# daanalytics Published articles for daanalytics. This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles. ## From Prompt to Query: Real Cost of Agents on Snowflake DevFeed: [From Prompt to Query: Real Cost of Agents on Snowflake]() Original publisher: [Read original article]() Author: Daan Bakboord Published: 2026-10-14T14:01:06Z Content type: article Language: en Sources: [Snowflake Builders]() Topics: [LLM observability](), [AI Infrastructure](), [AI Architecture](), [Multi-step AI processes]() Tags: [2026](), [account](), [agents](), [ai](), [ai-agent](), [analyst](), [bi](), [budget](), [chargeback](), [cloud-services](), [compute](), [cortex](), [cost](), [daanalytics](), [data-engineering](), [data-team](), [functions](), [index](), [indexes](), [inference](), [prompt](), [query](), [snowflake]() ### AI overview Agent spend on Snowflake spans model inference, Cortex Search, AI functions, warehouse compute, storage and cloud services, while query cost is difficult to tie to a prompt because telemetry views lack a shared key and may be delayed. The article recommends reconstructing sessions with user and time, query hashes, and durable query tags on every agent step, then copying usage data into owned tables. Track cost alongside latency, failures, plan steps, tool calls and queries per session; trace delivery may under-report cost, so use Snowflake usage views for credits and tokens. To control spend, set token caps and warehouse alerts, route questions by workload, and allocate budgets to teams while preserving an allowance for power users. In development, adaptive sizing and model selection help reveal workload patterns; in production, pin them to meet the chosen SLA. Version agents, canary releases where backend routing allows, and review cost and quality metrics weekly. A strong semantic layer and user training also matter: underspecified questions and missing prompt records make attribution and reliable answers harder. ### Source excerpt On 1 October 2026, the Snowflake User Group Netherlands met at the Snowflake office in Amsterdam. Amir Peres, CTO and co-founder of Yuki, gave the session From Prompt to Query: Understanding the Real Cost of Agents on Snowflake. A dashboard runs the same query every morning. An agent takes a different path every run. One prompt becomes spend in Cortex inference, Cortex Search, AI functions, and the warehouses that eventually run the SQL. Snowflake still has no single view that says what that interaction cost, or who was behind it. This is what stayed with the room. Why the old chargeback model breaks BI and pipelines are forecastable. You know the workload, the schedule, and roughly which warehouse size fits. An agent on the same account is a different product. You do not know the question, how hard it will be, or whether the person asking it would have written that SQL. Amir walked through one prompt as a chain: An orchestration model plans the work, calls tools, and often replans. Cortex Analyst turns the question into SQL through the semantic layer, and that SQL runs on a warehouse. Cortex Search serves retrieval from its own index. AI functions add inference on top of the scan. A single query that calls something like AI_COMPLETE once per row, across hundreds of thousands of rows, is enough to blow a budget. Under all of that sit warehouse compute, search indexes, storage, cloud services, and cross-region inference. Ask what Cortex costs, and the honest answer includes every line. Tokens are the part everyone talks about. Compute is where an agent gets expensive. Tokens are also the part a data team has the least leverage on: you can switch models and tighten prompts, and Snowflake's own pricing will keep moving. The warehouse, the semantic layer, and the limits in front of users are where you can still intervene. The same prompt is not deterministic. Run it again and the plan can change. Ship a new model or a new semantic layer and the SQL changes with it. A la