Project Description
I need our KarbonHQ workspace streaming continuously into Snowflake so that Claude can run live analytical prompts on up-to-the-second data. The pipeline must pull every element KarbonHQ’s API exposes—customer information, all task and project objects, plus the full communication history—and land it in Snowflake in a query-ready schema.
Once the data is in Snowflake, Claude has to “see” it immediately. Whether you choose Snowpipe, Kafka, Fivetran, or a custom Python microservice is up to you, as long as latency stays low enough to feel real-time from the user’s perspective.
Deliverables
• Automated connection to KarbonHQ that captures customer records, task & project data, and communication threads the moment they change
• Snowflake tables (or views) organised for straightforward analysis and semantic consistency
• Secure pathway, credentials management, and role-based access so Claude can query without exposure of sensitive keys
• Written hand-off notes: setup steps, object map, and a quick checklist to verify the sync is running correctly
Acceptance criteria
– A row inserted or updated in KarbonHQ appears in a Snowflake SELECT within seconds
– All three data domains are accounted for and match source counts after initial back-fill
– Claude can execute an example analysis prompt and return accurate aggregated results drawn from Snowflake
I’m happy to test side-by-side as you progress, but the finished job is considered complete only when the latency, accuracy, and documentation standards above are met.