AliceOS
AliceOS is Agentifai's enterprise voice AI platform for banking and healthcare. Its seven-layer agentic architecture lets teams build, orchestrate and govern AI agents that authenticate callers and complete real transactions inside tightly regulated contact centres.
What is AliceOS?
AliceOS is the enterprise platform built by Agentifai, a Portuguese company based in Braga, to create and run AI voice agents inside banking and healthcare organisations. Rather than a single conversational model, it is presented as an orchestration layer: one secure place where a company integrates and governs its AI agents, its foundation models and its data sources together.
The platform rests on a seven-layer agentic architecture. A Creator Agent turns a spoken or written objective, or an uploaded rulebook, into a working agent, defining its workflows, persona and toolset. Domain Agents carry the regulatory and operational knowledge of financial and clinical work. Knowledge and Tool Agents connect to internal knowledge bases, CRMs and enterprise APIs so the system acts rather than merely answers. An Orchestrator routes each request to the right specialist and keeps the conversational context intact; Agentifai says it already manages more than a hundred agents in production. Workflow and Micro Agents split long processes into narrow, low-error steps. A Guardrails layer watches every response and API call against the safety, ethical and compliance limits the organisation has set. A Supervisor Agent closes the loop with telemetry on performance, interaction quality and latency.
Around that core sits a full contact centre: native computer telephony integration, queues, skills, subject-based routing, transfers, escalation and team management, so that AI agents and human operators work in one environment. Agent Assist prompts operators live, while automated quality assurance scores conversations against configurable criteria.
The catalogue is deliberately vertical. More than 120 banking blueprints cover balance lookups, card activation, IBAN and NIB transfers, photo-to-payment and credit simulation; over 100 clinical blueprints cover triage, appointments, prescriptions and follow-up campaigns. Step-up multi-factor authentication runs inside the conversation itself.
Agentifai names Santander, Banco BPI, Caixa Geral de Depósitos, Vivalto Santé, Lusíadas Saúde, Ribera Salud, Luz Saúde and Viamed among its customers, and claims more than a billion interactions processed, over 84% of transactional flows automated and over 82% authentication success. Those figures are self-reported.
What it does
- Handle inbound and outbound customer calls with voice agents that grasp intent and carry the request through
- Authenticate callers with step-up multi-factor verification inside the conversation, before any sensitive record opens
- Execute banking operations: balance lookups, card activation, IBAN and NIB transfers, photo-to-payment, credit simulation
- Book, reschedule, confirm or cancel medical appointments and handle prescription refills and pharmacy routing
- Triage inbound clinical queries and run proactive follow-up campaigns for surgery, screening and chronic care
- Orchestrate several specialised agents on one conversation and hand over to a human operator with full context
- Score conversations automatically against configurable quality, script and compliance criteria
When to use AliceOS / When not to
A quick filter to help you decide if AliceOS is the right fit.
When to use AliceOS
- Retail and commercial banks automating authentication, card, payment and servicing journeys end to end
- Hospital groups and clinic networks facing high volumes of appointment, triage and follow-up calls
- Contact centre leaders who want AI agents and human operators managed from a single platform
- IT and AI teams in regulated organisations that need governance, audit trails and guardrails over every interaction
- Compliance and risk officers who require documented certifications, an Article 28 DPA and contractual data residency
When not to use AliceOS
- Individuals, freelancers and small businesses: there is no self-service sign-up and no published price
- Teams outside banking, healthcare and regulated customer service, since the domain logic is deliberately vertical
- Buyers who need to compare costs before speaking to sales: no pricing page, no plans and no trial are published
- Developers hoping for public API documentation or a sandbox to prototype against
- Organisations expecting the AI to take binding clinical, credit or compliance decisions on its own
How to use AliceOS
A typical end-to-end flow, from setup to results.
- Send the 'Talk to us' form on the website with your name, company and email, or use its 'Request a Demo' button
- Scope the deployment with Agentifai: verticals, channels, modules to enable and the cloud region, either the EEA or the United States
- Sign the commercial framework, which covers the Master Services Agreement, the service levels and the Article 28 Data Processing Agreement
- Connect AliceOS to your systems through configurable APIs: scheduling, hospital information systems, core banking, CRM or custom HTTP endpoints
- Choose the foundation models you want the agents to run on, among OpenAI, Anthropic, Gemini, Grok and others
- Start from the pre-built banking or clinical blueprints instead of designing every flow from scratch
- Let business teams create and iterate agents in the Agent Creator by uploading rulebooks, workflows or documents in plain language
- Set the guardrails, the compliance boundaries and the financial thresholds that pause an agent for human approval
- Configure teams, roles, permission profiles, skill groups and routing rules in the settings module, and enable SMS two-factor or enterprise SSO
- Run and refine in production using the operations dashboard, Supervisor Agent telemetry and automated quality assurance
Pros & Cons
Pros
- Vertical by design: banking and clinical logic ships pre-built instead of being assembled on a generic platform
- Named, checkable production references including major banks and European hospital groups
- Unusually dense public trust centre, with SOC 2 Type 2, ISO/IEC 27001, ISO/IEC 42001:2023, HIPAA and PCI DSS v4.0.0
- An Article 28 Data Processing Agreement, a Business Associate Agreement for US healthcare and contractual data residency in the EEA or the United States
- Model neutrality: the customer picks the foundation model rather than being locked to the vendor's own
- Native CTI removes the need to stitch telephony, AI automation and workforce management together
- Pricing is announced as pay-for-success, with nothing charged when a conversation is handed to a human agent
Cons
- No price is published anywhere: no pricing page, no range, no order of magnitude
- No free plan, no trial and no self-service sign-up; the only entry point is a sales form
- No public terms of service either: the sitemap carries only the privacy and cookie policies, and contractual documents are handed over during the sales cycle
- No public API documentation, even though the platform integrates through APIs and token-based authentication
- The named list of AliceOS sub-processors, like the SOC 2, ISO and penetration test reports, sits behind an access request
- Multilingual and polyglot speech are claimed, yet not a single supported language is named
- Performance figures are self-reported and the customers behind them are anonymised as 'Bank A' or 'Healthcare Provider C'
Pricing & Plans
Agentifai publishes no price for AliceOS. There is no pricing page, no plan, no free plan and no free trial anywhere on the site. The company describes a pay-for-success model billed per automated conversation, with nothing charged when a conversation is handed over to a human agent. Because no amount, currency or volume threshold is disclosed, the cost of a deployment can only be obtained from the sales team, and commercial terms are set in the Master Services Agreement.
Data, GDPR & hosting
A consolidated view of how AliceOS handles your data.
GDPR overview
GDPR implementation is documented rather than merely asserted. Agentifai is established in Braga, Portugal, so no Article 27 representative is required and none is named. A Data Protection Officer is designated and reachable at dpo@agentifai.com. The privacy policy cites its legal bases article by article, from Article 6(1)(f) for running the site to 6(1)(a) for forms, 6(1)(b) for pre-contractual exchanges and 6(1)(c) for legal obligations, and names the Portuguese supervisory authority CNPD. Rights of access, rectification, erasure, restriction, portability, objection and withdrawal of consent are all listed. For the platform, Agentifai offers a GDPR-compliant Article 28 Data Processing Agreement and, for US healthcare, a Business Associate Agreement. Its trust centre displays SOC 2 Type 2, ISO/IEC 27001, ISO/IEC 42001:2023, HIPAA and PCI DSS v4.0.0.
Who owns the data?
On the AliceOS platform Agentifai acts only as a processor: the customer, whether a bank, a hospital group or another enterprise, remains the data controller and keeps ownership of the personal data. Agentifai processes it solely on that customer's documented instructions, under a GDPR Article 28 Data Processing Agreement, and passes it on only to the customer's own systems, to approved sub-processors and to the customer's staff through the operations dashboard. The roles reverse on the corporate website: there Agentifai is itself the controller for data collected through its contact and recruitment forms. The published privacy policy, dated 12 March 2023, covers that website alone.
Reuse rights
Nothing on the site grants a visitor any licence to reuse Agentifai's own content, and for the platform the question is reversed: it is the customer, as controller, who decides what becomes of the conversation data. AliceOS processes personal data only as instructed, and Agentifai states that customer production data is not used to train foundation models unless explicitly agreed in writing with a documented legal basis. Product improvement is said to rely on controlled development, testing and release processes rather than on customer content. Data leaves the platform only towards the customer's own systems, such as appointment APIs, ticketing, electronic health records or core banking integrations, towards authorised sub-processors, and towards the customer's own personnel. On the website itself, form data is used only to answer the enquiry or to assess the job application it was submitted for.
Data retention & training
Hosting summary
AliceOS production infrastructure runs mainly on Google Cloud Platform, on Google Kubernetes Engine, in EU or United States regions chosen per customer deployment. Agentifai states that customer production data is processed and stored in cloud regions defined in the customer agreement, typically within the European Economic Area or the United States, and that it does not move that data across regions without contractual agreement and appropriate safeguards. The architecture is described as portable to other providers such as AWS or Microsoft Azure where a contract requires it, and no customer-facing production workload runs in an on-premise data centre. Data at rest is encrypted with AES-256 or equivalent across databases, object storage and backups, and traffic is protected with TLS 1.2+ and Google Cloud Armor. The corporate website is a separate matter: its named sub-processors are Cloudflare, Atlassian and Zoho, and the privacy policy states that no international transfers take place for website data.
Things to keep in mind
Risks and trade-offs to weigh before adopting AliceOS.
- The no-training promise is conditional: production data may feed foundation models if a written agreement says so, and that agreement is invisible from outside
- Voice automation at this scale reshapes contact centre jobs; the vendor frames it as a hybrid model, but the stated goal of cutting operating costs and headcount by more than half is explicit
- Callers may not always realise they are speaking to a machine about their money or their health, and it is the deploying organisation, not Agentifai, that must inform them
- Deterministic execution and zero hallucination are vendor claims, not audited results; over-trusting them in a credit or triage flow is a real clinical and financial risk
- Depending on the contract, data may be hosted in the United States, which deserves careful thought for European health or banking records
- Automated quality scoring of conversations also means continuous measurement of human operators, with the workplace consequences that follow
- Leaning on pre-built blueprints can erode internal knowledge of the processes themselves, leaving teams less able to handle the cases the agents escalate
Setup & Integrations
Technical difficulty
Deployment is a contractual project, not a sign-up. Expect a demo, a scoping phase, then integration work against core banking, electronic health records, scheduling systems or custom HTTP endpoints, with secure credential handling throughout. The effort is deliberately split: business teams build and iterate agents in plain language through the Agent Creator, while IT keeps control of architecture, security and governance. Pre-built blueprints, predefined roles and a settings module for teams, permissions and skill groups reduce the initial build, but this remains an enterprise implementation requiring internal IT involvement.
Deployment
Integrations
Behind AliceOS
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Alternatives
Tools that compete with or complement AliceOS.
Frequently asked questions
What is AliceOS?
Who is behind AliceOS and where are they based?
How much does AliceOS cost?
Is there a free trial or a free plan?
Will my data be used to train AI models?
Where is the data hosted?
Which certifications does Agentifai hold?
Is a Data Processing Agreement available?
Does AliceOS offer a public API?
Which language models can it run on?
Should you pick AliceOS?
AliceOS is a rare thing in the AI agent market: a platform that has chosen depth over reach. Agentifai does not try to serve every industry. It serves two, banking and healthcare, and it serves them with pre-built domain logic, embedded multi-factor authentication and a deterministic execution model designed for moments where a mistake is a regulatory incident rather than an annoyance. The seven-layer architecture, the native contact centre and the 120-plus banking blueprints all point the same way: the product exists to shorten the distance between a proof of concept and a compliant deployment.
The evidence behind that positioning is unusually concrete. Santander, Banco BPI, Caixa Geral de Depósitos, Vivalto Santé, Lusíadas Saúde and others are named, with case studies attached, and the trust centre publishes far more detail than most vendors: certifications, hosting regions, retention rules, an Article 28 agreement and an explicit position on model training.
The reservation is commercial transparency. There is no price, no plan, no trial, no public terms of service and no API documentation. A prospect cannot form even a rough idea of cost, or read the contractual terms, before entering a sales conversation. The performance figures are impressive but self-reported, with the customers behind them anonymised. And the promise that production data is not used to train foundation models carries an exception, activated by a written agreement, that any buyer should read closely.
For a regulated organisation with real call volume and a compliance department to satisfy, AliceOS deserves a place on the shortlist. For anyone who wants to try before they talk, it is simply not that kind of product.
- Choosing a selection results in a full page refresh.
- Opens in a new window.