Task automation and AI-powered workflows are taking a big step forward. As a result, Opal has launched a new feature called the agent step, changing how we interact with AI models. This update is more than a minor improvement—in fact, it shifts static and linear processes into dynamic, interactive experiences.
Takeaways 🚀
✅Transform workflows into interactive experiences
✅Automate tasks with agent intelligence
✅Personalize outputs through dynamic routing
✅Enhance collaboration with interactive chat
✅Boost efficiency using memory features
In this article, we will explore how to build agentic workflows in Opal. Specifically, we will discuss how this new feature works and outline the benefits it brings to users who want to automate complex tasks while also keeping creative and technical control. Furthermore, we will show you how these changes make AI-powered work both easier and more adaptable.
From Static to Interactive: The Evolution of Opal
Previously, creating a workflow in Opal meant setting up a series of fixed steps. You would pick a specific model to handle each task. For example, if you wanted to generate a story, you first had to decide how many pages it would have and what questions the user needed to answer. As a result, this process worked, but it often felt rigid and limiting.
Now, the agent step changes this process. Instead of picking a fixed model, you can add an “agent” during generation. Importantly, this agent doesn’t just follow instructions; it understands your goal and actively works to achieve it.
How does the agent step work?
The heart of this innovation lies in its proactive approach. Specifically, the agent examines your objective and then figures out the best path to achieve it on its own. To do this, it:
- Objective comprehension: It analyzes what you want to achieve, not just what command you entered.
- Tool selection: It decides which models or tools (such as web search or video generators like Veo) are needed to complete the task.
- Dynamic execution: Instead of following a pre-set script step by step, it adapts based on the information it receives or generates.
Use Case: The Interior Designer Example
Now, let’s look at a practical example: interior design.
In the previous version of Opal, the “Interior Designer” worked like a one-way street. First, you would upload a photo of an empty room, then choose a style (like “mid-century modern”), and finally get back a static image. However, if you didn’t like the result, the process stopped there or you had to start over from the beginning.
With the new agent step, you now work together with Opal—much like teaming up with a real design partner. Consequently, this collaborative approach transforms the process into a more engaging and personalized experience.
- Initial interaction: You upload the photo and describe your vision.
- Proposal and feedback: The agent generates an initial concept.
- Iteration: If something isn’t quite right, you can provide specific feedback on elements such as colors, furniture, or layout.
- Refinement: The agent uses this feedback to refine its understanding of your style, and can even research specific design substyles to deliver a result that feels unique and personalized, far from generic templates.
Because of this seamless interaction, the agent doesn’t just generate an image. Instead, it focuses on fulfilling your unique design vision every step of the way. As a result, it takes action and selects the right tools needed to help you achieve your goals.
New Tools to Empower Your Agents
To enhance the agent’s versatility and effectiveness, Opal now offers three new features that work together to create a more seamless experience.
1. Long-Term Memory
AI often forgets information between sessions. However, Opal solves this by giving agents memory. Now, Opals remember important details like:
- The user’s name.
- Style or brand preferences.
- Ongoing task or shopping lists.
As you continue to use Opal, it gradually becomes even smarter and more helpful. For instance, in a video idea generator, the agent remembers your brand identity over time. Consequently, it creates new ideas that match your company’s style, so you don’t have to provide the same information repeatedly.
- The user’s name.
- Style or brand preferences.
- Ongoing task or shopping lists.
As you continue using Opal, it gradually becomes smarter and more helpful. For instance, imagine you are working with a video idea generator agent. Instead of requiring you to repeat your brand guidelines every session, the agent now remembers your brand identity over time. As a result, it can create ideas that consistently match your company’s style without extra effort on your part.
2. Dynamic Routing
Nevertheless, control is still key. With dynamic routing, you can easily set up different paths for the agent to follow based on your own rules. This allows you to guide your workflow in a way that best fits your needs while giving the agent flexibility to respond intelligently to various situations.
For example, imagine you set up an Opal to prepare executive briefings. With dynamic routing, the agent first checks whether your meeting is with a new or an existing client:
- New client: The agent automatically searches the web for company background and recent news.
- Existing client: The agent reviews internal meeting notes to provide relevant context.
Therefore, the agent moves to the right step as soon as it meets the rules you set.
3. Interactive Chat
Sometimes, you might not provide enough information at the start. In these cases, the agent step can now initiate a chat with you, asking for any missing details or clarifying questions before moving forward. This makes the process smoother and helps avoid confusion along the way.
Let’s go back to the interior design example. Suppose you upload a photo but forget to mention a style. In this situation, the agent will not simply guess. Instead, it will ask you directly: “Do you prefer a minimalist style or something more rustic?” By doing so, the agent ensures you receive results that truly match your preferences and needs.
Ultimately, agentic workflows in Opal allow people and AI to work together more naturally. By combining an autonomous agent that understands your goals with customizable steps that you control, you open up new possibilities for everyone. Whether you’re a beginner or an experienced developer, this flexible approach bridges the gap between creativity and technical precision.
Moreover, new users will find it easier to get started, since the agent can fix mistakes and make the process smoother on its own. At the same time, advanced users benefit from mixing precise, fixed steps with flexible agent actions. As a result, this combination lets you build complex prototypes and dynamic stories that you couldn’t automate before.
As these tools become part of daily workflows, we will begin to see a shift—not just from simple “content generation,” but toward genuine “creative collaboration” between people and AI agents. In turn, this evolution opens up even more opportunities for innovation and teamwork in every area where Opal is used.
