AI Solutions Dashboard

Internal tools used by the KantaSwara delivery team to build, manage, and deploy client agents.

Delivery Console Workflow

The AI Solutions dashboard (often called the Delivery Console) provides a unified view of all pending client requirements, active development tasks, and production agent health.

1. Requirement Queue

When a customer submits a new agent requirement, it lands in the Requirement Queue. AI Engineers can review the requested use case, target audience, and provided knowledge documents before claiming the ticket and initiating a kickoff call.

2. Builder Interface

This is the core engineering workspace where the agent is actually constructed.

• Prompt Engineering: Direct access to the system prompt configuration. Engineers define persona, strict conversational rules, and fallback behaviors.
• Knowledge Mapping: Tools to ingest client documents, manually edit vector chunks, and test retrieval accuracy using similarity search queries.
• Tool Configuration: Defining the JSON schemas for function calling (e.g., mapping a "BookAppointment" intent to a client's specific Calendar API format).

3. Internal Simulator

Before handing the agent over for User Acceptance Testing (UAT), engineers use the Internal Simulator to have live audio conversations with the draft agent directly through their browser, testing edge cases and latency.

4. Deployment Manager

Once UAT passes, the deployment manager allows engineers to provision Twilio numbers, map them to the finalized agent configuration, and switch the agent from "Development" to "Production" status.