Guildy.ai
A Gmail-based job search pipeline with AI interview prep.
Co-founder & Designer
2024 — Present
Solo design + AI-assisted development
At a Glance
Job seekers lose track of applications because the source of truth is scattered across email threads, calendar invites, and spreadsheets. Status updates from recruiters arrive in email and are easy to miss.
- -Built end-to-end as a solo designer using AI-assisted development tools
- -Currently in private beta with active job seekers
- -Stage inference accuracy above 85% with the two-signal threshold rule
- -Early users report the product replaced their manual spreadsheet tracking entirely
Overview
Guildy turns your Gmail inbox into an automated job search pipeline. It scans your email threads, detects job applications and recruiter conversations, and organizes everything into a clean pipeline with stage tracking. For each job, Guildy provides company intel, context-aware interview prep, and AI-generated practice questions tied to the specific role. I designed and built the entire product as a side project using AI-assisted development tools. The core idea is simple: job hunting is chaotic because everything important lives in email threads, calendars, and scattered notes. Guildy turns that mess into structure automatically.
My Role
- -Sole designer and co-founder responsible for all product decisions
- -Designed the full experience from Gmail connection through pipeline management and interview prep
- -Built the product using AI-assisted development tools (Cursor, v0, Claude)
- -Defined the stage inference logic and rules for reducing false positives
- -Ran the private beta and iterated based on user feedback
The Problem
Job seekers lose track of applications because the source of truth is scattered across email threads, calendar invites, and spreadsheets. Status updates from recruiters arrive in email and are easy to miss. Interview preparation happens ad hoc and is disconnected from the job context. Most people try to manage this with spreadsheets or job tracker apps, but those require constant manual entry. The tools that exist today are either too manual (Notion templates, Google Sheets) or too disconnected from where the action actually happens (email). Nobody had built a job pipeline that starts from the inbox and updates itself.
What I Learned from Users
Job seekers described their tracking system as "a spreadsheet I update when I remember to" and most were at least a week behind on status updates
The most stressful part of job hunting was not the interviews but losing track of where things stood across 10 to 20 active applications
Users wanted interview prep tied to the specific company and role, not generic question banks
People feared missing a recruiter email or forgetting to follow up more than they feared the interviews themselves
The Approach
I started by mapping the entire job search workflow from the moment a user applies to the moment they get an offer or rejection. Every signal in that workflow passes through email. Application confirmations, recruiter outreach, interview scheduling, feedback, offers, and rejections all arrive as email threads. That was the key insight. If we could reliably parse email threads, we could build a pipeline that updates itself. I designed Guildy to connect to Gmail via OAuth, scan threads for job-related signals, and create pipeline items automatically. Stage inference uses a hybrid approach: rules and keyword detection for common patterns, plus an LLM for edge cases. A key design decision was requiring two or more tracked signals in the same email to reduce false positives. The job details panel combines pipeline context with AI interview prep, showing company intel, next actions, and practice questions all in one place.
Key Design Decisions
Gmail-first pipeline with automatic stage inference
Instead of asking users to manually create entries, Guildy scans Gmail threads and creates pipeline items automatically. It extracts company names, role context, and current stage from email signals. The pipeline view shows all active applications with their inferred stage (recruiter screen, technical screen, onsite, offer, rejection). Users can always override stage assignments, but the goal is that most updates happen without any manual input. The desktop view shows the pipeline on the left and a rich detail panel on the right with company intel, email thread summaries, and context for the next action.
Guildy Desktop
Context-aware interview prep tied to each job
Each job in the pipeline has a details panel that goes beyond just tracking. It pulls in company intel (industry, size, location, recent news), shows the email thread summary, and surfaces the next action the user should take. Below that, Guildy generates interview prep specific to the company and role. This includes likely interview questions, suggested talking points, and company-specific research. The mobile view makes this accessible on the go so users can review prep right before an interview. The prep is not generic. It is tied to the actual job context Guildy has gathered from the email threads.
Guildy Mobile
Impact
- -Built end-to-end as a solo designer using AI-assisted development tools
- -Currently in private beta with active job seekers
- -Stage inference accuracy above 85% with the two-signal threshold rule
- -Early users report the product replaced their manual spreadsheet tracking entirely
What I Would Do Next
- -Add voice-based mock interview with scoring (rates answer quality and provides improvement tips)
- -Expand stage inference to handle edge cases like referral introductions and recruiter cold outreach
- -Build a token-based monetization model with a referral loop for earning additional usage
- -Add calendar integration so interview prep surfaces automatically before scheduled calls