Video 4/10: Building AI Agent with CrewAI, Building an Agent with Dynamic Input

In this video you will learn how to run a single CrewAI agent across multiple dynamic inputs in one execution using kickoff_for_each — and have it generate a separate output file for each input automatically.
This is Video 4 of the Building AI Agents with CrewAI series. Instead of hardcoding a task for one specific subject, you will use input interpolation with {placeholder} syntax inside your task description and expected_output fields, then pass a list of inputs to kickoff_for_each. CrewAI handles the iteration, runtime variable interpolation, and file output naming automatically.
One crew definition. Multiple inputs. Multiple deliverables.
——————————————————————————
🧠 What you'll learn
——————————————————————————
→ How to use {placeholder} input interpolation in task descriptions
→ How kickoff_for_each runs your crew over a list of inputs
→ How dynamic output_file naming works per input
→ How to scale a single agent workflow into a batch processing pipeline
——————————————————————————
📦 Stack
——————————————————————————
→ CrewAI
→ Gemini 2.5 Flash (via LiteLLM)
→ python-dotenv
——————————————————————————
🔗 Links
——————————————————————————
→ CrewAI Docs: https://docs.crewai.com
→ Video 1: https://youtu.be/brexE2qNQXo?si=CYI3NpvA0WxFacR8
→ Video 2: https://youtu.be/2kXhBxmjA_Y?si=tXTJjTKyqlxeNjcK
→ Video 3: https://youtu.be/WNmz0qUJ4UM?si=bQW-AQvqduu1FpiU

#CrewAI #AIAgents #Gemini #Python #LLM #AIEngineering #AyonaireAcademy #BuildInPublic #GenerativeAI

Видео Video 4/10: Building AI Agent with CrewAI, Building an Agent with Dynamic Input канала AI Research with Faruq