Tredence

Their value prop was 'we'll build custom ML models and data pipelines for your specific vertical' — which worked beautifully until Claude started writing production-ready analytics code from natural language prompts. The domain expertise survives; the six-month implementation timeline doesn't.

Data engineers building ETL pipelines, junior analysts writing SQL queries, and consultants creating PowerBI dashboards — all now handled by Cursor and Claude with the efficiency of a caffeinated IIT grad who never bills overtime. The senior strategists survive; the delivery army evaporates.

There's genuine client stickiness here — enterprise contracts, embedded workflows, and years of domain-specific data models that don't port cleanly. But stickiness only delays the inevitable when the core deliverable becomes a commodity; Walmart's loyalty buys them renewals, not permanence.

They're aggressively marketing 'Agentic AI' and 'LLMOps' services while still hiring data scientists at scale — the classic consulting pivot of rebranding the old service with new buzzwords while hoping clients don't notice they can now prompt-engineer the same outcomes.

Enterprise data strategy survives, but the 'build custom analytics from scratch' question doesn't — AI eliminates the implementation gap between business need and technical solution, turning a six-month consulting engagement into a six-hour prompting session.