PBC Vision
PBC Vision is a pre-trained AI agent for anti-money-laundering teams in banks and financial institutions. It monitors transactions, customer behaviour and external sources in real time, scoring risk dynamically and documenting the reasoning behind every single alert it raises.
What is PBC Vision?
PBC Vision is an artificial-intelligence agent dedicated to anti-money-laundering work, prevención del blanqueo de capitales in Spanish, hence the initials. It is one of the seven pre-trained agents in the Sensei Vision catalogue published by the Spanish company TXSTOCKDATA, S.L., based in Pontevedra with technical offices at the Tecnópole park in Ourense. Sensei Vision presents itself as part of Salvum Technologies, a group it describes as deployed in banks across Europe and Latin America. This record covers PBC Vision alone, not the rest of the catalogue.
The agent watches operations, customer behaviour and external sources in real time in order to catch laundering patterns before they escalate. The product page sets out three pillars. The first is dynamic risk scoring: transaction and behavioural data is cross-referenced with external sources and the score is recalculated every time relevant new information arrives. The second is explainability: each alert carries the weighting of the factors that determined it, so that, in the vendor's words, the analyst sees why an alert was generated and not merely that it was. The third is a complete alert cycle, from detection to escalation, follow-up and documented closure, with compliance reporting files produced automatically in the structure supervisors require.
The published flow has three stages: configuring the monitoring (risk typologies, alert thresholds, portfolio segments under enhanced scrutiny), monitoring AML alerts in real time with a risk level attached to each transaction, and generating the regulatory reporting files. PBC Vision can sit as a layer of intelligence on top of an existing monitoring system or act as the primary monitoring system, and Sensei's technical team assesses which model suits each institution. The baseline configuration covers the frameworks applicable in the European Union, AMLD and FATF, plus the international standards relevant to Latin American entities, and the vendor says it can be extended to specific jurisdictional rules. Every run is logged, with sources processed, signals detected, factors behind each alert and actions taken, and that history is offered to internal audit and supervisory bodies.
The vendor claims an 85% reduction in false positives and 63% adoption among active users; neither figure comes with a methodology, a date or a named customer. There is no published price, no trial and no self-service sign-up: the only way in is a demonstration request.
What it does
- Monitors transactions and customer behaviour in real time
- Cross-references internal transaction data with external sources
- Recalculates a dynamic risk score whenever relevant new information arrives
- Flags money-laundering typologies before they escalate
- Shows the weighting of the factors that triggered each alert
- Runs the full alert cycle: detection, escalation, follow-up and documented closure
- Generates regulatory reporting files automatically and assists analysts in handling cases
When to use PBC Vision / When not to
A quick filter to help you decide if PBC Vision is the right fit.
When to use PBC Vision
- Banks and financial institutions that already run an AML monitoring system and are overwhelmed by false positives
- Compliance teams that must justify every alert to a supervisor and want to see the weighting of the factors behind it
- Entities subject to the EU AMLD and FATF frameworks, or to the international standards relevant to Latin American institutions
- Institutions whose data cannot leave their own systems: the vendor states that data never leaves the entity's corporate environment, under a multi-tenant architecture with strict per-organisation access control
- Compliance departments looking for a modular rollout, since PBC Vision can be activated on its own, and comfortable working in Spanish, the only language of the site and of the documented interface
When not to use PBC Vision
- Individuals, freelancers and small businesses: the product is sold to financial institutions only
- Organisations outside the financial sector, since the models are pre-trained on banking terminology and context
- Buyers who want to sign up on their own: there is no self-service registration, no free trial, no free plan and no published price
- Teams that need public API documentation, a published security certification (SOC 2, ISO 27001, ENS) or a published data processing agreement before engaging, as none of these exists on the site
- Teams looking for something other than anti-money-laundering work: document verification, ESG scoring, credit-risk analysis, competitive benchmarking and cyber monitoring belong to other agents in the same catalogue, not to PBC Vision
How to use PBC Vision
A typical end-to-end flow, from setup to results.
- Start from the PBC Vision product page on senseivision.ai: there is no online sign-up, so the single entry point is a demonstration request
- Fill in the contact form, which can be opened with PBC Vision already selected, giving your name, company or institution, corporate email, an optional phone number, the products of interest and a message
- Expect a reply within 24 business hours; the vendor announces a demonstration built on your own use case, with no commitment and no generic presentation
- Agree with the Sensei technical team on the integration model: PBC Vision running alongside your existing AML monitoring system, or acting as the primary monitoring system
- Connect the platform, whose APIs plug into the institution's core systems, which the vendor says requires no migration and no interruption of ongoing operations
- Configure the institution's parameters rather than the model: risk thresholds, internal policies, portfolio segments and priority data sources, with no model retraining involved
- Set the monitoring rules specific to PBC Vision, namely risk typologies, alert thresholds and the portfolio segments placed under enhanced scrutiny
- Let analysts work the alerts, using the weighting of the factors that triggered each one to escalate, follow up and close cases with documentation
- Collect results through the channels already in use: internal dashboards, email or integration with your reporting tools
- Adjust thresholds, typologies and segments over time from the platform's management interface, follow the metrics dashboards, and go back to the Sensei team for any new system integration
Pros & Cons
Pros
- Explainability claimed down to the weighting of the factors behind each alert, which few AML tools expose
- End-to-end traceability designed for internal audit and supervisors: every run is logged and the history is said to be available at any time
- Coexists with the AML monitoring system already in place instead of replacing it, which limits the risk of a switch-over
- Models pre-trained on banking terminology and context, so no retraining project is needed before going live
- Modular catalogue: PBC Vision can be activated on its own, without deploying the rest of the range
- Baseline coverage of the AMLD and FATF frameworks, extendable to other jurisdictions, and a vendor statement that data never leaves the institution's corporate environment
- European publisher, controller established in Spain with the AEPD as supervisory authority, complete legal documentation (LSSI-CE legal notice, GDPR privacy policy, cookie policy, whistleblowing channel), and website analytics cookies processed on EU servers with no transfer outside the EEA
Cons
- No public pricing at all: no pricing page, no range and no billing unit, verified against the site's complete 15-URL sitemap
- No free trial, no free plan and no self-service registration; the only entry point is a demonstration request
- Performance figures (85% fewer false positives, 63% adoption) come with no methodology, no date and no named customer, and the testimonials give a first name, a surname and a title but never the institution
- No published security certification, neither SOC 2 nor ISO 27001 nor ENS, and no data processing agreement published or offered, on a product aimed at banks
- No public API documentation, although integration is presented as happening through REST APIs and webhooks
- Privacy policy and legal notice dated 16 February 2024 and limited to the website: nothing covers the processing of banking data by the agents, and no hosting country or subprocessor list is published for the platform
- Site and documentation available in Spanish only, with no language selector; two footer links, /agentes and /sensei-ai, return a 404 and the HTML carries a "SenseiAI - próximamente" comment
Pricing & Plans
No price is published. The site carries no pricing page, which was verified against its complete sitemap of 15 URLs, and no amount, tier, currency or billing unit appears anywhere. No free trial and no free plan are announced. PBC Vision is sold through contact with the vendor: a demonstration first, then a quotation per institution. The contact form carries a "Productos de interés" selector in which PBC Vision appears alongside the other agents, which suggests commercial packaging per agent, and the catalogue is modular, each agent being activatable independently according to the institution's priorities. Access to the website itself is free of charge; the terms state that the use of some content or services may require prior registration or subscription. Any figure has to be obtained by requesting a demonstration.
Data, GDPR & hosting
A consolidated view of how PBC Vision handles your data.
GDPR overview
Implementation is documented and explicit, for the website. The privacy policy declares alignment with GDPR (EU) 2016/679, Spanish LOPD-GDD 3/2018, RD 1720/2007 and LSSI-CE 34/2002, and sets out the seven principles of Article 5. Seven rights are listed: access, rectification, erasure, restriction, portability, objection and not being subject to automated decision-making. They are exercised in writing with the reference "RGPD-senseivision.ai", full name, a copy of the DNI, grounds, a notification address, date and signature, sent to Calle Río Umia 13, 36005 Pontevedra or contacta@senseivision.ai. The supervisory authority is the Agencia Española de Protección de Datos. Minimum consent age is 14. An SSL certificate is in place and breaches are notified without undue delay where risk is high. No DPO is named, no Article 27 representative is required as the controller is established in Spain, and no data processing agreement is published or offered.
Who owns the data?
The controller named on the site is TXSTOCKDATA, S.L. (NIF B94131489, Registro Mercantil de Pontevedra, Tomo 4107, folio 120, hoja PO-59889), Calle Río Umia 13, 36005 Pontevedra, +34 604 086 860, contacta@senseivision.ai. Its privacy policy, dated 16 February 2024, covers only the identifying data collected through the website's forms, on a consent basis, and states that user personal data is not shared with third parties. It says nothing about who owns the banking data the agents process. For the platform, the vendor states only that data never leaves the institution's corporate environment and that a multi-tenant architecture with strict per-organisation access control prevents data from being shared between organisations.
Reuse rights
Two perimeters have to be kept apart. For website data, section 7 of the privacy policy lists the purposes: facilitating and performing the commitments between the site and the user, maintaining the relationship created by the forms and answering requests or enquiries, as well as commercial personalisation, operational and statistical purposes, activities falling within the company's corporate purpose, data extraction and storage, marketing studies and improvement of the site. Cookies split three ways: technical ones on legitimate interest, PostHog analytics on consent and preference cookies on consent. On the agent side, the vendor states that the models are pre-trained on banking terminology and context, so customisation applies to the institution's own parameters, such as risk thresholds, internal policies and portfolio segments, and not to the model itself. Every agent run is logged with the sources processed, the signals detected, the factors behind each alert and the actions taken, and that history is made available to internal audit and to supervisory bodies. No written commitment published on the site covers whether banking data may be used to train models.
Data retention & training
Hosting summary
The only explicit hosting statement on the site concerns website analytics: PostHog analytics cookies are processed on servers located in the European Union, and no data is transferred to countries outside the European Economic Area. That statement covers the website, not the banking data the agents process. For the platform itself, the vendor states that data never leaves the institution's corporate environment, that its APIs connect to the institution's own systems, and that a multi-tenant architecture with strict access control per organisation prevents data from being shared between organisations, with activity logged and auditable. Beyond those statements nothing is documented: no hosting country, no cloud provider and no data centre is named, and no subprocessor list in the sense of a data processing agreement is published. One infrastructure fact, independent of the vendor's declarations: the public website is served through Fastly's anycast network and the resolved IP address geolocates to the United States, which reflects a CDN node for the marketing site and says nothing about where banking data is held. Residency commitments have to be requested in writing.
Things to keep in mind
Risks and trade-offs to weigh before adopting PBC Vision.
- The 85% reduction in false positives and the 63% adoption rate are vendor claims, published as animated counters with no methodology, no date and no named customer. A compliance team that adopts a score without challenging it would be relying on an unverified figure, so keep human judgement above the model's output
- The testimonials name a person and a role but never the institution ("European regional bank", "credit institution", "corporate banking"), so they cannot be traced back to a reference customer you could call
- The interface shown on the product page is an illustration built on obviously fictional entities. The AML typologies, the OFAC, EU, UN and SEPBLAC sanctions lists, the PEP detection and the SAR case file it displays are what the vendor's mock-up shows, not a documented or certified integration list
- The privacy policy and the legal notice are dated 16 February 2024 and cover the website only. Nothing published describes how banking data is processed, stored or retained by the agents, nor how long execution logs and regulatory reporting files are kept
- Two points to raise with the vendor: the privacy policy states that personal data is not shared with third parties, while the cookie policy updated on 9 March 2026 names PostHog as its analytics provider; and that same cookie policy points to an address on a .com domain distinct from the site, which serves only an empty page
- No security certification (SOC 2, ISO 27001, ENS) and no data processing agreement is published, and there is no public API documentation to size the integration effort in advance, so everything has to be obtained from the vendor before signing
- Checking the counterparty takes extra care: the company's incorporation date is not published on the site and public sources disagree, two postal addresses coexist (Calle Río Umia in Pontevedra in the legal documents, Tecnópole in Ourense on the contact page), and the Sensei brand prefix is also carried by the publisher's earlier products, which have nothing to do with anti-money-laundering, so confirm which legal entity you are contracting with
Setup & Integrations
Technical difficulty
Demanding, and never self-service: there is no online sign-up and the Sensei team runs the deployment. The platform connects to core systems through SaaS or API, which the vendor says needs no migration or interruption, the duration depending on the number of agents and the core systems involved. No model retraining is required, which shortens go-live. Configuration covers risk thresholds, internal policies, portfolio segments and priority data sources, then PBC Vision's typologies, alert thresholds and enhanced-scrutiny segments. Expect to mobilise compliance and IT, with access to core and monitoring systems. No public API documentation exists to size the work beforehand.
Deployment
Behind PBC Vision
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
Does PBC Vision replace an existing AML monitoring system?
How is the traceability required by supervisors ensured?
What control does the institution keep over typologies and thresholds?
Which AML frameworks does the baseline configuration cover?
Do the models have to be retrained on the institution's internal data?
How long does integration take?
Can PBC Vision be deployed on its own?
How does the vendor say data stays inside the institution's environment?
Is there a public price or a free trial?
What retention period and minimum age apply to the website's data?
Should you pick PBC Vision?
PBC Vision is a narrow, vertical product: one AML agent for financial institutions, not a general-purpose tool. Its two claimed differentiators are worth a meeting if they match your problem: explainability down to the weighting of the factors behind each alert, and end-to-end traceability built for internal audit and supervisors. It is designed to slot into an existing arrangement rather than replace it, which lowers the stakes of a first deployment: a bank drowning in false positives can run it over the monitoring system it already has.
The publisher is a small Spanish company whose legal documentation is serious but limited to the website. The commercial opacity is deliberate: no price, no trial, no sign-up, and a demonstration as the only door. The evidence on offer is thin, with metrics that carry no methodology, customers who are never named and no published certification.
So go in with a list. Ask for a demonstration on your own data rather than the illustrated interface. Ask which sanctions lists and typologies are actually connected, and how. Ask for written commitments on where banking data sits, how long execution logs and reporting files are kept, and who may access them, because none of that is published today. Ask for a data processing agreement and for whatever security certification exists. Ask for the pricing structure per agent and for a reference customer you can call.
A few signs point to a young site: two footer links return a 404 and one product is announced as coming soon. That is not disqualifying for a B2B product sold by demonstration, but it argues for checking in the meeting what the site does not document. For a compliance team with a concrete false-positive problem and the patience for a procurement conversation, PBC Vision is worth the call.
- Choosing a selection results in a full page refresh.
- Opens in a new window.