I've been bringing one of my older projects back to life.
GrafiTalk is a local-first communication workbench for developers and freelancers.
The idea is simple:
Writing the code is often easier than writing the update for the client.
Instead of starting with ChatGPT every time, GrafiTalk turns your project context into structured communication like:
• Status Updates
• Client Updates
• Handover documents
• Debug Reports
• Weekly Summaries
The core engine is deterministic and works without AI.
AI is an optional layer that I'm currently planning. It won't replace the workflow—it will only help polish the generated text (grammar, tone, translation, shortening, etc.).
The roadmap I'm working on includes:
🌍 Translation into 30+ languages
🔊 Read Aloud with configurable language and voice
🤖 Optional local AI (Ollama/Qwen) for writing refinement
📄 More communication templates (Release Notes, Feature Summary, Meeting Summary, Bug Report...)
📦 Better export and workflow polishing
The philosophy stays the same:
GrafiTalk prepares. Humans send.
I'm currently auditing and polishing everything before the next public release.
I'd love to hear:
What part of communicating with clients or teammates do you find the most repetitive?
The distinction that stood out to me is making AI optional instead of foundational. A lot of products ask users to trust AI with the entire communication process, while your philosophy keeps ownership with the human and uses AI only to refine the message. "GrafiTalk prepares. Humans send." communicates that boundary surprisingly well.