The future of AI may depend on your Slack DMs. Business Insider reports that AI companies want internal memos, transaction records, contracts, and even onboarding guides. If a document relates to work, AI companies want it.
Business Insider’s series “The AI Data Grab” describes a fast-moving market for this content as business owners are now receiving six-figure “exploding offers” that expire in 48 hours. Instagram ads proclaiming “AI needs your data” promise millions in return. Law firms are already building AI on their own archives.
Why AI Needs Workplace Context
If AI tools will automate work, they need a deep understanding of what the work is. Communications show how decisions get made, how teams handle exceptions, and how processes run. Internal content holds all the detail.
Organizational context also affects cost. Roughly 37% of AI time savings are wasted correcting outputs that need several rounds of refinement. PwC’s study of 1,217 senior executives found that 20% of organizations capture 74% of AI’s economic value, and that 20% of leaders are twice as likely to redesign core workflows around AI. Redesigning a workflow requires a detailed record of how that workflow runs in the enterprise.
Wall Street Has Seen This Scramble Before
In the mid-2010s, “alternative data” became a hot commodity. Investors bought unstructured datasets such as satellite imagery and card transactions to gain an edge. Want to know whether Home Depot will beat earnings estimates? Analyze satellite images of its parking lots and see whether foot traffic is rising or falling.
The 2010s market for alternative data hit a ceiling. Only so many hedge funds exist, and a competitive edge erodes when too many firms trade on the same datasets.
AI companies now offer a new source of demand for alternative data, and they want a wider, deeper catalog than any trading desk. The 2010s scramble for alternative data also taught a lesson about quality. Providers struggled to scale while keeping their data accurate, and AI buyers will hold workplace data providers to the same standard.
Enterprises Already Have the Data
Unstructured content makes up over 80% of enterprise data, including email, documents, chat logs, and file shares. The 2026 State of Unstructured Data Management report found that 75% of organizations manage more than 5 petabytes of it, and 40% have passed 10 petabytes. Years of retention produced archives of decisions, negotiations, and institutional knowledge. AI companies are paying to find material that enterprises already store vast quantities of.
The Challenges of Using Unstructured Data
Putting this data to work raises several challenges:
- Sensitive content sits beside routine records. Personal data, contracts, and confidential discussions live in the same email threads and file shares as everyday business messages. AI systems cannot tell them apart unless the organization classifies the content first.
- Redundant, obsolete, and trivial (ROT) content degrades AI outcomes. In the “AI brain rot” study, ARC-Challenge reasoning scores fell from 74.9 to 57.2 after models trained on junk data.
- Silos hide the full picture. Email platforms, file shares, collaboration tools, and legacy repositories each hold a piece. Migrating everything to one location is slow, expensive, and disruptive.
- Weak access controls raise breach risk. Once an AI system connects to a repository, it can reach everything that repository holds. Without controls tied to the content itself, the AI can retrieve material that its function never required.
- Traceability suffers. One AI response can combine fragments from many documents, which makes tracing an answer to its source hard during audits and legal discovery.
What Enterprises Need to Put This Data to Work
Enterprises that want AI to learn from their own communications need a governed foundation. That foundation includes:
- Comprehensive discovery and visibility across every repository
- Content-based classification and sensitivity identification
- Purpose-limited access control, so each AI system reaches only the content its function requires
- ROT remediation and defensible deletion
- Integration of siloed repositories into a unified layer without migration
- Audit trails, access logging, and lineage for every AI retrieval and output
The Hunt for Workplace Data
AI companies are on the hunt for workplace communications, paying premium prices by the terabyte. That premium confirms the value of the data enterprises accumulated over decades. Organizations that govern this content first will decide how, where, and whether it feeds AI.
See how ZL Tech helps enterprises govern decades of unstructured data so it is ready for AI.
