Dashboard tour

Overview#

Your savings at a glance, across API calls and the Agent SDK:

• tokens and cost saved, and how many requests were optimized;
• input tokens over time, before and after Contextwaise;
• savings by channel (API call or Contextwaise SDK) and the time Contextwaise adds, as a median and p95;
• every agent, with the channels it uses, and the latest activity;
• model routing advice, when you send models.

Requests#

Every request, newest first, with a switch between API calls and Contextwaise SDK results. Open an API request to see each candidate with its score, decision and reason, the budget and threshold applied, the quality check's verdict and the time each step took. This is the first place to look when an answer seems to miss context.

The SDK view lists tool results per tool: what each returned, what the model read, the reduction, and how often agents expanded a trimmed result.

Sources#

Savings broken down by the source label of your candidates (rag, web, tools…), to see where your waste comes from.

Agents#

Every agent sending context, with its token budget: how much of its budget each agent uses on average and at most, and how often it hits it. Change a budget here and the next requests use it, no deploy needed.

Playground#

Paste a question and some context, or start from an example, and see what Contextwaise keeps, without writing code. Playground requests are tagged as such in Requests and Agents.

Billing#

Your plan, the context tokens processed this month against what it includes, what Contextwaise saved you this month, and the other plans. Usage counts the tokens you send to optimize, plus the tool results the SDK sends for scoring.

Settings#

• Optimization defaults: the default budget, aggressiveness, and the quality check.
• Models: your default model, and prices for models outside the catalog.
• API keys: create and revoke keys.