← All work

01

SeamlessApply

SaaS / Career Technology

Challenge

Job search at scale is still manual. Candidates hunt across four or more boards, retype the same screening answers, skip cover letters, and track everything in spreadsheets. Most people apply to five or ten roles a week when they need fifty-plus to get interviews.

Solution

ASSOSIATIX designed and built SeamlessApply—an end-to-end job application platform. Users define a Seam (a campaign with criteria and resume) once. The system scrapes matching roles from Indeed, LinkedIn, Workday, and Remote, uses GPT-4 to fill forms and screening questions, writes personalized cover emails, and applies—either on a daily schedule or in-browser through a Chrome extension.

Implementation

We owned the product end to end: campaign architecture, a 12-model data layer, 12+ application surfaces, 60+ APIs, a Manifest V3 Chrome extension, dual auth for web and extension, cron automation with per-campaign error isolation, and production logging on every operation. Resume PDFs land in Google Cloud Storage, are parsed with GPT-4, and feed every AI answer on that campaign.

Outcome

A production SaaS that collapses find → fill → write → apply into one system: multi-board scraping, reusable Q&A memory, auto-apply with a full audit trail, and an admin dashboard for operators.

What we built

Seam campaigns

Each Seam is a named search: title, location, radius, job type, experience, boards, remote flags, keywords, resume, exclusions, and an email template. Users can run several at once—for example frontend in a city and full-stack remote worldwide.

Scraping and results

Apify scrapers pull Indeed, LinkedIn, Workday, and Remote into one job table. The results surface polls for criteria changes, filters by type, company, location, date, and application status, and paginates a dense card grid.

AI form fill

The Chrome extension reads fields on any application page and asks the API to answer them from resume context. Parsing is defensive—direct JSON, then regex, then field-by-field—so a messy model reply never takes the flow down.

Indeed Q&A memory

Screening questions are matched on stable field IDs, not wording. The first answer is generated, reviewed, and saved. After that, the same field pre-fills instantly with no extra AI call.

Auto-apply and cron

Company emails are cached, then discovered. Cover templates substitute company, role, and name. Every send is deduplicated and logged. A daily cron walks enabled Seams, applies to new matches, isolates failures per campaign, and rate-limits outbound mail.

Auth, admin, observability

The web app uses NextAuth sessions; the extension uses JWT against the same APIs and user table. Operators get subscriber and match charts plus client management. Winston, performance timing, and error tracking sit on every critical path.

$./start-project.sh

READY TO BUILD
WHAT'S NEXT?

Tell us where you are and where you want to go. We'll help you choose the right path—storefront, system, product, or a focused growth sprint.