Published on
Every firm wants to empower its teams with AI.
We are entering an era where AI has accelerated the way firms streamline workflows for their data and operations teams, advisors and investors. Firms rely on that intelligence to generate alpha and automate tasks through AI agentic workflows. But it's no secret that there’s a big problem.
Gartner finds that by the end of 2025, at least 50% of generative AI projects were abandoned after proof of concept due to poor data quality.1
Nearly every RIA, private bank and institutional firm wants to implement AI agentic workflows, but most struggle to harness it. Either the data is difficult to govern, inconsistent or can’t be shared at scale. Poor data quality introduces significant risk.
The AI expectations gap
Data and operations teams at financial firms are being asked to do more, build more sophisticated analyses, connect more data sources and support a growing number of AI requests. At the same time, advisors and investors want faster access to enriched data and contextualized insights. The list of AI needs includes:
Advisors want natural language queries to understand portfolio performance
Operations teams want automated reporting
Data teams want tools to build advanced models and analytics
Firms want AI agents for operational leverage
The expectations are clear, and the need is there, but the foundation can’t keep up. When data is unstructured, ungoverned and difficult to orchestrate, the fuel for AI engines is locked away.
AI doesn’t fail because of the models
Most AI initiatives don’t stall because the models aren’t powerful. And they don’t stop because the chatbot can’t generate answers. They fail because the data underneath those tools isn’t complete, consistent or governed. So instead of intelligence, firms have:
Chatbots returning incorrect information
Analytics tools operating on different datasets
AI agents stalling when the data isn’t aligned
Teams falling back into spreadsheets, trying to reconcile the numbers
The tools aren’t the problem. Fragmented data is. Over time, data drifts out of sync across systems. Different teams rely on different data sources. Pipelines break, definitions change and a lack of confidence in the data erodes trust. Now, AI isn’t accelerating workflows, it’s only amplifying the gaps.
Tool-based data strategies are limiting
Many firms built their analytics environments over time with dashboards that were layered on top of warehouses, point integrations feeding reporting tools and manual pipelines connecting systems together.
For a while, it worked. But as firms scaled, complexity grew. Data volumes increased and:
APIs struggled to handle deeper data models
Different integrations pulled data at different times
AI tools began relying on data that wasn’t fully synchronized
Suddenly, the system of record firms rely on wasn’t so reliable. And the risk of inaccurate or ungoverned AI grew with it.
Modernization means moving beyond the dashboard layer
AI and agentic workflows require something deeper than visualization tools.
They require a unified data foundation that is structured, governed, historically consistent and scalable across data sources.
The firms pulling ahead in the AI race are rethinking the data stack and how teams access data. They’re moving toward platforms that centralize, model and connect data in ways that support different sides of the firm. AI then becomes an outcome of their infrastructure.
Why firms are choosing Addepar Data Exchange
Addepar Data Exchange (ADX) gives firms an enriched data foundation that flows throughout their entire ecosystem, purpose-built for financial data. ADX combines the technical strength of Databricks with Addepars' expertise in financial data to truly unlock collaboration between technical teams and business users within a firm.
Data teams gain a managed data environment powered by Databricks to build more sophisticated analytics and AI workflows using Python/SQL. At the same time, downstream users — advisors, operations teams and investors — benefit from the insights generated on top of that trusted data within the Addepar platform.
If firms already have data sources and BI tools, Addepar can feed data directly into those systems, increasing accuracy without forcing teams to rebuild their stack or replace existing data lakes.
Key features of ADX
Data model ->
Provides a structured financial data foundation across portfolios, ownership structures, transactions and valuations, enabling analytics between teams.
Ingress ->
Allows firms to securely ingest external data directly into ADX, automating pipelines, and reducing the need for manual integration or fragile API workflows.
Data archive ->
Preserves point-in-time data history for governance, reconciliation and auditability providing firms with a way to trace how data has evolved over time.
Reference:
Why 50% of GenAI Projects Fail — And How to Beat the Odds, Gartner, January 2026.