What I Built

I built ContextBridge, a Chrome extension that helps you switch between AI models without losing the context of your work.

Ever hit a usage limit while working on a project with ChatGPT, only to realize that switching to another AI means explaining everything all over again? ContextBridge solves this by extracting your conversation and using Gemma 3 (4B) to turn it into a structured handoff containing project details, important decisions, constraints, current blockers, and next steps. You can then copy it into another AI and continue where you left off.

Built around open-weight AI running locally through Ollama, ContextBridge is designed for developers, freelancers, students, and researchers who rely on AI for ongoing work.

The business opportunity is simple: people who use AI professionally could pay for a Pro version with saved project context, longer conversation handling, reusable handoff templates, and project workspaces—helping them spend less time repeating themselves and more time getting things done.

Code

GitHub Repository:

Prathamesh2007-o7 / Context_Bridge

ContextBridge

Carry your working context between AI conversations.

ContextBridge is a Chrome Manifest V3 extension that captures a visible AI conversation, uses a local Ollama model such as Gemma 3 (4B) to create a structured continuation handoff, and lets you edit, copy, or export the result as Markdown.

Features

Capture visible conversation messages from supported ChatGPT, Claude, and Gemini page layouts (best-effort DOM extraction; sites can change their markup).
Review and edit extracted text before it is processed.
Generate a structured handoff covering project state, decisions, constraints, failed attempts, blockers, and next steps.
Run inference against Ollama on your own computer; no ContextBridge cloud backend or remote AI API is used.
Edit the final handoff, copy it to the clipboard, or export a .md file.
Keeps the current session in the popup only; Clear session removes the current text.

Requirements

Google Chrome or another Chromium browser that supports Manifest V3.
…

View on GitHub

How I Built It

I built ContextBridge as a Chrome extension using HTML, CSS, JavaScript, and Chrome's Manifest V3.

At its core is Gemma 3 (4B), an open-weight model running locally through Ollama. The extension captures the conversation from a supported AI website, lets users review the extracted messages, and sends the context to the local model to generate a structured handoff.

Gemma organizes the conversation into key information such as project state, important decisions, constraints, blockers, and next steps. Users can then edit the result, copy it, or export it as Markdown and continue their work with another AI model.

By using local inference instead of a paid cloud AI API, ContextBridge gives users more control over their conversation data while making open-weight AI useful for a practical, everyday problem.

Why Does Open Innovation Matter?

Open innovation made ContextBridge possible without depending on a paid, closed AI API.

By using Gemma 3 (4B) through Ollama, I could run the AI locally, experiment with prompt design, and generate conversation handoffs without sending chat content to a third-party AI inference service. It also means there are no per-request API charges for generating handoffs, although users still need compatible hardware to run the model.

More importantly, open-weight AI makes it easier for developers to build on existing models, customize their behavior, and create tools that solve real problems.

ContextBridge is my attempt to turn that openness into something practical: helping people move between AI tools without repeatedly explaining their work from scratch.

DEV Community

ContextBridge — Switch AI Models Without Losing Your Context