Profile memory
Dheeraj NaguruEducation, experience, reusable answers, and long-form context stay in one source of truth.Turn every jobform into a guidedhigh-velocity flow.
Loopboy combines a protected Next.js backend, structured candidate memory, and a Chrome extension that understands the current screen. The result feels less like autofill and more like an application control room.
UI modeMotion-first landing
Extension modelMV3 + protected API
Autofill postureConservative by design
Live targeting
Current field, full page, or selectionMove from broad autofill to exact field control without leaving the page.Protected planning
Server-side GeminiYour API key stays off the extension while unresolved questions still get high-context answers.Autofill posture
ConservativeManual on low-confidence fields.Prompt-aware answers
Per-user memory
Exact field selection
Human review
Current-page autofill
Prompt-aware narrative answers
Per-user profile memory
Protected Gemini planning
Selection mode for exact fields
Chrome extension + Next.js backend
Current-page autofill
Prompt-aware narrative answers
Per-user profile memory
Protected Gemini planning
Selection mode for exact fields
Chrome extension + Next.js backend
Product Surface
Designed like a control surface instead of another startup template.
The page leans into hierarchy, atmosphere, and motion. It should feel intentional before a user ever opens the extension.
Product direction
A control room, not a brochure
Loopboy keeps the important parts visible at once: candidate memory, targeted fill actions, and the exact state of the current application screen.
LLM layer
Prompt-aware answers
Open-ended fields can be drafted against the visible job description, your structured profile, and long-form context.
Safety
Human-in-loop by default
Low-confidence and high-risk fields stay manual. You keep final control while the system removes the repetitive work.
Architecture
Built for your own stack
Profile storage, auth, and fill-planning live in your own Next.js app, so the extension stays thin and the key material stays server-side.
Execution Flow
Every step stays visible, from page capture to exact-field application.
This isn’t framed as “AI magic.” The workflow is explicit, inspectable, and designed around speed with operator control.
Capture the surface
The extension scans the visible form, labels, options, and relevant job-page context on the current screen.
Resolve the obvious
Deterministic heuristics map direct inputs like identity, compensation, education, work authorization, and repetitive screening answers.
Escalate only when needed
GSAP drives the motion layer on the homepage. Gemini only handles the unresolved or narrative fields in the product workflow.
Apply with precision
Fill the whole page, the current field, or a hand-picked selection. The interface stays fast even when the form is messy.
Why it feels different
Loopboy is built like a product you’d want to use daily, not a demo you forget in a week.
The homepage now mirrors the product thesis: precise control, visible state, and motion that explains the system instead of decorating it.
Selection modePick exact fields like a visual tool, then fill only that set.
Long-form memoryResume text, cover-letter guidance, and custom context stay available to the planner.
Own the backendAuthentication, storage, and planning remain in your own Next.js stack.
Ready to tune it