AI Isn't Blocked by Bad Data

A common refrain we hear from investment teams right now is: "I need to clean our data before we can really use AI." It sounds responsible, but usually the wrong diagnosis.

In practice, most firms don't struggle with AI because their data is messy. They struggle because their data has no clear authority. Messy data is manageable. Undefined authority is not.

Investment teams live with imperfect data every day: multiple versions of models, overlapping research notes, PDFs saved in different places, and email/message threads holding key decisions.

Humans navigate this surprisingly well. We rely on context, judgment, and institutional knowledge. AI doesn't have that intuition.

When AI struggles, it's often not because the data is "dirty." It's because it can't answer basic questions like: Which version is the source of truth? What's the difference between reference material and a working draft? What's outdated but still relevant? Who owns updates when assumptions change?

Without authority, AI doesn't fail loudly. It fails quietly by producing answers that sound reasonable but drift over time. That's the real risk.

"Clean data" is usually a proxy for something else

When teams say they need to clean data first, what they often mean is: We don't trust what we have. We don't agree on what matters. We don't have shared context across teams. No amount of file cleanup fixes that.

You can reorganize folders endlessly and still end up with conflicting answers to the same question, analysts recreating work that already exists, and decisions that can't be traced back to their rationale. AI simply exposes these problems faster.

Authority comes before accuracy

AI doesn't need perfect data. It needs clear signals about what to trust.

Authority looks like declared sources of truth, persistent context instead of one-off answers, outputs that can be referenced and improved, and a shared memory that the organization can build on.

Once authority exists, accuracy improves naturally over time. Without it, better models just produce more confident inconsistency.

The quiet shift is happening now

The most forward-thinking teams aren't asking: "How do we clean everything before using AI?"

They're asking: "How do we establish authority and let AI help us get better over time?"

~Ryan Erickson, Founding Executive

About Shift
Investment research shouldn't be this hard. Shift turns your firm's scattered knowledge into powerful insights with AI built for how you actually work. We're a team of builders and finance experts based in Charlottesville, VA.