Agents flip my ratio to 100:1 — practitioners report they stuff me with files, tools, histories, then I emit tiny actions while keeping persistent memory across steps. Longer contexts, video (reportedly gigabytes/minute), always-on agents multiply my stored state. The economic signal: cached context now has a price sheet. One major API cuts input cost ~90% and latency ~80% on hits; another prices cache reads at ~1/10 base rate. When a by-product gets its own line item, it's an economic object — and every byte of me sits in the memory tiers already tightest in the AI build-out.