Speaking at a Salesforce conference in the United States (Marketing Cloud Next / Data 360 session)
More than 20 speaking engagements a year, including international conferences
Today’s Topic
How to maximize the value of combining SaaS with Claude
Combining Claude with a business system such as Salesforce opens up enormous possibilities. But the benefits appear only when the underlying data structure is something AI can understand.
Start with Claude Code to create a data structure AI can work with.
What’s Happening Now
AI adoption is accelerating. But the answers still fall short.
Searching internal knowledge with AI
Analyzing sales data with AI
Supporting business decisions with AI
The problem is not that AI is not smart enough. The information it needs is not organized.
Our Story
For 12 years since 2013, Salesforce has grown at the heart of our operations.
At the Leave a Nest Group, we centralized sales, research, and education data in Salesforce and made it the foundation of our operations.
The more successful a business system becomes, the more complex it grows over time.
The Structural Problem
Business systems become more complex as they grow
12 Years of Results · Our Actual Data
Our custom data objects accumulated to 135
Added that yearPrior cumulative total* Company-specific data containers called “custom objects” in Salesforce
The Turning Point
AI agents suddenly made the problem impossible to ignore
People
Experience tells them which information matters
Tacit knowledge fills in duplicates and exceptions
AI agents
They process large volumes of information as-is
Multiple similar datasets create ambiguity
Irrelevant data can become part of the evidence
A system held together by human tacit knowledge is unreadable to AI.
Our Choice
Project Vanilla Core
In software, “vanilla” means the unaltered standard state.
Before adding more AI, return the business platform to a state AI can understand.
This is reform by subtraction, not addition. A simple structure built on standard features gives AI agents the foundation they need to perform.
How We Make It Happen
Claude Code × MCP, CLI, and APIs: untangling 12 years of complexity
Claude CodeAnalyze, plan, and implement
⇄Connected through three paths
MCPA protocol connecting AI and systems
CLICommand-line tools
APIInterfaces for system integration
⇄Read data and apply changes
Metadata (design information) Business data
With access to both metadata and business data, we can move continuously from understanding the AS-IS to implementing the TO-BE.
What Actually Changed
Improvements postponed for lack of time finally started moving
Before
We knew what needed improvement
Investigation alone took enormous time
So the work stayed on the backlog
After
Claude Code supports analysis and execution
Multiple improvements move forward in parallel
A continuous improvement cycle emerges
The value of generative AI is not merely saving time. It is scaling improvements that were previously impossible.
In Your Organization
This is not just a Salesforce story
Business applications such as Notion and kintone
Core systems such as CRM and ERP
Internal databases maintained over many years
The longer a system has been used and expanded, the more it needs to be organized for the AI era.
Closing Message
Before making AI smarter, create a business environment AI can understand.
Preparing for the AI era is not only about building new systems. It also means returning the systems we have nurtured to a state AI can understand. Claude Code is a powerful partner for that work.
Claude CodeAnalyze, plan, and implement
Metadata (design information)
