If you fork this tier, keep the FAISS-vs-Chroma comparison table but update the "Default for this tier is..." recommendation if your own testing points the other way for your corpus — don't just leave the generic advice unexamined once you have real data.
Say explicitly in the Overview what "user" means in your fork — is it an individual, or does it map to something else in your domain? This tier's isolation is per-user; be clear about that so nobody assumes team-level isolation they don't have (that's the advanced tier).
Keep the "Which tier do I need?" table pointing at both neighbors so people don't get stuck fighting this tier's assumptions when a different tier already fits.