Applications Are Dead: Long Live the Data‑Centric Enterprise

“Business applications as we know them will collapse in the Agent era.” — Satya Nadella, Microsoft CEO

1. Why SaaS Applications Thrived—and Why It’s Time Is Up

SaaS pioneers like Salesforce, ServiceNow, SAP, and Workday flourished by doing at least three things very effectively:

1. Capturing functional data (Sales, Customer, HR, IT tickets, etc.)
2. Presenting that data in a usable format back to functional leaders via their UIs.
3. Keeping the customer’s data or data ownership within the context of the application vendors’ control

However, with the advent of AI agent workflows, AI agents do not need a traditional application user interface to be effective. Agents speak MCP, A2A, and API, not GUI. Today, in 2025, with an AI agent, records can be automatically retrieved from Salesforce, combined with inventory data from SAP, and integrated with other data sources to take action or publish business insights into Slack—all without a human ever needing to log in to an application.

2. AI Agents Are The Rage, But Governed Data Access Is the Bottleneck

The industry, and SaaS application vendors specifically, are leaning into AI agents, but have limited access to mission‑critical data, i.e., based on their historical focus, they only have access to certain functional data, such as HR, IT, Finance, etc. The Workday’s or ServiceNow’s AI agents are thereby limited to the data they have access to. Compounding the “AI agent effectiveness” issue FOR THE APPLICATION VENDOR is that organizations have sprawling data architectures, different sources, on-prem, in the cloud, and cross-cloud, which are not easily accessible.

3. The Trojan‑Horse Playbook

Most application vendors, with SAP, ServiceNow and Salesforce taking the lead with Data Cloud and their pending acquisition of Informatica, have recognized the data bottleneck threat, and their response has varied, but in simple terms, they use the business benefits of their AI agents as a lead to gain access to more of your or the organization’s data.

• Salesforce bundles Agentforce on top of Salesforce’s Data Cloud, enabling customers to stream their ERP, web, IoT, etc. data into the Salesforce platform.
• SAP embedded Databricks inside SAP’s Business Data Cloud, opening its application to data feeds and ingesting customers’ enterprise data via Databricks.

It’s a brilliant product strategy – but it also betrays the real and underlying fear: without the application vendor’s access to all the customer’s data, their AI agents can’t perform effectively.

4. A New Moore’s Law for AI Agents 

Are AI Agents ready to displace SaaS applications today? The application vendors themselves believe so, given their hyper enthusiasm for AI agents, and the verdict is that it’s just a matter of time. Researchers tracking multi‑agent benchmarks find that the task duration an agent can handle doubles about every 7 months. If that efficiency curve holds, an organization that starts planning today could run near‑autonomous 24‑hour workflows on the organizations and their partner data in a couple of short years —far faster than any multi‑year SaaS migration.

5. Blueprint For AI-Agent Success: Build a Data‑Centric Architecture

I believe Larry Ellison stated, “He who owns the database owns the customer.” In the era of AI agents, the statement has evolved to “Those that have access to all the data own the customer.”

The reality today is that:

• The customers themselves hold or have access to all or most of their own data. 
• AI agents can be vendor‑neutral and are dependent on data access.
• With MCP, A2A, AG-UI etc. the application interface guards are gone, and the data layer is the new battleground.

How could an organization build a Data-Centric Architecture?

1. Starburst Data for unified data access (on-prem, in the cloud, and cross-cloud) without data movement (aka zero-copy) AND secure and compliant data management with any AI agent.
2. Object storage (S3, GCS, ADLS) as the system of record.
3. Access to AI agents (OpenAI GPT‑o3, Anthropic Claude, etc.) hosted in a VPC.
4. AI Agent event orchestration and prompting

6. Implications for the Big Four

As the example of Data-Centric Architecture illustrates, customers can today bypass application vendors by taking control of their own data access and data management and using a vendor-neutral AI agent to perform business tasks on their own data. 

• Salesforce must convince customers to connect and pour non‑CRM data into Salesforce’s Data Cloud or risk being reduced to a commodity UI.
• ServiceNow’s Workflow Data Fabric is the right directional approach, but only if it can access all the relevant customer data.
• Workday excels in HR logic yet faces the same constraint: without finance, project, and learning data, its AI agents can’t reason holistically about, for example, talent.
• SAP has bet on Databricks and has thereby excluded itself from customers running their data platforms exclusively on Snowflake

In all cases, the power now lies with the data-savvy and AI-focused organization, not the application vendors.

7. What to Do Now

Despite the focus of SaaS Application vendors on AI and AI agents, AI is essentially bypassing these vendors. The winners of this new AI era will be the organizations that treat their data and data access—not applications—as the strategic asset to unlock the value of AI within their organizations and reap the benefits. 

1. Stand up a unified and governed data ecosystem, ideally with Starburst Data. Starburst ensures both data access and secured, governed, and compliant data management with the AI agent.
2. Pilot a vendor-agnostic AI agent such as the Starburst AI agent or Sierra.ai on one of your business problems (e.g., quote‑to‑cash) and measure the delta vs., e.g., Salesforce’s AI agent.
3. Go it alone with a GSI such as Accenture, Deloitte, PWC, etc., or negotiate with SaaS vendors from a position of strength: You own all the necessary data, and therefore, you are in the driving seat for AI agent success and broader AI impact for your organization.

What is certain is that a major platform shift is underway, driven by the value of AI and customers’ broader control of their data and access to vendor-neutral agents.