AI Tool Development with Python, LangChain, and Google Cloud

I build AI-powered web tools that solve day-to-day workflow problems in digital marketing, content operations, and career decision-making. This case study focuses on practical tools I designed, built, and use in my own work, rather than practice projects or one-off demos.

To keep the interface fast and lightweight, the frontend is built with static HTML, CSS, and JavaScript. More complex AI processing is separated into a Python backend running on Google Cloud Functions. This architecture gives non-technical users a smooth experience while keeping API keys out of the browser.

View AI Tools

Background and Purpose

Although generative AI is highly flexible, using it for professional work often means repeatedly writing long prompts to define context, output rules, formatting requirements, and decision criteria. For recurring tasks such as content drafting, translation, copy adjustment, and user research, a purpose-built web tool with predefined inputs and structured outputs is much more efficient and reliable than starting from a blank chat window each time.

To streamline these workflows, I built dedicated web interfaces that let users run customized AI processes by filling in a few simple fields. Because the tools grew out of my own day-to-day needs, building them has also been a practical way to strengthen my problem-solving skills as a digital marketer through software development.

AI Tools I Developed

Below are the AI-powered tools I built to solve specific real-world tasks through focused AI workflows.

AI User Interview

Given a topic, this tool models pain points, behavior patterns, and decision-making criteria from the perspectives of multiple virtual personas. I use it during content planning, UX research, and requirements definition to develop stronger user hypotheses and uncover perspectives that might otherwise be missed.
» View the AI User Interview tool

AI Image Prompt Generator

This hub helps users create and organize high-quality prompts for image generation, with templates for use cases such as social media posts, blog headers, ad banners, and presentation slides. It guides users in refining composition, artistic style, color palette, and negative prompts before they run an image generator.
» View the AI Image Prompt Generator

Character Count

This utility quickly counts the characters in a block of text. It streamlines copy checks for character-limited formats such as SEO meta titles, meta descriptions, search ads, and social posts.
» View the Character Count tool

AI Character Count Adjustment

This AI-powered feature suggests shorter or rewritten versions of text to match specific length limits and use cases. It reduces the manual effort required to trim copy while keeping the final text natural and faithful to the original meaning.
» View the AI Character Count Adjustment tool

Job Change Decision Score

This diagnostic tool compares a user's current workplace with potential career opportunities and turns complex decisions into objective scores. By rating factors such as responsibilities, compensation, growth prospects, and work-life balance, it helps users evaluate career moves more rationally instead of acting on temporary anxiety or impulse.
» View the Job Change Decision Score tool

AI Writing

This tool brings translation, proofreading, and email drafting together into a single, unified workspace. It eliminates the need to construct new prompts for every task, allowing users to quickly write, refine, and translate professional copy between English and Japanese.
» View the AI Writing tool

Funny Writing Creator

This creative tool turns ordinary descriptions or explanations into humorous analogies, metaphors, and more engaging copy. It is useful for brainstorming sessions, social media posts, article hooks, or adding a lighthearted touch to internal presentations.
» View the Funny Writing Creator

AI User Interview: A Practical Use Case

The AI User Interview tool is one of my favorite examples because it streamlines the planning phase before any code is written. When working with AI coding assistants such as Codex or Antigravity, the quality and relevance of the generated code depend heavily on the user and product context provided upfront.

Before writing a technical specification, I use this tool to map out target users, especially their contexts, pain points, expectations, and decision-making drivers. Feeding this richer background into AI coding assistants helps them make smarter UI/UX and architectural decisions, rather than generating code from narrow instructions alone.

Technical Architecture

Each tool uses a decoupled architecture that separates the lightweight static frontend from the serverless API backend that handles AI processing. The UI is built with vanilla HTML, CSS, and JavaScript, while AI workflows and external API integrations are managed by a Python backend running on Google Cloud Functions.

• Static HTML/CSS/JavaScript: Provides a fast, responsive user experience and strong SEO discoverability, both of which are critical for web-based marketing tools.
• Python: Handles backend API calls, prompt formatting, and sensitive processing so that internal logic and proprietary prompts remain secure.
• LangChain & LangGraph: Orchestrates complex, multi-step workflows that require prompt chaining, conditional branching, and state management. This allows the backend to process user inputs dynamically, run intermediate steps, and format the final response in a single execution loop.
• Google Cloud Functions: Powers the AI backend in a serverless environment. This setup scales automatically during traffic spikes and keeps maintenance costs near zero during quieter periods.
• Gemini API: Serves as the primary LLM engine. Google Cloud Functions handles authentication so that API keys are never exposed to the client browser.

Future Plans

Rather than chasing vanity metrics, my focus is on building tools that address real operational needs, running them in production, and improving them based on actual usage. As I monitor search traffic, user engagement, and feedback, I plan to update this case study with concrete performance data and optimization examples.

When developing new AI tools, I will continue to let practical challenges in digital marketing, content creation, and career planning shape the roadmap, so every feature solves a real problem instead of simply following tech trends.

View AI Tools