How Addepar Scales Investment Workflows with Databricks AI Agents

A unified data and AI foundation for financial services

Addepar is a global technology and data platform that empowers investment professionals to turn complex financial information into actionable intelligence. Registered investment advisors, family offices, private banks and global institutions rely on Addepar to unify portfolio, market and client data and deliver a total portfolio view across public and private markets.

Data and AI are fundamental to this mission. Addepar now manages nearly $9 trillion in assets on its platform, and clients rely on security, quality and consistency to make informed, high-stakes decisions. To support this, Addepar moved from a collection of older systems and database tools to a single data intelligence platform on Databricks running on AWS. That platform ingests hundreds of disparate data feeds, standardizes and enriches them and then delivers the results to clients through products, APIs and data sharing.

Building on Databricks for scale, governance, and collaboration

Addepar chose Databricks to unify engineering, analytics and AI on a single, governed data platform. Collaborative notebooks and SQL let internal teams work in one place, while Unity Catalog provides the fine-grained permissions and access controls that a global financial services footprint demands.

The result is a single source of truth that engineers, analysts, and now AI systems can all depend on.

This decision has produced a clear business impact. Since adopting Databricks, Addepar has reduced pipeline costs by 60% versus legacy infrastructure—driving more than $2 million in infrastructure and data processing savings—and achieved a 5x improvement in the speed of delivering new pipelines and integrations. That acceleration helps onboarding, client delivery and experimentation, while the Databricks and AWS combination gives Addepar the scale and reliability needed to grow with its clients.

Addison: a native AI experience embedded in the platform

Building on its unified data foundation, Addepar has introduced Addison, a native AI experience embedded directly within the platform. Addison is designed to provide trusted guidance and actionable insights that are grounded in Addepar’s core data and workflows.

Addison goes beyond a chat-based interface, to:

• Live inside Addepar’s core platform, integrated directly with portfolios, solutions and workflows.
• Understand the “nouns and verbs” of finance in the context of Addepar’s data model.
• Combine Q&A, proactive insights (push) and action-oriented workflows into a single experience.
• Surface relevant market news alongside portfolio data, helping advisors connect client holdings to current market events.
• Run on Addepar’s core calculations engine, referencing the same portfolio metrics and performance calculations used across the platform.

For investment professionals, Addison acts like a digital partner:

• Pull: Advisors ask questions like, “Break down this portfolio’s alternatives allocation,” “If rates rise by 50 bps, what is the projected impact on fixed income duration?” or “Identify any accounts that have drifted more than 3% from the target,” and Addison responds using live, governed data.
• Push: Addison surfaces notifications and events, such as emerging risks, opportunities or anomalies in portfolios, without requiring explicit prompting.
• Act: Advisors initiate workflows, such as running a financial plan,, or exploring alternative allocations, understand portfolio trends and behaviors – while Addison helps orchestrate the underlying data and steps across Addepar tools and workflows. These capabilities are designed with humans in the loop, keeping investment professionals firmly in control of decisions and actions.

The vision is that natural language, workflows and intelligent agents become the primary way users interact with Addepar. By offloading tedious data manipulation and orchestration to Addison, investment professionals can focus more time on relationships and strategic decisions.

Safe, scalable GenAI for financial services

Because Addepar’s clients operate in highly regulated domains, Addison’s architecture must be safe and scalable in ways that generic consumer models, such as direct calls to public LLMs, cannot match. Addepar prioritizes security, data privacy and governance, and has designed its AI stack accordingly.

By transforming its infrastructure on Databricks, Addepar utilizes Unity Catalog, with permissions and access controls deeply integrated into its environment. Those same controls surface in Addison. A combination of cutting-edge frontier models are served and hosted within Addepar’s environment via Databricks Model Serving, and are tracked and managed with MLflow, delivering consistent lifecycle management and auditability.

Keeping both data and models inside the Addepar ecosystem is critical for personally identifiable and client‑identifiable data across Addepar’s global infrastructure footprint. It helps the company meet client expectations around risk, compliance and legal or jurisdictional concerns.

This approach means Addison is not just an LLM endpoint. It is an AI system that inherits the same governance guarantees as the rest of Addepar’s platform, something that would be significantly harder to achieve with fragmented tools or unmanaged external APIs.

From LLMs to agents with Agent Bricks, Foundation Model Serving and MLflow

Simple LLM prompts can be powerful, but making them reliable and repeatable enough for production financial services workflows is difficult. It requires orchestration, validation and iteration to reach the level of consistency advisors and investors need.

Addepar is now adopting Databricks Agent Bricks as the next evolution of its AI journey, starting with Supervisor Agent that coordinates Genie‑powered analytics behind the scenes. Addison uses these Supervisor flows to move from “LLM plus prompt” to trusted, agentic workflows, where the system can execute sequences of actions on behalf of advisors with their oversight. What was previously a disjoint, manual process of wiring together prompts, tools and validation logic is now centralized and simplified by Agent Bricks, including early use of multi‑agent Genie workflows to power internal Slackbots and advisor experiences.

Addison leverages LLMs served from Databricks Foundation Model APIs, which provide access to state-of-the-art models from a variety of model providers through governed serving endpoints. Production financial services workflows demand transparency, audibility, and fine-grained evaluation of AI accuracy. Addepar leverages Databricks Managed MLFlow to power traceability and granular insights into individual agent workflows. Addepar also now uses MLFlow to develop, evaluate, and iterate on Addison's performance and behavior.

For Addepar, all of this means it can define agent workflows, such as multi-step portfolio analyses, planning flows or automated insight generation, test them rigorously, and deploy them with governance, all on the same platform that powers its core data. This is a uniquely Databricks value proposition: unified data, governance and agent orchestration in one place.

Collaboration and data sharing as a force multiplier

Databricks has also changed how Addepar collaborates internally and with clients. Previously, different types of users inside Addepar and at client organizations often worked in a transactional way using spreadsheets, extracts and one-off API exchanges. Collaboration was limited and disjointed.

With Databricks Notebooks and Unity Catalog, Addepar can now share data, code and SQL in a single environment with the right access controls. Teams can work on data and models in the same place, and that collaboration extends to AI. They can share models, configurations and prompts with consistent context. For clients, being able to view the same data simultaneously builds trust, reduces miscommunication during onboarding and ongoing operations, and supports a more accurate and transparent view of portfolios.

A partnership focused on outcomes

Addepar provides the foundational data platform for the investment ecosystem, bringing together complex portfolio, market and client data to power the workflows investment professionals rely on every day. To support the scale, security and innovation the platform requires, Addepar works closely with technology partners like Databricks and AWS, whose capabilities help power key elements of its data infrastructure. These partnerships are built around open exchange and shared success rather than a simple vendor transaction.

As Databricks continues to advance its data and AI capabilities, Addepar expects Addison to become the primary way many users experience the platform. By combining a unified, governed data foundation with GenAI and agents, Addepar helps investment professionals cut through complexity across portfolios, data and workflows to make more informed decisions and deliver better outcomes for the clients they serve.

Attend Databricks AI Days in a city near you to learn how to take control of your data and build AI agents that drive business impact.