Amazon Lex
Conversational bots with the tech behind Alexa.
❓ What is it?
A service for building voice and text chatbots using intents (what the user wants), utterances (how they say it), and slots (the parameters to collect), with automatic speech recognition and language understanding built in.
💡 Why does it exist?
A large share of support conversations are routine and structured — check a balance, book a slot. Lex automates those flows deterministically, with fulfilment logic in Lambda, and escalates the rest to humans.
⏱️ When should you use it?
Use Lex for TASK-oriented bots with defined flows (ordering, booking, FAQ-with-actions) and IVR automation in Amazon Connect. For open-ended knowledge conversation, a Bedrock-powered assistant fits better; the two can be combined.
🗺️ Where does it fit?
At the conversation edge: web/mobile chat widgets, Connect phone flows, and messaging channels; each fulfilled intent typically invokes a Lambda function that does the real work.
🔌 How do you integrate it?
Define intents with sample utterances and slot types, wire a Lambda for validation/fulfilment, test in the console, and deploy bot aliases to channels; enable sentiment analysis if routing needs it.
🧩 Commonly integrated with
🎯 Exam angle (AIF-C01)
- Vocabulary check: intent / utterance / slot — the exam expects you to identify each in a scenario.
- Lex = structured task bots; Bedrock/Q = open-ended generative conversation. Pick by how bounded the dialogue is.