MarvelX vs Weav.ai (2026)

Case documents are collected from the insurer's systems, and each decision is written back into them. Inbound, documents arrive by upload or are collected straight from the customer's systems, so the platform reaches into the insurer's systems to fetch case material instead of waiting to be fed. Outbound, once the output is delivered to the customer's system, the data is wiped from MarvelX's.

The decision is written back into the customer's system of record as the normal course of operation, and the zero-retention design depends on that write-back, which makes it stronger evidence than a general claim about integration. ClaimOS MaX, the earlier product, was described as integrating legacy and modern insurance systems, connecting to current tools quickly, updating the claims or case management system automatically and integrating through APIs and webhooks.

Integrations is a named platform component, listed again on the trust center under Product Security, so it is governed as part of the product, not offered as professional services. The counterparty class is the hard one in this market. Claims and policy administration systems are the legacy systems that make the category difficult, and the older site promised to connect with current tools and systems in minutes, not months. A cloud marketplace listing is a purchase channel, not an integration. No named counterparty system, connector catalog or protocol is confirmed, so what is known is the two way flow with the customer's systems.

Sourcemarvelx.ai home solution pipeline and security sections, platform navigation, trust.marvelx.ai product security sectionread 2026-09-14

Data moves both ways between Weav.ai and external and internal systems, through agentic methods or traditional APIs. A named pre-built catalog covers Guidewire core systems, Verisk, Nearmap, OSHA, HPD and Google data sources, and Nexus audit systems. With the Rapid Insurance Connector framework, Weav.ai or the customer's own IT team can build a new connector within a day. The FNOL Processor writes outbound with authentication, creating the claim directly in the customer's claims administration system. Reach spans core admin, data providers, audit, CRM and ITSM through MCP clients.

SourceWeav.ai, weav.ai/platform and /agentic-toolsread 2026-09-07

The claims flow runs as a numbered pipeline, worked by distinct agents under rules the insurer writes. The case moves through five stages. Documents arrive by upload or collection from the customer's systems, the case is read whole with every field scored, and the agents then decide on coverage, fraud and settlement, with every conclusion citing its source, before an expert gate and improvement in the Agent Gym.

Extraction, reasoning across policy and evidence, three distinct decisions and then approval add up to conditional, multi stage execution, not a single model call, and the worked example of fourteen documents for one claim shows the fan in the sequence handles. The agents are part of the product architecture.

AI Agents and Workflows are separate platform components, and five agent functions, among them Data Extraction, Claims Processing and Fraud Detection, cover the claims lifecycle. The composition belongs to the customer, held to its rules so the same result comes every time, and a published customer account describes an agent customized to its specific products. Rules the insurer authors, applied deterministically, are orchestration configured, not hard coded. No orchestration runtime is named, and there is no workflow graph or versioning.

Sourcemarvelx.ai home solution pipeline, platform and solutions navigationread 2026-09-14

The Agentic Decision Engine coordinates ingestion, clearance, triage, risk, exposure and loss history analysis through to underwriting and premium audit recommendations. Ingestion evaluation agents run alongside extraction agents, and the MCP layer orchestrates agentic workflows across multiple models. So an insurance case runs end to end as multi-step, multi-agent execution on Weav.ai's own platform.

SourceWeav.ai, weav.ai/platformread 2026-09-07

Work starts when case documents arrive, including documents collected from the insurer's systems with nobody at a screen, and it runs around the clock. The first step of the pipeline is documents arriving by upload or collected straight from the customer's systems. Collection straight from the insurer's systems is work reaching the agent without a person, and upload is the alternative, not the requirement.

In claims the arriving document is the trigger (first notice of loss, a repair schedule, an invoice), so document arrival is the domain's native event. The input is varied, not a single feed, with policies, photos, invoices and handwriting, and fourteen documents on one worked claim. Continuity is MarvelX's own and explicit, though it sits on the older security page, not the rebuilt home page, which promises 24/7 workflow continuity to run critical processes nonstop, even outside business hours.

Real-time fraud signals are monitored, and updates go out consistently across customer channels. Channel coverage is real but narrow, with documents in several formats including handwriting and photographs, plus a customer-support agent with its own page implying an inbound service channel. There is no named scheduler, subscription model or queue interface, and while webhook support has been stated, no current page names a webhook.

Sourcemarvelx.ai home solution pipeline step 01, marvelx.ai/security 24/7 workflow continuity tileread 2026-09-14

The FNOL Processor takes first notice of loss from wherever it arrives, a submission inbox, portal or phone, creates the claim and keeps updating it as new information comes in. The Insured Portal Builder sends document requests and reminds on outstanding items on a schedule. Email is among the connected surfaces. Both tools that carry this intake from several channels need a consultation at launch, and the core platform starts ingestion inside existing systems with no trigger surface for the customer, no webhook, schedule or queue the customer configures.

SourceWeav.ai, weav.ai/agentic-tools and /platformread 2026-09-07

Each claim decision is grounded in that case's own documents, with every conclusion citing its source, but no knowledge store survives between claims. The pipeline is bounded by the documents. They arrive by upload or are collected straight from the customer's systems, and the AI reads the whole case, policies, photos, invoices and handwriting, scoring every field, with the worked example naming fourteen documents for one claim.

The grounding is real and tracks provenance, since every conclusion cites its source, and citing source documents is genuine grounding. There is no kept structure to add customer knowledge to. MarvelX keeps zero data retention, processing claims in a transient state and wiping the data once the output is delivered to the customer's system.

A product that wipes the data after delivering the output assembles context per claim and discards it instead of keeping a retrieval structure over the customer's knowledge, by MarvelX's own design choice, not by omission. No vector store, index or retrieval layer is named. The Vault, a named platform component, is the one place a persistent document store for a customer could live, and MarvelX does not describe it. The domain specific models are trained on insurance data, which is training, not retrieval.

Sourcemarvelx.ai home solution and security sectionsread 2026-09-14

The platform ships a pre-configured library of knowledge graphs representing business entities, their relationships and hierarchy by industry and line of business, configurable by business users, grounding the score cards and next best action recommendations. The graph persists, is maintained as a library and can be queried across cases, so it is a maintained retrieval structure, not context assembled per run.

SourceWeav.ai, weav.ai/platformread 2026-09-07

Nothing persists on MarvelX's side, and it says so outright, with claims data wiped once the decision is delivered. MarvelX sells zero data retention, processing claims in a transient state and wiping the data once the output is delivered to the customer's system. The trust center points the same way, listing certificates of destruction under Data Security. The absence is visible in MarvelX's own words, not inferred from a silent site, and for an insurance customer it is a deliberate design choice meant to reassure.

There is no memory layer of any kind, with no session state, no durable artifact the agent writes, no scope, lifetime or expiry, and no carry-over between claims. The claim that agents learn and improve over time, adapting to new data and regulatory requirements, is absorbed learning in a model, not context an agent reads back. There is no conversational surface either. Work arrives as documents from the customer's systems, not as a user turn. A vault in a zero retention design is more likely a credentials or document transit store than an agent memory.

Sourcemarvelx.ai home security section, trust.marvelx.ai data security sectionread 2026-09-14

Beyond the knowledge graphs and scorecards, which are the application's decisioning assets and not memory, there is no memory layer with a set scope and lifetime. Deleting those assets deletes the product configuration, not a remembered state, and there is no per-user, per-team or per-workflow memory with a lifetime. There is no chat session state either.

SourceWeav.ai, weav.ai/platform and /agentic-toolsread 2026-09-07

An expert approval gate stands before each decision goes out, with uncertain cases routed to a person under rules the insurer writes. MarvelX says it twice. Every approval passes the customer's gate, and in the five-step pipeline, step four sits between the decision and the outcome, where the customer's experts agree, the case is held to the customer's rules for the same result every time, and doubt goes to a person, under the captions Your rules and not a model.

An explicit routing of uncertain cases to a human, with the customer's rules deciding instead of model judgment, is a shipped review-and-approve surface, and it is a gate before commit, not control after it. The commitment is consistent across the site. The hero promises to drive claim costs down to 1 euro while keeping the customer's experts in control, and the older security page carries a human-in-the-loop tile about keeping control where it matters so that automation never sacrifices oversight.

Human specialists shift from manual entry to orchestration and quality control, and MarvelX's economics, with experts focused on the exceptions instead of the routine, only work if exceptions actually reach them. A command center for reviewing decisions, with anomalies flagged for human review, appeared on the previous site and is not in the current navigation. There is no approval queue interface, escalation matrix or configurable threshold. The gate is known from repeated commitments of MarvelX's own and its place in the pipeline.

Sourcemarvelx.ai home hero and solution pipeline step 04, marvelx.ai/security human-in-the-loop tileread 2026-09-14

Humans in the loop have full control to review and approve critical decisions. Ingestion evaluation AI agents evaluate the extraction and send conflicts and low confidence data to human review, and audit mechanisms and human in the loop checkpoints are among the stated business benefits. So Weav.ai ships its own review and approve surface, with a set rule for what reaches it.

SourceWeav.ai, weav.ai homepage and /platformread 2026-09-07

A scoped ISO 27001:2022 certificate with its statement of applicability is available, alongside SOC 2 attestation and named access and data controls. MarvelXai B.V. holds an ISO/IEC 27001:2022 certification for its ISMS, covering the development, deployment and management of AI-driven software solutions for the finance and insurance industry, including data privacy management and processes supporting responsible AI development and deployment, with Certi-Trust as the certifying body and an ISO 27001 statement of applicability published as a document.

A scoped certificate with an SoA says what was certified, not merely that something was. The trust center also lists SOC 2 Type 1, with annual SOC 2 Type 1 reporting, Type 2, AIUC-1 and GDPR, with documents behind an access request, and MarvelX's security page states SOC 2 Type I attested. Access controls include access log management and automated account management, the policies include an access control policy, and data security covers access monitoring, data asset classification and certificates of destruction. A second data control is unusually strong.

MarvelX keeps zero data retention, processing claims in a transient state and wiping the data once the output is delivered to the customer's system. MarvelX runs on AWS, GCP and Azure, with attestation reports reviewed annually, which speaks to the host, not to MarvelX. No single sign on, SAML or role based access control is named as a product feature, the one gap in an otherwise deep surface, and the certificate documents themselves are gated.

Sourcetrust.marvelx.ai overview, compliance, product security, access control and data security sections; marvelx.ai home security section; marvelx.ai/securityread 2026-09-14

Single sign on works with Okta, Azure AD, Auth0 and other SAML compliant authentication systems, so identity runs through the customer's own provider, and Weav.ai is SOC 2 Type 2 compliant. There is no trust center, no named RBAC or audit log, no detail on the attestation beyond one sentence, and no compliance page a customer can read without a call.

SourceWeav.ai, weav.ai/platformread 2026-09-07

Audit logging is a listed product feature, and every conclusion the agents reach carries its source. Audit logging sits under Product Security alongside Data Security and Integrations, as a property of what the customer buys, not of how MarvelX runs its office, Continuous Monitoring adds event and audit log management, and Access Control adds access log management. The record of each run is specific to this product.

Every conclusion cites its source, and the pipeline runs from messy documents to an audited decision. A decision that carries the document and field it was derived from is a reconstructable run, a record of why this agent decided this, not a dashboard of throughput. The company sells itself as the audit-ready AI workforce for modern insurance carriers and TPAs, and on a regulated claims product the audit trail is the thing being sold. What is logged and cited is the agent's decision on a claim.

The fraud monitoring the agents perform on the customer's transactions is the product's output, not a record of the agents. A command center dashboard with a transparent audit trail on every claim appeared on the previous site and is not in the current navigation. There is no log schema, retention period, export path or SIEM integration, and the trust center lists the capability by name without describing it.

Sourcetrust.marvelx.ai product security and continuous monitoring sections, marvelx.ai home solution sectionread 2026-09-14

Built-in audit mechanisms keep a clear trail of underwriting decisions with guideline adherence, the premium audit solution tracks adherence to state regulations, and the document generation tool keeps every generated document for regulatory review.

The Business Research Agent shows the supporting evidence behind every recommendation, and the Guideline-to-Scorecard agent traces every line back to its source sentence, which is decision rationale, not just reporting. There is no runtime observability for agents and no trace, replay or per-action log, and in the private cloud deployment monitoring and logging are under the customer's own control. Decision documentation is deep, but there is no runtime audit of what an agent did.

SourceWeav.ai, weav.ai homepage, /platform and /agentic-toolsread 2026-09-07

Customers each get a private, isolated cloud environment of their own, but there are no regions, no option to run in the customer's own account and no way to choose. MarvelX guarantees private deployment sovereignty, deploying in a private, isolated cloud environment dedicated to each organization. MarvelX runs the environment, which is dedicated to the customer, not owned by them. A dedicated tenancy is a real step above shared multi tenant SaaS. No region is named, there is no option to deploy into the customer's own account or data center, and there is no residency menu.

The trust center says only that MarvelX uses providers like AWS, GCP and Azure, which describes its supply chain, not a customer choice. A published customer account says MarvelX let it develop and deploy an AI claims agent, customized to its specific products, trained on its own data and running in its own environment within weeks. A customer's phrase in a testimonial is not a deployment option, and it sits beside MarvelX's own sentence describing the opposite shape.

Sourcemarvelx.ai home security section and customer account, trust.marvelx.ai operational security sectionread 2026-09-14

The platform deploys in the customer's private cloud with full control over security, data, monitoring and logging, and supports all private cloud deployments including on-premises. It runs on Google Cloud, Azure, AWS, on-prem and SaaS. Deployment into the customer's own environment is the lead offer. There is no list of named regions, so the choice is an environment, not a region.

SourceWeav.ai, weav.ai/platformread 2026-09-07

Five named agent solutions each have their own page, but on a single workflow product several look like stages of one pipeline, not agents a customer adopts separately. Each has its own durable URL, alongside a platform level AI Agents page. The product has also been described as shipping prebuilt insurance agents for claims assessment, fraud detection and customer communication, plus reusable templates such as compliance report templates, not a broad catalog for users to compose.

Nothing settles whether those five are agents a customer selects among, or five descriptions of what one claims platform does. On this product the second reading is live. Data Extraction, Claims Processing and Report Generation are consecutive stages of the same pipeline, not alternatives, and removing one would break the others. Customer Support and Fraud Detection are the two that look separable. Either way, these are named, prebuilt, domain specific agents shipped as product.

Sourcemarvelx.ai solutions navigation read across two pages, home solution pipelineread 2026-09-14

A browsable catalog holds eleven named standalone agents.

They are SoV Analysis, Loss-run Analysis, Business Research Agent, Payroll Agent, Contextual Document Summarization, Agentic Document Generation, Vehicle Schedule Analysis, Employee Census Agent, FNOL Processor, Insured Portal Builder and Guideline-to-Scorecard Agent, each signed up for individually, each doing its own whole job and priced per unit in shared tokens.

Beneath them the platform ships pre-configured libraries of knowledge graphs and scorecards by industry and line of business. The suite opens through a sign up form that grants 100 free tokens, and the FNOL Processor and Insured Portal Builder require a consultation first.

SourceWeav.ai, weav.ai/agentic-tools and /platformread 2026-09-07

MarvelX runs its own domain-specific models and names no provider or customer choice of model. It uses its own models trained on insurance data with reinforcement learning and generative AI. The older security page sells domain-specific models with lightning-fast accuracy, customized to the unique needs of each business, and the rebuilt home page captions its expert gate Your rules and not a model.

A product positioned on proprietary vertical models and customer-authored rules has no commercial reason to expose a model selector, and does not. No provider is named, and there is no selector, disclosed routing or bring your own key path. What the customer does control is the training data and the rules, not the engine.

A published customer account describes an agent customized to its specific products and trained on its own data, and step five drills it in the Agent Gym. Training a vendor's model on the customer's data is not selecting which model runs. AWS, GCP and Azure on the trust center are infrastructure providers, not a second model. MarvelX makes no negative statement either. The site is silent, and the positioning points away from model choice.

Sourcemarvelx.ai home solution pipeline step 04 and customer account, marvelx.ai/security domain-specific models tile, trust.marvelx.ai AI and operational security sectionsread 2026-09-14

The platform is model-agnostic and orchestrates agentic workflows across multiple models, balancing quality, responsiveness and costs. It supports commercial models from OpenAI (GPT family) and Anthropic (Claude family) and open source models from Meta (Llama family), Deepseek and Mistral, with Gemini also listed and vendor-assisted fine-tuning for specialty lines. Weav.ai does the routing across these named providers. There is no customer or admin selection surface, and private cloud deployment makes customer choice plausible, but Weav.ai does not describe one.

SourceWeav.ai, weav.ai/platformread 2026-09-07

Integrations is a governed platform component, but nothing shows the platform being callable from outside. It is part of the platform, and the trust center lists Integrations under Product Security, so it is governed as part of the product. A named, governed integrations component plus a two-way flow with the customer's systems is a real extension surface.

The product has been described elsewhere as integrating through robust APIs and webhook support, but no current page names an API, an SDK or a webhook, so that claim is unconfirmed. What is missing is an API or SDK through which an external system or another vendor's agent drives MarvelX. There is no developer entry, documentation host, SDK, MCP server or published reference. The direction that exists runs the other way, MarvelX reaching into the insurer's systems to collect documents and deliver outputs.

Sourcemarvelx.ai platform navigation read across two pages, trust.marvelx.ai product security sectionread 2026-09-14

Outside systems can call in, but the details are not published. The Weav.ai MCP server lets other agentic systems communicate with Weav.ai agents to orchestrate agentic workflows, but no endpoint, auth scheme or reference documentation is published. Every page calls the architecture API-first, yet there is no developer portal, API reference or docs subdomain. The MCP clients to Salesforce and ServiceNow and the Rapid Insurance Connector framework reach out to other systems and are not a way in.

SourceWeav.ai, weav.ai/platformread 2026-09-07

A changed agent is drilled in a named harness, the Agent Gym, before it goes live. Agent Gym is listed first among the platform components, with a durable URL. In step five of the pipeline the AI improves itself, with the agent drilled in the Agent Gym before it goes live. A training and testing stage standing between a changed agent and production is a pre-deployment harness in the strict sense, and it puts a change under test. The governance side supports it. The trust center includes an AI system development and evaluation policy among its policies and lists AIUC-1, an AI-specific certification, among its compliance items.

What is drilled is MarvelX's own claims agent on this customer's products and data, customized to the customer's specific products and trained on its own data, per the published customer account. Customer subject matter experts also take part in training and refining agents, and assessment accuracy is measured. What is known of the harness is its name and the promise of testing before it goes live. There are no scored cases, golden set or published accuracy measure.

Sourcemarvelx.ai home solution pipeline step 05 and platform navigation, trust.marvelx.ai policies and compliance sectionsread 2026-09-14

Ingestion evaluation AI agents evaluate the extraction and surface conflicts and low confidence data on every case in production, so quality is measured live and always on. Nothing puts a change under test. There is no evaluation harness, scored test cases, release gate or controlled experiment after deployment. The scorecards evaluate the insurance risk, the customer's artifact, not the agent, so they are the product and not an evaluation tool.

SourceWeav.ai, weav.ai/platformread 2026-09-07

There is no screen in the loop. The agents work on documents and write results back into the insurer's system through integrations and APIs. There is no hosted or local browser, desktop session or remote computer control. The action path is documents in and a write into the customer's system out, from upload or collection from the customer's systems at one end to delivery into the customer's system at the other.

Document AI is MarvelX's headline capability and can read like visual automation, since its agents handle everything from extracting data in messy handwritten documents to drafting settlement reports, and the pipeline reads policies, photos, invoices and handwriting, scoring every field. Reading a photograph of a damaged vehicle or a handwritten claim form is perception applied to a file, not an agent operating an interface.

Breakage when UI elements change has no meaning against PDFs, photographs and a system write back. MarvelX's failure mode is a document it cannot parse, not a layout it cannot find. Insurance claims platforms are the kind of legacy systems other products drive by screen automation, and MarvelX connects to them through an Integrations component instead. MarvelX describes the architecture in enough detail that an undocumented screen-driving path is implausible.

Sourcemarvelx.ai home hero and solution pipeline, platform and solutions navigationread 2026-09-14

The Business Research Agent researches the business across public web sources at case time, builds a profile of its actual operations and recommends classification codes, with supporting evidence behind every recommendation. That is fetching web sources at run time as part of the work, by an unstated mechanism, and it feeds the live classification decision, not the knowledge graph library. There is no hosted or local browser, desktop session or computer control, and the tool opens through the suite's public sign up form.

SourceWeav.ai, weav.ai/agentic-toolsread 2026-09-07