Tuesday, June 24th, 2025
Artificial Assistants in Italian Wealth Management market
The Wealth Management market is growing continuously, but the number of bankers and advisors has remained fairly stable. Artificial Assistants represent the immediate answer to increase productivity, empowering advisors and bankers in their daily activities and strategic decisions. Not a replacement, but an acceleration: concrete, rapid and scalable results, already successfully tested overseas and ready to revolutionize the European market.

Management Market
The Italian Wealth Management market is growing almost uninterruptedly. In the last five years alone, trust in Financial Advisory networks and Italian Private Banks has increased at a sustained pace. Total client assets managed by advisory networks reached € 909 (1) billion with a growth of + 8.0% (CAGR 19-24). This dynamic is also replicated by the evolution of Wealth managed by Italian Private Banking players, which reached € 1,257 billion (2) at the end of 2024 with a growth (CAGR) of +7.7% in the 19-24 period.
The industrial ingredient of this growth, as well as the scarcest, are people. In fact, if we look at the evolution of Financial Advisors, we realize that the number has remained almost stable over the years at around 23 thousand and the same goes for the number of Private Bankers (approx. 16 thousand). What has grown exponentially is their ability to attract clients’ assets. In fact, all productivity indicators have grown brilliantly so far (first and foremost the average portfolio of the advisor has increased by 43% reaching €39 million in 2024, approx. €80 million the average portfolio in Private banking, also thanks to the market effect).
However, finding the right people for this job is increasingly difficult and increasingly expensive, and for players ensuring their loyalty is equally difficult. For the next five years, if we conservatively assume a growth rate equal to half the historical one, the client assets managed by the Financial Advisor Networks would reach a value of €1,105 billion in 2029, while Private Banking would reach €1,517 billion. With the same headcount, it would therefore be necessary for advisors and bankers to increase their productivity by a further 20%.
To achieve this goal or to hypothesize an even more ambitious one in which the market grows at the same rate as in the past 5 years, one can imagine that Artificial Consultants or, more precisely, Artificial Assistants of the current Advisors will gradually be
inserted alongside consultants and bankers.
This is equivalent to hypothesizing:
• That Artificial Assistants ensure an increase in productivity of current consultants and bankers of at least 20 % (40% if we were to imagine a market growth equal to that of the last 5 years).
• Consequently, the size of the average portfolio, also considering the market effects, reaches just under 100 million Euros per Private Banker and just under 50 million Euros per Consultant at the end of the period.
The economic value of an initiative such as the one described – assuming that only half (10%) of the increase in productivity of advisors and bankers is due to the inclusion of Artificial Assistants and that the additional assets generate an average profitability (Net profit of 40 bps – amounts to at least 80-90 million in increased net profit per year for Wealth management players.
The strategic value goes beyond the economic one: acquiring the ability to integrate Artificial Assistants into your organization means increasing: your growth capacity, including the possibility of bringing new opportunities to customers more quickly; your cost competitiveness and consequently your resilience to shocks; your capacity for innovation and roll out of new opportunities that will gradually emerge from artificial intelligence applications.
(1) Exentially analysis on Assoreti data
(2) Exentially analysis on AIPB data
The Role of Artificial Assistants and Collaboration with Bankers
It is well known that Artificial Assistants and Artificial Agents are massively intruding into the virtual headcount of many organizations. Examples range from coding, marketing, legal professions and increasingly many professions based on cognitive and textual skills.
In the most advanced experiences, the presence of artificial people has allowed to streamline customer care management by 65% in the first month of activity alone, thus freeing up staff for cross-selling and up-selling activities on customers.
But even in traditionally more conservative industries like wealth management, AI has proven to be a powerful enabler of traditional advisory teams, particularly in tasks such as the following:
Activity AA contribution and interaction with the Consultant
- Presentation production to support the client / prospect meeting (pitching)
Preparation and management of client meetings
- Suggestion of arguments and points of contact for soft talks through cross-referencing public news and social media
- Generate meeting memos and takeaways
Research and analysis consultation
- Tailor-made research with specific economic scenarios on the reference sectors
- Automatic KIID search and consultation
Consultation of product documents
- Production of benchmarks on product performance
- Product Commissions Comparison vs Market
Production of customized portfolio analyses
- Data analysis and aggregation for producing reports on top portfolio stocks
- Performance attribution and decomposition: comparison of the client's portfolio with similar benchmarks / strategies
- Stress testing and backtesting to assess portfolio resilience in stressful situations
- Buy /sell recommendation on individual portfolio stocks based on sectors of interest
- Personalize recommendations based on feedback received in previous interactions
Production of customized proposals
- Integrated pricing tool based on historical exceptions and customer margins
- Verification and simulation of tax impacts/benefits
Contractual and compliance checks
- Quick access to preliminary AML/KYC customer checks, pending centralized analysis
- Automatically generate pre-filled forms and contracts based on information collected in the meeting
- Customized notifications about expiring customer documents or KYC update requests
- Automated document storage and retrieval with intelligent search
Portfolio Rebalancing
- Automated notifications on relevant market events that impact portfolio performance by suggesting changes in asset allocation
In none of the cases mentioned is the Advisor work “automated” and even less is the Advisor replaced. In each of these cases the Advisor’s work is rather “augmented” and enhanced by technology.
Analysis of a significant case of application of Artificial Assistants
Among the many ongoing cases of use of Artificial Assistants in Wealth Management, the most interesting case of “augmented” Advisory is that of LPL in the United States.
LPL Financial is a leading broker-dealer in the United States, managing more than $1.67 trillion in assets across 7.9 million client positions.
LPL's network includes more than 28,000 independent financial advisors. LPL Financial set out to launch its “AI Advisor Solutions” initiative with the expectation of helping advisors maximize their time with clients and grow their businesses at scale. As a network serving more than 5 million families through independent financial advisors, LPL found that many advisors were struggling to keep up with the growing demands for personalization and service.
According to a survey cited by LPL, nearly 1 in 3 advisors say they do not have enough time to dedicate to their clients. Therefore, LPL's vision was that AI could be the key to automating repetitive tasks (administrative tasks, document preparation, system updates) and allowing advisors to focus on high-value interactions. Unlike other companies that have developed their own chatbot in-house, LPL has chosen an approach of " curation " and integration of third-party solutions.
In 2023, it identified and tested three fintech providers considered to be at the forefront in specific AI application areas, including them in its Vendor Affinity Program. These three tools (along with Microsoft Copilot) were later announced as AI Advisor Solutions and include:
- an AI meeting management platform designed for financial advisors, which automates meeting preparation (agenda), note-taking, minute generation, and CRM updates. LPL piloted the meeting management application with selected advisors in 2023, finding that it dramatically reduces the workload generated by meetings (according to the vendor, it allows a meeting to be processed in 5 minutes instead of 60, cutting up to 90% of administrative time thanks to automation of notes, tasks, and summary emails). This tool has been integrated with the systems already used by LPL advisors (calendars, CRM, Zoom/Teams) and officially launched in 2024.
- a marketing platform for advisors, which LPL already had in use, enhanced with mobile generative AI. Through an app, advisors can create communication contentfor clients (newsletters, social posts, messages) with the help of AI that suggests texts and images in line with the desired tone -of-voice. This helps independent advisors, often without a dedicated marketing department, to remain engaging for clients without investing too much time.
- an advanced investment research and analysis tool that uses AI to sift through market data and studies, providing quick answers to complex financial queries and generating useful visualizations or insights for the advisor. It integrates a rich financial database (on stocks, funds, macroeconomics) with conversational chatbot-like features.
All three of these vendors were evaluated and piloted throughout 2023 (some in beta). LPL conducted the selection process based on advisor ease of use and security/compliance to ensure the solutions met regulatory standards (e.g., communications archiving) and integrated seamlessly into the advisors’ tech stack. Once validated, LPL has made agreements so that these tools will be available at favorable conditions to all consultants affiliated through its program (Vendor Affinity Program). The formal announcement took place in November 2024: LPL introduced AI Advisor Solutions as part of its suite of solutions for advisors and simultaneously launched training and support to drive adoption.
Results achieved
Since the first few months, there is evidence that many LPL consultants have started using at least some of the AI solutions. In particular, the meeting management solution has been well received: consultants who have adopted it report significant daily time savings (up to an hour per day) thanks to the automation of post-meeting notes and activities; a survey of its LPL users indicated that 84% prefer it over traditional methods for meeting preparation and follow-up, and that on average they save an hour of work for each day of use. On an annual basis and based on 23,000 consultants, this represents a huge impact in terms of aggregate productivity. On the marketing and research front, the solutions released by LPL have given consultants additional capabilities without increasing staff: an independent consultant can now launch a campaign on client groups in a few AI-driven clicks, or answer client questions by having AI quickly analyze key financial facts. An important qualitative finding is that “advisors are optimistic but cautious”: there is a willingness to use AI, but many are initially testing it on a small scale to verify reliability. LPL is therefore seeing gradual but increasing adoption as advisors gain confidence.
The program also has a marketing value in recruiting: showing that LPL offers its affiliates cutting-edge AI tools to help attract new advisors from the competition. In terms of results for end customers, it is early, but an improvement in the customer experience is expected: consultants can respond faster, with more curated materials (thanks to AI support in Office) and maintain more frequent and personalized contact (thanks to marketing automation suggestions).
One possible future indicator will be organic growth: LPL is confident that widespread adoption of these tools will translate into +20% business growth for its advisors in the coming years, in line with the beliefs expressed by 90% of advisors on the potential of AI.
As a final note, LPL stated that over time AI will not be limited to the back-office but will permeate nearly all applications in wealth management, creating increasingly personalized experiences and “elevating the value of the financial advisor” in the eyes of clients. Theencouraging initial results suggest that LPL will continue to expand and invest in the AI Advisor Solutions program as a pillar of its growth and service strategy.
The technical and operational feasibility of Artificial Assistants
The adoption of Artificial Assistants is no longer a question of “if", but of "when and how". The technologies that enable them – in particular generative AI models – have already demonstrated their maturity and reliability in regulated, complex and cognitively intensive contexts. The implementation of Artificial Assistants in the banking and insurance context is today technically feasible, data-secure, and operationally sustainable on an industrial scale.
Data Protection: AI and Compliance Can Coexist
One of the main elements of attention in the adoption of AI systems is the protection of personal and sensitive data, in line with the GDPR and sector regulations (such as EBA/ESMA guidelines or IVASS for insurance). In the financial and insurance sectors, where customer data represents highly sensitive and confidential information, ensuring regulatory compliance and data security is a top priority.
The nature of the data processed in these sectors - which includes financial information, risk profiles, health data for life insurance policies, and numerous personal identifiers - requires a particularly rigorous approach. The leading GenAI model providers today offer architectures that allow for the isolation of customer data, ensuring that no information is used to train the model itself or is accessible to third parties.
The implementation solutions available cover the entire spectrum of security needs:
- Dedicated and isolated cloud environments: Microsoft Azure, AWS, and Google Cloud offer private and dedicated instances of their best foundation models (such as GPT-4o, Claude 3.7, Llama 3.3, Gemini 2.0), with contractual guarantees of data segregation and compliance with European and Italian regulations.
- On-premises solutions: For organizations with extreme security requirements, it is possible to deploy open-weights models (such as Llama-3, Mistral, DeepSeek) on completely internal infrastructures, keeping sensitive data within the corporate perimeter.
- Hybrid Architectures: Combining the power of cloud models with the isolation of sensitive data through retrieval techniques augmented generation (RAG) and internal vector databases, allowing AIs to access proprietary documents without ever exposing the underlying data.
In particular, Artificial Assistants can be:
- Deployed in private or on-premises environments, if necessary, for full data segregation;
- Logging and audit trail tools to track every interaction and ensure accountability;
- Equipped with security controls for automatic classification of information, preventing the generation or processing of sensitive data outside of corporate policies.
Modern generative AI systems also incorporate advanced protection mechanisms:
- Jailbreak detection: Systems that detect and block attempts to manipulate AI through engineered prompts to extract data or bypass security protections.
- Content filtering: Platforms like Azure or AWS integrate automatic filters that prevent the generation of inappropriate content or the processing of potentially sensitive information. For example, Azure OpenAI includes a content filtering system that uses multi-class neural classification models to detect and filter harmful content across four main categories (hate, sexual content, violence, and self-harm) with different configurable severity levels. These settings can be customized separately for prompts and completions, allowing for granular control over the handling of sensitive data.
- Boundary control: Precise definition of the perimeter of knowledge and data that the Artificial Assistant can access, also depending on the user who is using it, limiting its ability to interact with critical internal systems. At the design stage, solutions are configured to have no active memory or to have data retention regulated according to internal policies. In this way, Artificial Assistants operate as " stateless advisors", generating value without the risk of data leakage.
Reliability and accuracy: AI as a support, not a substitute
Reliability is another key aspect: Artificial Assistants do not make autonomous decisions, but support the consultant, automating repetitive tasks, suggesting content and providing insights based on verifiable sources. In the context of financial services, where information accuracy is crucial, modern generative AI systems can be configured to maximize accuracy and minimize the risk of “hallucinations” or misinformation:
- Grounding verified sources: Output can be tied to certified internal sources (e.g. KIID documentation, internal research, compliance framework), drastically reducing the possibility of generating unverified information.
- Advanced Contextual Retrieval Systems: Modern Artificial Assistants integrate RAG (Retrieval Aggregation) technologies. Augmented Generation) that allows you to consult internal documents, updated regulations and proprietary knowledge bases in real time before formulating responses. The most advanced versions implement " agentic RAG ", where the system can autonomously decide which documents to consult, formulate intermediate search queries, and progressively verify the information through multiple reasoning steps, always maintaining human supervision on the final decisions.
- Traceability and citation of sources: A fundamental aspect is the ability of these systems to indicate precisely from which specific company document a piece of information was extracted. This allows the consultant to instantly verify the origin of the data, consult the original document with a simple click, and validate the reliability of the information. This functionality offers a double advantage: on the one hand, it allows you to efficiently navigate through thousands of documents, quickly finding relevant information; on the other, it provides an immediate verification mechanism in case of uncertainty, underlining the importance of human supervision in the process.
- Transparency and Confidence Levels: Systems are designed to transparently indicate not only the sources but also the confidence level of the generated response, explicitly signaling when they are operating in areas of uncertainty that require greater human attention.
- Performance Monitoring: Quality Framework assurance based on custom datasets created specifically for each use case in the financial context. These evaluation datasets are purpose-built to test the accuracy of assistants in real-worldscenarios and allow you to iteratively improve system performance in areas that require greater accuracy.
The relationship between the advisor and the AI is fundamentally symbiotic, not substitutive. The AI amplifies human capabilities while the human brings judgment, relational experience, and responsibility. The human advisor always maintains control and ultimate responsibility for the recommended action, with the AI acting as a powerful tool to extend its cognitive capabilities.
The result is an “augmented” advisor who can spend more time interacting with the client and developing personalized strategies, while the assistant handles the more mechanical and repetitive aspects of the job.
Development and adoption times: weeks, not months
Unlike other traditional IT projects that can drag on for years, the adoption of Artificial Assistants has surprisingly rapid development times. Based on our field experience in implementing AI solutions for the financial and insurance industry:
- A first prototype AI Assistant can be activated in 4-6 weeks, focusing on a limited use case (e.g. meeting preparation, portfolio analysis, product comparison, compliance support);
- The pilot phase with selected users typically takes an additional 2-4 weeks, during which feedback is collected, initial KPIs are measured, and quick adjustments are made;
- Large-scale deployment typically takes 2-3 months, including onboarding, integration with existing systems (CRM, Microsoft/Google suite, data lake) and training of Advisors;
- Subsequent iterations to expand functionality can be done in 2–4-week cycles, allowing for continuous refinement based on real user feedback.