PROJECT SUMMARY
Giving users control over an AI scheduling agent Blockit is designed to act like a personal scheduling assistant—handling the back-and-forth of coordinating meetings automatically by understanding your calendar, preferences, and constraints. The system learns over time and negotiates meeting times across calendars, messages, and email threads. But scheduling is highly contextual. Users often want the assistant to behave differently depending on the situation: scheduling with investors versus internal meetings, protecting deep-work time, or prioritizing urgent requests. The challenge was giving users fine-grained control over the agent’s behavior without turning the product into a complex rule-builder.
The Big PICture
Giving users control over how the scheduling agent behaves
Blockit operates as a scheduling agent that manages meetings according to a universal set of preferences—things like availability windows, meeting lengths, and calendar priorities. These defaults work well for most situations, but scheduling is inherently contextual. Users frequently need the assistant to behave differently depending on the meeting: prioritizing investors, protecting deep-work time, sending different follow-ups, or sharing different availability.
Historically, Blockit attempted to solve this through features like codewords and templates, which allowed users to override default behavior in specific scenarios. However, adoption was limited. Only a small percentage of meetings were scheduled using these tools, and many users didn’t fully understand how they worked. At the same time, we were introducing a new memory system that allowed the agent to adapt preferences dynamically over time.
The challenge was to rethink how users should influence the agent’s behavior across different situations—without forcing them to manage complicated rules or configuration.
Understanding the real user problems
Rather than starting with the existing features, I focused on identifying the core problems users were trying to solve.
Users wanted to:
- Trust that Blockit behaves correctly in specific scenarios
Certain meetings—like investor calls or internal planning sessions—carry different expectations for scheduling behavior. - Apply different priorities to different types of meetings
Not all meetings should be treated equally. Users needed a way to signal when something should override normal calendar preferences. - Control the agent’s actions for certain meeting types
This included things like sending specific scheduling links, using different meeting lengths, or applying different follow-up behaviors.
These needs revealed that the real problem wasn’t just “overriding preferences”—it was helping users express intent about how the agent should behave in different contexts.
final thoughts
Design principles
To guide the redesign, we established a set of principles for how customization should work within an AI-driven product.
1. Control without complexity
Users should be able to influence the agent’s behavior without building complicated automation rules.
2. Design around real scheduling scenarios
Customization should reflect how people actually think about meetings—investor calls, recruiting chats, internal syncs—rather than abstract settings.
3. Meet users where they are
Users shouldn’t have to preconfigure every possible scenario. The system should allow preferences to evolve naturally as people use the product.
4. Keep the agent in the loop
The product should remain agentic-first. Configuration should enhance the assistant’s behavior, not require users to constantly manage settings.
The solution: scenario-based templates
The redesigned system introduced scenario-based templates that allow users to define how the scheduling agent should behave in specific contexts.
Instead of configuring abstract rules, users create reusable templates that capture the behaviors they want for a given type of meeting—such as priority level, meeting length, links to share, or follow-up behavior.
Templates allow users to:
- signal priority differences between meetings
- define behavior patterns for recurring meeting types
- guide how the agent responds in specific situations
By grounding customization in real meeting scenarios, the system gives users meaningful control while keeping the experience approachable.
