Should Data Engines Start Feeding AI Agents Instead of Dashboards?
Notes on the role of data engines
Welcome to Infinite Curiosity, a newsletter that explores the intersection of Artificial Intelligence and Startups. Tech enthusiasts across 200 countries have been reading what I write. Subscribe to this newsletter for free to receive it in your inbox every week:
For years, the playbook for data infrastructure followed a predictable formula: extract data, transform it, and load it into a centralized destination. This destination is usually a dashboard or a data warehouse.
This worked when human decision-making was the primary bottleneck, but that world is quickly fading.
The real opportunity today is in consolidating data not for human consumption but for AI agents that can reason, act, and automate workflows. This shift unlocks massive startup opportunities and fundamentally changes how companies operate.
The Old Way: Consolidating Data for Static Dashboards
Data pipelines were built to serve dashboards. Companies aggregated information from multiple sources—databases, APIs, logs—and funneled it into BI tools like Looker or Tableau. Many companies still do! The idea was that decision-makers needed clean and structured data to analyze trends and make informed choices.
This approach worked well when companies were optimizing for human intuition. Analysts built dashboards, executives reviewed them, and teams made decisions based on historical data. The cycle was slow, but it fit the constraints of manual decision-making.
The Problem: Dashboards Are Dead Weight
Dashboards are supposed to provide clarity. But in reality, they often create inefficiencies. Most dashboards become static reports that executives check sporadically. The process of surfacing insights, interpreting them, and translating them into action is slow and prone to human bias.
Data intelligence tools require significant resources (teams of analysts, ongoing maintenance, constant tuning) to remain useful. And even when dashboards deliver valuable insights, they don’t do anything with them.
The burden of action still falls on human operators, making the system sluggish and inefficient.
In a world where AI can process data and act on it instantly, forcing humans into the loop is a bottleneck. Instead of surfacing insights for human review, companies should be thinking about how to enable AI to execute decisions directly.
The New Way: Consolidating Data for AI Agents
The paradigm shift happening now is clear. Data consolidation isn’t just about visualization anymore. It’s about action. Instead of building data pipelines that serve dashboards, companies should be building pipelines that serve AI agents.
Consider how this shift is already playing out:
Finance: Instead of CFOs analyzing financial reports, AI agents monitor key cash flow, flag anomalies, and rebalance budgets automatically.
Sales and Marketing: Instead of sales teams interpreting CRM dashboards, AI agents analyze deal progress, forecast revenue, and generate personalized outreach campaigns.
DevOps and Security: Instead of engineers reviewing system health dashboards, AI agents detect anomalies, trigger rollbacks, and patch vulnerabilities autonomously.
This approach eliminates the latency of human decision-making. AI agents can ingest structured and unstructured data—spreadsheets, logs, emails, support tickets—synthesize insights in real time. And then execute tasks without waiting for human intervention.
By removing dashboards from the equation, companies accelerate decision-making and reduce operational drag.
The Startup Opportunity: AI-Native Workflows
This transformation opens up multiple avenues for startup founders and investors. Some of the most compelling opportunities include:
AI-Native Data Pipelines: A new generation of data tools is needed to structure, normalize, and route data specifically for AI agents rather than BI dashboards. This would be the equivalent of “Fivetran for AI Agents”.
Verticalized AI Agents: Most SaaS tools today are passive reporting systems. The next wave of enterprise software will be AI-driven, where the product doesn’t just provide insights but actively runs business functions.
Agent Management Platforms: As companies adopt multiple AI agents, they’ll need infrastructure to manage permissions, memory, workflows, and reasoning across systems. The company that builds the "Zapier for AI agents" could define this space.
LLM-Optimized Data Structures: Traditional databases and warehouses are optimized for human querying. A startup that rethinks data storage for AI-first environments through vector embeddings, agent-aware data lakes, or LLM-specific indexing could gain significant traction.
The VC Perspective: Where are the Opportunities
If you’re a founder, the key takeaway is simple: the world is heading towards a lower number of dashboards. Start building decision loops where AI agents act rather than waiting for humans to interpret insights.
For VCs, the most valuable companies in this space won’t just be the ones building individual AI agents. We’re witnessing a fundamental shift in how companies operate. The biggest outcomes will come from startups building the infrastructure, data layers, and orchestration systems that enable AI-first business operations.
If you're a founder or an investor who has been thinking about this, I'd love to hear from you.
If you are getting value from this newsletter, consider subscribing for free and sharing it with 1 friend who’s curious about AI: