Expert.ai Platform
Expert.ai Platform is an enterprise web workspace for designing and operating natural language solutions. It pairs EI-Authoring, for training explainable models, with EI-Flow, for visual document workflows published as REST APIs, aimed at regulated, document-heavy organisations.
What is Expert.ai Platform?
Expert.ai Platform is the web platform that Expert.ai S.p.A. documents on docs.expert.ai as "the easy to use Web platform to design and operate AI-based natural language solutions". It bundles two applications: the authoring app, EI-Authoring, and NL Flow, also documented as EI-Flow. EI-Authoring is a collaborative web workspace where data scientists, knowledge engineers and data analysts train, tune and deploy predictive models that categorise texts or extract information from them. The same workspace manages thesauri, assembles corpora reusable as training or test libraries, and lets teams explore the knowledge graphs available per language and derive customised versions from the factory release. EI-Flow, a component of the EidenAI Suite, is the multi-user side: a visual designer for building, deploying and running document analysis workflows. A published workflow is an executable pipeline that analyses a single document or a batch, synchronously or asynchronously, orchestrating one or several models with pre- and post-processing blocks, and every published workflow is exposed as a REST API. Ready-made components cover parallel and conditional branches, loops, scripting steps, text extraction with optional OCR and language detection, while an SDK covers custom components. Models can be symbolic, hybrid machine learning, LLM and document interpretation models supplied by the vendor, or external models reached through LiteLLM; workflows can themselves act as Model Context Protocol tools for other AI models. A catalogue of pre-built Knowledge Models covers areas such as PII anonymisation and pseudonymisation, contract analytics, ESG sentiment, corporate crime, hate speech, macroeconomics and life sciences. Compute capacity scales automatically with incoming traffic, and document similarity search works by indexing a set and then querying it. A companion Windows IDE, Studio, a rebrand of IntelliJ IDEA Community Edition, exchanges models and document libraries with the platform in both directions. The publisher is Expert.ai S.p.A., an Italian company listed on Euronext Growth Milan, with around 200 employees including 130 data and AI specialists, and more than 30 years of experience. Its stated approach, Hybrid AI, combines neuro-symbolic AI, knowledge graphs, machine and deep learning, LLMs, and generative and agentic AI, aimed at six industries: insurance, pharma and life sciences, digital information services, banking and financial services, manufacturing, and the public sector.
What it does
- Train, tune and deploy predictive language models that classify documents or extract information from them
- Design document analysis workflows visually and publish them as REST APIs
- Anonymise or pseudonymise personal data with the ready-made PII models
- Index a document set and retrieve documents similar to a given one
- Build and manage thesauri, then extract mentions of their concepts from text
- Explore language-specific knowledge graphs and derive customised versions from the factory release
- Expose published workflows as Model Context Protocol tools for other AI models
When to use Expert.ai Platform / When not to
A quick filter to help you decide if Expert.ai Platform is the right fit.
When to use Expert.ai Platform
- Insurance, banking, pharma and public-sector teams that must analyse large volumes of unstructured documents with a traceable decision path
- Data scientists and knowledge engineers who need explainable, hand-tunable language models rather than black-box classifiers
- Computational linguists and lexicographers working with thesauri, corpora and language-specific knowledge graphs
- Solutions architects and integration teams that want document workflows exposed as REST APIs and reusable as Model Context Protocol tools
- Organisations under sovereignty constraints, since the offering ships as public cloud SaaS, private cloud or an on-premises licence
When not to use Expert.ai Platform
- Freelancers, small businesses and low-budget projects: there is no free plan, no entry tier and no published price
- Anyone wanting to sign up and test immediately, since access goes through a demonstration request form and a sales conversation
- Users looking for a consumer conversational assistant; this is a modelling workbench for technical teams
- macOS and Linux workstations, because the companion Studio IDE requires Windows 10 or later on AMD64 hardware
- Teams needing a non-English interface, as the user interface and the whole documentation set are published in English only
How to use Expert.ai Platform
A typical end-to-end flow, from setup to results.
- Submit the demonstration request form on the vendor's site: there is no self-service sign-up, and the team calls back to schedule a session
- Open EI-Authoring, create a project and assemble a corpus of documents to serve as a training and test library
- Train and tune a categorisation or extraction model, test it, then deploy it
- In EI-Flow, create a workflow with the wizard or import an existing one
- Edit the block graph, add processors and flow operators, and define the input properties
- Publish the workflow in synchronous or asynchronous mode, then test it
- Create an API key, bind it to the workflow, and call the REST resource matching the publication mode; a dedicated resource returns the workflow status
- Manage runtimes: create a runtime, grant access, regenerate the connection token and export a runtime package when needed
- Administer user accounts, service accounts, groups, labels and custom component manifests
- Optionally install Studio on Windows and connect it to the platform; connectivity to artifactory.expert.ai is required to download knowledge graphs
Pros & Cons
Pros
- Explainability is the central argument: the symbolic and hybrid approach is presented as auditable, which carries weight in regulated sectors
- An unusually deep assurance base: ISO 9001:2015, ISO/IEC 27001:2022, ISO/IEC 27017 and ISO/IEC 27018, plus a SOC 2 Type II report covering the Platform product specifically
- Deployment freedom: public cloud SaaS, private cloud or an on-premises licence
- 23 ready-to-use business Knowledge Models, including PII anonymisation and pseudonymisation
- Open to external models through LiteLLM, and published workflows can be reused as Model Context Protocol tools
- Detailed public technical documentation, readable without any authentication
- An established, listed European publisher with around 200 employees and a designated Data Protection Officer
Cons
- No public pricing information whatsoever: no pricing page, no range and no tier; seven candidate pricing routes were tested and all return a 404
- No free trial and no free plan is documented; the only entry point is the demonstration request form
- The list of processed languages is not published: only French and Italian are named, and only through two sentiment models
- No data processing agreement is published or announced as available, and no named third-party sub-processor list exists, the privacy policy citing categories of recipients only
- No data hosting country or region is stated anywhere on the site
- No documented mechanism to exclude customer data from model training, and no published position on the subject
- Studio is Windows-only (10 or later, AMD64), and the Company Data page that stands for legal notices is three lines long, mixing the Italian corporate name with the American subsidiary's address
Pricing & Plans
No price is published. Neither a free plan nor a free trial is documented anywhere on the site, and no starting price, currency or billing unit could be established: no amount appears on any of the pages reviewed, the seven candidate pricing routes all return a 404 page of constant size, and none of the 344 URLs listed in the sitemap concerns pricing. The commercial model is contact-based: access runs through the demonstration request form and an enterprise contract negotiated with the vendor. The offering is available as SaaS on public cloud, as private cloud, or under an on-premises licence, which implies three distinct billing arrangements, but no figure is attached to any of them. For the record, the Terms of use state that the services offered on the website are voluntary and free of charge; that statement concerns the website, not the Platform product.
Data, GDPR & hosting
A consolidated view of how Expert.ai Platform handles your data.
GDPR overview
GDPR implementation is explicit. The privacy policy is drafted under Articles 13 and 14 of Regulation (EU) 2016/679 and entered into force on 05/07/2024, under European and Italian law, with CCPA/CPRA and United States state-law provisions added on top. A Data Protection Officer is designated and reachable in writing. Access, rectification, erasure, restriction, objection and portability (Article 20) are set out, along with the right to lodge a complaint with a supervisory authority. No Article 27 representative is named, and none is required: the publisher is established in the European Union. The quality and information security policy claims alignment with the GDPR, the AI Act, HIPAA, DORA, NIS2 and Italian Legislative Decree 231/01, backed by ISO 9001:2015, ISO/IEC 27001:2022, ISO/IEC 27017 and ISO/IEC 27018 certifications and a SOC 2 Type II report on the Platform product.
Who owns the data?
The data controller is Expert.ai S.p.A., with registered offices at Via Fortunato Zeni 8, 38068 Rovereto (Trento), Italy, VAT number 02608970360. Expert System Enterprise Corp. and the relevant group company act as joint controllers depending on the geographic area; the named group entities cover the United States, Spain, France, Germany, the United Kingdom and Canada. A group Data Protection Officer can be reached in writing at dpo@expert.ai. Scope matters: this privacy policy, in force since 05/07/2024, governs the group's websites, not the enterprise contracts under which the Platform processes customer content. That second regime is set by the commercial agreement, which the site does not publish.
Reuse rights
The privacy policy states the purposes as provision of the service, marketing and personalisation subject to consent, recruitment, legal compliance, and the prevention of abuse and fraud; a marketing soft opt-in applies and can be stopped through the link provided. Recipients are processors (individuals, companies and professional firms), other group companies, and public authorities on lawful request. Data may be transferred outside the European Economic Area, and the policy specifies that such transfers can be based on an adequacy decision or on the Standard Contractual Clauses approved by the European Commission. Nothing published states whether customer content submitted to the Platform may be reused to train models: the site takes no position in either direction, so the permitted reuse of data on both sides has to be fixed in the enterprise contract rather than read off the website.
Data retention & training
Hosting summary
No hosting country or region is stated anywhere on the site, and the corresponding fields are therefore left empty. What the documentation does say is that the offering is available as SaaS on public cloud, as private cloud, or under an on-premises licence, so the physical location of the data depends on the delivery mode chosen at contract level rather than on a vendor-wide default. The privacy policy is drafted under European and Italian law, the group being headquartered in Italy, and it states that personal data may be shared with recipients located outside the European Economic Area on the basis of an adequacy decision or of the Standard Contractual Clauses approved by the European Commission; that is a transfer clause, not a hosting declaration. Two certifications bear on the cloud perimeter: ISO/IEC 27017 for cloud security and ISO/IEC 27018 for the protection of personally identifiable information in the cloud. Buyers with residency requirements should treat hosting location as a contractual point to settle, since the public site does not answer it.
Things to keep in mind
Risks and trade-offs to weigh before adopting Expert.ai Platform.
- No public price means no way to budget or compare before engaging with sales, which weakens the buyer's position from the first conversation
- The privacy policy covers the group's websites, not the enterprise contracts under which the Platform processes your documents: the customer data regime is entirely contractual
- The site takes no published position on whether customer data is used to train models, so demand a written answer before sending sensitive content
- No named sub-processor list and no stated hosting country or region: data residency has to be pinned down in the agreement
- The Company Data page names Expert.ai S.p.A. but shows only the American address of Expert System Enterprise; the Italian registered office appears solely in the Terms of use and the privacy policy
- The Organization JSON-LD block published on every page carries an incorrect URL, https://www.expert.ai.com/, which suggests it has never been reviewed
- Studio is Windows-only, with no macOS or Linux build, and the company's incorporation date is not published on the site, which limits the background checks a buyer would normally run
Setup & Integrations
Technical difficulty
High, and it starts before any software: with no self-service sign-up, setup begins with a sales conversation, then a project. The authoring application explicitly targets data scientists, knowledge engineers and data analysts, and Studio targets data scientists, computational linguists and software engineers. Workflow design is visual and wizard-assisted, requiring no code, but custom components go through an SDK and API consumption means publishing a workflow, creating an API key and binding it. Administration adds runtimes, service accounts, groups and component manifests. Studio requires Windows 10 or later, AMD64, 8 GB of RAM (16 recommended) and connectivity to artifactory.expert.ai.
Deployment
Integrations
Supported languages
Behind Expert.ai Platform
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
What is the Expert.ai Platform?
Is there an API?
Can it be installed on our own infrastructure?
How much does it cost?
Is there a free trial?
Which certifications does the vendor hold?
Which languages are processed?
Can external LLMs be plugged in?
Who publishes the tool?
Is our data used to train the models?
Should you pick Expert.ai Platform?
Expert.ai Platform is an enterprise tool for technical teams, not a self-service product. Its strongest argument is the combination of explainability, security assurance and deployment freedom: symbolic and hybrid models are presented as auditable, the vendor holds ISO 9001:2015, ISO/IEC 27001:2022, ISO/IEC 27017 and ISO/IEC 27018 certifications plus a SOC 2 Type II report covering the Platform product specifically, and the offering runs as public cloud SaaS, private cloud or an on-premises licence. For a regulated organisation with heavy document volumes, that combination is rare, and the 23 pre-built Knowledge Models plus the openness to external models through LiteLLM shorten the distance between a pilot and production. The weakness is commercial opacity, and it is total. Nothing about price is published: no pricing page, no range, no tier, no free plan and no free trial. Evaluation therefore starts with a sales conversation, and budgeting beforehand is impossible. The publication gaps extend to the data regime: no hosting country or region is stated, no named sub-processor list exists, no data processing agreement is published, and the site takes no position on whether customer data feeds model training. The privacy policy, in force since 05/07/2024, covers the group's websites rather than the enterprise contracts under which the Platform would process customer content, so all of this has to be negotiated and obtained in writing. The publisher itself is solid: an Italian company listed on Euronext Growth Milan, around 200 employees including 130 data and AI specialists, more than 30 years of experience, and technical documentation that is genuinely detailed and open to all. Expect a project rather than a sign-up, and expect to ask in writing for everything the website does not say.
- Choosing a selection results in a full page refresh.
- Opens in a new window.