Agentic Systems · Live
An automated interview prep system that turns a calendar event into a structured prep document — with zero manual effort.
I designed, built, and use this system myself.
Senior professionals in active job searches are interviewing for multiple roles across different functions simultaneously. Each interview requires tailored preparation: company research, relevant stories, smart questions. That prep takes 1-2 hours per interview and gets skipped entirely when schedules get tight. The result: candidates show up underprepared for roles they are genuinely qualified for, not because they lack the experience, but because the prep system does not exist.
Job seekers navigating multi-track job searches across different role families. People who apply to adjacent roles tapping into their cross-skills or exploring a career pivot need prep that adapts to the role, not a one-size-fits-all script.
Interviewing across multiple role families means the STAR scenarios, questions, and positioning need to shift with every conversation. Manually rebuilding that context before each call can take too long and may happen inconsistently. The design constraint I set: the system should require zero effort once it is running. Tag a calendar event, forward a job description, and the prep arrives in your inbox before the meeting. No dashboards to check, no prompts to write each time.
That constraint shaped the architecture: a calendar trigger for the baseline prep, a separate email parser intake for job description context, and Claude doing the synthesis between them.
Claude is used at one specific moment: synthesizing career context, STAR scenarios, and meeting details into a structured prep document. It is not a chatbot and it is not answering general questions.
The system prompt contains a condensed version of 20 years of career context: key roles, metrics, 10 STAR scenario themes with story summaries, and writing rules. Claude selects the three most relevant scenarios based on the meeting details and generates a prep document in a direct, confident voice with no filler.
The result is prep that sounds like the candidate, not like AI output, because the source material is specific and the instructions are tight.
No custom code. No servers. No dashboards to maintain.
Post-meeting reflection loop: a third Zap that fires after the meeting end time and prompts a quick reflection capture; what went well, what to improve, what to follow up on.
Prep quality rating: a simple one-click rating in the delivery email so the system learns which prep formats are most useful over time.
Role family tagging: structured JD intake that tags the role family and routes to a role-specific system prompt variant for more precise scenario selection.
Agentic systems work best when the trigger is something you are already doing. Tagging a calendar event and forwarding an email are zero-friction actions. The automation does the rest.
Prompt quality is the product. The output is only as good as the career context and scenario library built into the system prompt. Investing in that content upfront is what separates a useful prep doc from a generic one.
Two Zaps beat one complex Zap. Separating the calendar trigger from the JD intake kept both pipelines simple, testable, and easy to modify independently.
Ready to see it in action?
No signup. No friction. Jump in and explore the app yourself.