Three-year savings model.
Last updated: 2026-05-05
ROI
The commercial argument for Mycelium is that a customer-owned memory layer prevents the switching cost when an AI vendor changes. This page is the math behind that argument. Inputs are yours. The formula below is public and auditable. The result is your number, not a vendor claim.
Calculator
Read the full formula below. Adjust inputs to see your numbers update.
Estimated three-year savings
The formula
- Switching cost = 30% of one year of AI spend. Source: aggregate of published vendor-migration studies; the 30% figure covers data extraction, re-onboarding, retraining, and the productivity dip during migration.
- Switches in three years = 3 / your stated switch interval.
- Three-year switching cost (status quo) = switching cost × switches in three years.
- Mycelium prevents most of the switching cost because the memory layer is portable: the data, schema, and skills survive the model swap. Mycelium-prevented cost = three-year switching cost × 0.85 (the 15% residual covers the model-specific prompt engineering and skill calibration that does not transfer).
- Mycelium three-year cost = annual platform license (Lite or Enterprise) × 3, plus estimated usage. Usage is workload-specific; the calculator uses 30% of platform license as a conservative usage estimate. Engagement letter has the actual unit rates.
- Three-year net savings = Mycelium-prevented cost minus Mycelium three-year cost.
What this calculator does and does not do
It estimates the order-of-magnitude switching-cost savings against a public formula. It does not estimate productivity gains from agents that work better, time-to-decision improvements, headcount avoidance, or revenue lift. Those are real but they are not in this number. The estimated number on this page is the floor, not the ceiling.
Mycelium · founded 2026