From Horses to Cars: How AI is Reshaping Business Software
Why the B2B software stack of today will be obsolete in 3-5 years—and what product managers should do about it.
A few years ago, when I was a product manager at Salesforce, we were building Experience Cloud, a platform to create B2B and B2C portals. In every sales call, we proudly pitched how easy it was for customers to build and customize their portal.
Site templates to give them a head start.
Pre-built components they could drag and drop.
Builders for pages, dashboards, workflows, and automation.
Predefined reports, dashboards, email templates, and quote forms—a favorite for finance teams.
Our advantage was making complex implementations easier through structured templates, builders, and automation tools.
But here’s the irony: What was once a competitive advantage will soon become obsolete.
The Old vs. New Playbook: How to Think About the Shift
This article isn’t just about CRM. It’s about every B2B software category that has been built on structured workflows, admin interfaces, and prebuilt automation layers.
The premise is simple:
The software stack making money today will become obsolete in 3-5 years.
AI isn’t just automating small tasks—it’s replacing entire layers of software.
The companies that recognize this shift and embrace reinvention will thrive. The ones that rely on incremental AI features will get disrupted.
To illustrate the scale of this transformation, let’s go through several key areas where AI is rewriting the software playbook, comparing the old model (how things work today) with the AI-driven future.
From CRM Databases to AI-Driven Orchestration
Old Model: Structured Data and Rigid Relationships
Business software has long relied on structured databases with predefined tables, fields, and relationships. CRM, ERP, and finance systems all store data in fixed schemas, requiring updates whenever business needs evolve.
This worked well when data was simple, but today’s interactions—emails, chat logs, call transcripts, and product usage—don’t fit neatly into a structured model. Maintaining a 360° customer view now requires syncing CRM with e-commerce, support logs, marketing automation, and third-party data sources, which is costly and complex.
New Model: AI-Orchestrated Customer Data Across Multiple Sources
AI removes the need for predefined schemas, allowing businesses to move from rigid, structured databases to AI-driven orchestration of unstructured data. This shift is already happening: Snowflake, and Databricks are leading the way with data lakehouses that unify structured CRM records. Salesforce itself is investing heavily in CDPs (Customer Data Platforms), recognizing that structured CRM data alone isn’t enough to power AI-driven experiences.
Example:
AI enables real-time understanding of unstructured data:
AI listens to customer interactions across email, chat, and calls and extracts insights dynamically.
The salesperson simply asks, “What’s the latest on Acme Corp?”
AI retrieves all relevant customer details—without requiring structured fields—and recommends next steps.
The End of Dashboards: AI-Generated Insights
Old Model: Predefined Dashboards with Limited Flexibility
Businesses rely on prebuilt dashboards and reports that come with software solutions like CRM, ERP, and BI tools. These dashboards serve a fixed purpose. Any business change requires manual customization, additional development, or installing third-party extensions.
New Model: AI-Generated Reports On-Demand
AI eliminates predefined templates by dynamically generating reports and dashboards on request. Instead of configuring filters and queries, users ask AI directly, and it builds the best visualization based on real-time business data.
Example: ChatGPT and Claude’s AI-Generated Reports
ChatGPT and Claude already generate reports from spreadsheets, databases, and external business systems through simple prompts.
New Model: The leader asks AI, "Show me my revenue growth by product line over the past six months, with trends and key risks." AI analyzes data, selects the best graphs, and presents a tailored report instantly.
UX on the Fly: The End of Static User Interfaces
Old Model: Fixed Interfaces and Menu-Driven Navigation
Traditional business software has fixed UIs, requiring users to click through menus, dashboards, and forms to find what they need. Even with customization options, users must adapt to the software’s structure rather than the software adapting to them.
New Model: AI-Generated, Adaptive Interfaces
Instead of navigating through static screens, users interact with AI conversationally—and the system builds an interface dynamically based on their needs. The UI adapts to the user’s workflow, surfacing only the most relevant data, tools, and actions.
Example: How AI is Changing UX
The rep simply asks AI, “Show me my top priority accounts and summarize the last interactions,” and the system instantly generates a personalized interface with key details, relevant actions, and AI-powered insights—no manual navigation required.
The End of Builders & Admin Interfaces
Old Model: Complex Builders and Admin Panels
B2B software often provides drag-and-drop builders, configuration tools, and admin consoles to customize workflows. These tools come with complexity: hundreds of settings pages (Salesforce has 700+), expensive maintenance, and static configurations.
New Model: AI-Driven Configuration and Adaptive Systems
AI eliminates manual configuration by understanding business needs and adjusting settings dynamically. Instead of a complex admin panel, users interact with AI conversationally, and the system self-optimizes based on usage patterns.
Example: How AI is Replacing Traditional Builders
The admin simply tells AI, “Route enterprise leads directly to senior reps and prioritize follow-ups within 24 hours,” and the system configures it instantly.
Replacing Static Workflows with AI Agents
Old Model: Rigid, Rule-Based Workflows
Business workflows rely on predefined automation rules—static sequences of actions. Workflows are rigid. Once configured, they follow fixed rules. And they lack intelligence. Workflows execute exactly as programmed and cannot adapt to changes.
New Model: AI-Driven Autonomous Agents
AI agents orchestrate processes dynamically, making real-time decisions based on context, business goals, and historical data. AI agents eliminate the need for manual setup: AI creates, executes, and refines workflows automatically.
Example: How AI is Replacing Traditional Workflows
AI autonomously analyzes the deal size, customer history, and risk profile, then decides the best approval process dynamically. Instead of waiting for static approval steps, AI can fast-track or escalate deals based on real-time conditions.
The Ford & Horses Parallel
In the early 1900s, people laughed at the idea of cars replacing horses. The automobile industry seemed expensive and impractical—until Ford’s Model T transformed transportation.
AI-native software is following the same trajectory. What seems unrealistic today will become the new default in 3-5 years. The only question is:
Will you embrace the shift or wait to be disrupted?
For the all product and engineering team, their role will be redifined as well - I still look for the next name of the next org where you have smaller teams that can prototype and iterate faster with a continious feedback loop.