I want to build an AI agent that can pose the same research question to at least five different large-language-model APIs (e.g. GPT-4, Claude, Cohere, Palm, Llama-2 via Hugging Face) and capture every answer in real time inside a single Google Sheet. The workflow should look like this: 1. I type or paste a prompt once. 2. The agent forwards that prompt concurrently to each chosen LLM. 3. As soon as each model replies, the agent writes a new row to my Google Sheet with these columns: timestamp, prompt, model name, full response, and any cost/latency data you can easily expose. 4. When all models have answered, the sheet shows a complete, side-by-side record that I can analyse for research purposes. Language and stack are flexible; pick whatever lets you move fast (Python, JavaScript or another option you prefer) as long as setup is straightforward and credentials remain secure. Deliverables: • Clean, well-commented source code • Setup guide for API keys, Sheet auth and local/cloud deployment • A demo Google Sheet pre-configured with the required columns • Brief read-me on how to add or swap LLM endpoints in the future Acceptance criteria: running a single command should open a prompt box (CLI or simple web form), send the question to at least five live LLMs, and populate the sheet exactly once per model with no manual intervention.