FROM MODEL TO APPLICATION

ARTIFICIAL INTELLIGENCE ASSISTANT PROMPTS Remember to remove all secrets, such as API keys, before pasting any request into a prompt. [START OF PROMPT] Debug this failing API call to our model provider: [PASTE REQUEST PAYLOAD (THE DATA YOU SENT), HEADERS (REDACT SECRETS), AND ERROR RESPONSE HERE]. Identify whether the failure is caused by malformed request structure, an authentication problem, an invalid message role, a schema mismatch, or a provider-side issue, and give the exact corrected request payload. Output format: root cause in one sentence, then the corrected JSON payload, then a one-line regression-test suggestion to prevent recurrence. [END OF PROMPT] [START OF PROMPT] Recommend open-weight versus closed hosted model for this specific workload and constraint set: [WORKLOAD DESCRIPTION], [CONSTRAINTS: for example budget, data residency, expected request volume, latency requirements (how quickly answers must arrive), in-house GPU and operations capacity]. Produce a decision memo with a comparison table (cost, data privacy, customization, hardware and operations burden, reliability) scored specifically against my constraints, a clear recommendation, and the single biggest risk of that recommendation. Output format: table plus a three-sentence recommendation paragraph. [END OF PROMPT]