Back to index
OtherAbout 2 hours plus a challenge lab; longer contact centre learning path available·Free with a Google Skills account for many learners; labs can require credits or a subscription

Create Conversational Agents with Stateful Flows (Dialogflow CX) on Google Skills

3.8

Short, hands-on and done inside the real Google Cloud console, which is the best way to learn Dialogflow CX. The naming churn is exhausting, and the course teaches the product far better than it teaches conversation design.

What We Liked

  • Real console labs rather than slides
  • Experienced builders can skip straight to the challenge lab for the skill badge
  • Teaches flows, pages, intents and entities, which are still the backbone of predictable voice agents
  • Part of a wider contact centre path covering architecture and generative features

What Could Be Better

  • Dialogflow CX, Conversational Agents, Customer Engagement Suite: the naming changes faster than the course does
  • The old Dialogflow CX console is marked deprecated, so screenshots and third-party tutorials go stale quickly
  • Skill badges test that you completed a lab, not that you can design a good conversation
  • Little on measuring containment honestly or on when to hand off to a human

Detailed review

Google has renamed this product so many times that it is worth spelling out. Dialogflow CX is now presented as Conversational Agents, part of what Google calls Customer Engagement Suite, and the documentation itself warns that some products and features are in the process of being renamed. The old Dialogflow CX console is listed as a deprecated interface, and generative playbooks and flows are being merged into a single console. Cloud Skills Boost has also been renamed Google Skills.

If you search for any of these you will find tutorials that no longer match the screen in front of you. This course, Create Conversational Agents with Stateful Flows, is the one I would start with. It is intermediate, takes about two hours, and covers the pieces that make a voice agent predictable: flows, pages, intents, entities and state handlers. You work in the actual console. If you already know the product you can jump to the challenge lab and still earn the skill badge.

It is also tied into Google's Gemini Enterprise Agent Ready programme. There is a wider learning path for contact centre engineers and architects that adds generative fallbacks, data store agents and integration with telephony, and I would do the architecture course from that path next. The weakness is the one I see in most vendor conversational AI training. It teaches you how to build in the tool, not how to decide what the agent should do. There is almost nothing on writing prompts and responses that callers understand, on escalation to a person, or on how containment rates get gamed when a team is judged on them. Those are the questions that decide whether a contact centre bot annoys customers. Search interest for Dialogflow tutorials is tiny, around 10 a month in the US, and falling, which I read as people searching for the new names rather than losing interest in the product. My take: take it if your employer is on Google Cloud, and do it soon, because the next rename is probably coming.

Pair it with a design-focused course so you learn why as well as how.

[ final ]

The verdict.

The right starting point if your company runs its contact centre on Google Cloud. For general conversational AI design, Rasa's CALM course explains the thinking better.