Choose what your memory spends.
Retrieval and synthesis do different amounts of work. Memrain exposes model usage and spending controls so you can choose which capabilities to enable and evaluate their cost.
Illustrative example
Can we try richer retrieval without leaving model spend open-ended?
Set the relevant daily budget, enable only the features being evaluated and review the spend log against representative questions.
Illustrative scenario, not a live product result.Understand the cost of the selected features.
Default search uses a Titan query embedding without a chat-model call. Indexing creates embeddings for content, and optional reranking, fact extraction, synthesis or research agents can add model usage. Infrastructure and your AI client’s own subscription are separate costs. Free software does not mean a free AWS deployment, and a configuration tier is not a paid Memrain subscription plan.
Set limits around an evaluation.
The operator can configure supported daily USD caps for credentials or enrolled people, plus an optional deployment-wide cap. Agent jobs and some batch workflows also require explicit budget controls. These limits apply to priced model calls in the relevant paths; parts of the accounting are approximations. They do not cap the whole AWS bill or charges from an independently connected AI provider.
Review actual use before widening the rollout.
The admin spend view and operational logs help reveal which features are being used. Start with a small corpus and a fixed set of real questions, then inspect the resulting activity. A scheduled extraction or synthesis feature can spend outside an interactive search, so include background work in the evaluation. Discuss the deployment shape before assuming a flat subscription price.
Keep the context. Move the work forward.
Connect your knowledge to the AI you already use. Find your starting point and make the next conversation count.