‘Beyond CRM’: How Wilson Sonsini and Salesforce are preparing law firm data for agentic AI

# Summary At ILTACON Wilson Sonsini's CIO Jeffrey Lolley and Salesforce's Kiel Parker explained how a long-standing CRM relationship has evolved into a platform for AI-driven intelligence. Their work focuses on preparing a legal firm's fragmented data so agentic AI can deliver practical value: faster pitches, cleaner RFP responses, and decision-support for associates.

# What they built

# Data and governance—concrete choices The firm identified three problems and designed for them:

• Fragmented sources: Accept that data will remain distributed and build a data layer that aggregates metadata and context.
• Clean data requirement: Prioritize getting clean, structured data before applying AI—buying tools that sit on dirty data does not work.
• Client obligations and compliance: Create a governance layer that enforces policies and works across models, keeping client constraints central.

Salesforce's 'harmonisation engine' was shown in demo as the metadata ingestion and mapping component inside Salesforce Data Cloud that helps make the aggregated layer usable.

# Adoption and trust Lolley described a phased licensing and adoption approach. Instead of forcing company-wide change, the firm demonstrated clear wins: lawyers went into meetings with information they otherwise would not have had and continued using the platform. This incremental approach increased trust and drove broader use.

# How agentic AI fits in now Wilson Sonsini frames agentic AI as an accelerant to existing work rather than an immediate overhaul. Near-term value examples include:

• Speeding pitch and RFP responses by surfacing institutional knowledge and relevant touchpoints.

The firm aims to "structure the firm round" these tools so AI becomes part of normal workflows and thought processes.

# Practical implications for other law firms

• Start with data hygiene: focus on cleaning and structuring the data you already have before layering AI tools on top.
• Build governance early: design controls that can operate across different LLM providers to avoid lock-in and to respect client obligations.
• Deliver visible wins first: adopt incrementally and surface immediate benefits to users so trust grows organically.
• Treat AI as workflow change: plan to reorganize processes so automated decision-support replaces low-value manual checks instead of just adding tooling.

# Bottom line