Project

Exempliphai
Automation in the Browser

A Chrome extension case study focused on reducing the friction of repetitive web forms with a reliable hybrid approach.

Overview

Exempliphai is a browser-extension project focused on a real, repetitive problem, filling out web forms over and over. Keith Azodeh built Exempliphai as an automation-first tool that keeps the user’s profile information structured and reusable, then applies that information to application pages in a consistent way. Where AI helps, it is used for interpretation and short responses. Where reliability matters, the logic stays deterministic.

This combination is important. Pure "AI autopilot" approaches tend to be unpredictable on complex forms. Pure "hard-coded autofill" approaches break when sites vary. Exempliphai is designed around a pragmatic middle ground, use stored profile data as the source of truth, use site-aware field mapping to fill reliably, and use AI only for the parts that are naturally fuzzy.

The problem

Job applications are a clear example of a form-heavy workflow. Many portals ask for the same core information repeatedly, contact details, work history, education fields, and short answers. The cost is not just time, it is attention. Repeating the same work increases errors and discourages consistent applications.

Keith Azodeh approached this problem as a workflow automation challenge. The question is not "can we fill fields," it is "can we fill fields in a way that stays understandable, debuggable, and respectful of user control."

Solution approach (hybrid automation)

1) Keep a structured profile

Exempliphai centers around a user profile and resume-based information. The concept is simple, enter or import your key details once, then reuse them. This is the stable base that deterministic form filling depends on.

2) Deterministic form filling on the page

On target pages, the extension uses content scripts to locate relevant inputs and apply the stored values. Done well, this includes handling common control types (text inputs, selects, date controls), and it includes safe fallbacks when a field cannot be confidently mapped.

3) AI only where it helps

Some fields are not just "a value," they are a small piece of writing. For those, AI can be useful when it is grounded in known information (the user’s resume and stored profile) and when the user remains in control. Exempliphai’s design keeps that boundary explicit, it does not treat generated text as the only truth.

Architecture at a glance

Exempliphai’s architecture is a typical modern extension layout, but with a deliberate separation of concerns:

Popup UI for user profile management and configuration.
Content scripts for page interaction and deterministic field filling.
AI integration used as a helper for interpretation and suggestion generation.

That separation makes the system easier to reason about. Keith Azodeh can adjust AI prompts and heuristics without rewriting the entire field mapping layer, and can expand site support incrementally.

Why this matters (transferable patterns)

Even if your problem is not job applications, the patterns in Exempliphai are broadly useful for automation projects:

Keep a structured data source of truth, then apply it repeatedly.
Use deterministic logic for reliability-critical steps.
Use AI as an assistant for interpretation, not as the entire system.
Design the UX so users can see what happened and correct it.

If you are considering a similar tool, Keith Azodeh can apply these same patterns to internal workflows, customer support operations, or any process built around repetitive web interfaces.

Explore adjacent work by Keith Azodeh:

Exempliphai Automation in the Browser