowlee is a horizontal agentic operating system where insurance is one tenant among many: the same runtime that manages claims automation also manages sales pipeline, talent screening, legal review, and ops events, with a Brain that compounds intelligence across all of them. Pick MarvelX when the specific problem is automating insurance workflows in a regulated FSI context. Pick Knowlee when insurance is one of several business functions you need an AI workforce to run as a coherent system.

What each platform actually is

MarvelX (marvelx.ai, Amsterdam, founded 2025, $6M Seed led by EQT Ventures and Plug and Play, founded by former bunq Head of Data & AI) is an agentic AI platform for regulated industries, starting with insurance. Its ClaimOS product is live with Companjon for embedded insurance, automating up to 90% of claims processing — connecting to existing insurance systems, centralizing operational data, and turning siloed manual claims tasks into intelligent automated flows. The founding team's FSI background (bunq is one of the fastest-growing European neobanks) shapes how MarvelX thinks about compliance, data integration, and regulated-industry deployment.

Knowlee is a horizontal agentic OS that runs a fleet of AI agents across multiple business functions — insurance claims, sales, legal, talent, ops — as a single coherent system. Insurance workflows in Knowlee (claims processing, policy review, underwriting support) run on the same runtime as sales research jobs and legal review jobs, feeding the same Neo4j Brain that accumulates signals across all verticals. An insurance signal — a claims pattern, a high-risk customer segment, a fraud indicator — can surface as relevant context in a sales job, a legal job, or an ops job without manual routing between systems.

Architecture difference: vertical insurance OS vs. horizontal agentic OS

MarvelX: depth in regulated FSI, starting with claims

MarvelX's architecture centers on the claims event as the primary unit of work. ClaimOS connects to the existing systems of record in an insurance operation — policy management, customer data, payment infrastructure — and turns the claims workflow into an intelligent automated flow: ingest the claim, validate against policy, assess the loss, route edge cases to human review, resolve and pay. The 90% automation rate for Companjon suggests this is not a surface-level integration but a genuinely deep replacement of manual processing steps.

The regulated-industry framing matters. MarvelX was built from the start for FSI compliance requirements — not retrofitted. The bunq pedigree means the founders understand what "connecting to existing systems" means in a banking or insurance context: core system integrations, data sovereignty requirements, audit trail obligations that differ from standard SaaS.

Knowlee: insurance as one tenant of a wider OS

Knowlee's insurance job tier handles the same claims workflow as a typed job with declared inputs (claim event, policy data, validation rules), outputs (resolution, payment trigger, escalation flag), governance metadata (risk_level, data_categories, human_oversight_required, approved_by), and an audit trail. The difference is the Brain layer. An insurance job in Knowlee writes claim patterns, customer risk signals, and fraud indicators to the Neo4j graph alongside sales contact history, legal clause patterns, and talent screening outcomes. Cross-vertical compounding is structural.

The MCP Model Context Protocol fabric means an insurance job can trigger downstream jobs automatically: a resolved claim can update a CRM record, flag a cross-sell opportunity, trigger a legal review if a clause is disputed, or update an account risk score visible to the sales team.

Side-by-side comparison

Where MarvelX wins

MarvelX is the right tool when the problem is specifically insurance claims automation in a regulated FSI environment:

• 90% claims automation with a live production reference. Companjon is a real embedded insurance deployment, not a pilot. MarvelX is operating in production at a level of automation that most platforms cannot match for this specific workflow.
• Regulated-industry depth from the ground up. The bunq founding team brings genuine FSI pedigree. MarvelX was not built for generic enterprise automation and then adapted to insurance — it was built for regulated industries from the start.
• Existing system integration. Insurance operations run on legacy core systems. MarvelX plugs into those systems rather than requiring a rip-and-replace. That integration depth is a real deployment accelerant for established insurers.
• EQT Ventures + Plug and Play backing. EQT Ventures and Plug and Play both have FSI investment and accelerator networks. For an insurance company evaluating a young vendor, that institutional backing provides credibility and ecosystem access.
• Vertical focus as a feature. If the problem is insurance claims and only that, MarvelX's narrow scope is an advantage. Simpler procurement, faster deployment, no OS-level overhead.

Where Knowlee wins

Knowlee is the right tool when insurance is one of several functions that need an AI workforce, or when cross-vertical compounding matters:

• Multi-vertical operation. Insurance claims alongside sales intelligence, legal review, talent screening, and compliance monitoring — Knowlee runs them as one coherent fleet. MarvelX solves one vertical.
• Cross-vertical compounding. A claims pattern that surfaces as a sales signal (this customer segment has high loss ratios — reconsider the cross-sell strategy), or a fraud indicator that informs a legal risk flag — that compounding requires a shared Brain. MarvelX's intelligence is insurance-scoped.
• EU AI Act governance as schema. Knowlee's job metadata is structural at creation for every job type. For organizations under EU AI Act high-risk obligations across multiple functions, that structural compliance is a meaningful advantage over per-vertical compliance implementation.
• Brain that compounds across verticals. Insurance signals → sales context → CSM signals → legal flags — the value of the Neo4j Brain grows with each vertical that writes to it. A single-vertical OS does not compound this way. See multi-agent orchestration.
• Operator-grade runtime across functions. A COO or platform lead who needs one interface showing what every AI agent is doing across insurance, sales, legal, and ops gets that from Knowlee's kanban. MarvelX's UI is claims-workflow-specific.
• No dependency on a single vertical's roadmap. As the AI workforce expands beyond insurance, Knowlee scales horizontally without adding a new platform per function.

Decision framework

The insurance operations leader. You process insurance claims — embedded insurance, travel insurance, device protection — and the bottleneck is manual processing at scale. You need 90%+ automation with a system that plugs into your existing infrastructure and satisfies FSI regulatory requirements. → MarvelX / ClaimOS is the right starting point. It is purpose-built for this problem and has a live reference in a comparable deployment.

The COO or CTO building a multi-function AI workforce in regulated industries. Insurance claims are a significant need, but so are sales intelligence, talent screening, and legal review. You need all of them coordinated around the same account and customer record, feeding the same Brain, with one audit trail that satisfies regulators across all functions. → Knowlee is the right architecture. Insurance is one tenant; it competes on shared infrastructure with every other function.

The holding company or platform team. You run multiple regulated businesses — insurance, lending, compliance, advisory — and need each to benefit from shared intelligence. A claims pattern at the insurance unit should inform the risk pricing at the lending unit should inform the compliance posture at the advisory unit. That cross-entity compounding requires a shared Brain. → Knowlee, explicitly. That is the cross-vertical compounding model.

For more on multi-vertical architecture see Knowlee vs CrewAI and agentic OS vs agent platform in 2026. For governance context, see agentic process automation explained and agentic workforce platforms comparison.

Book a deployment review | See the platform | Compare with Taktile