TwinMail (side project): AI-assisted email workflows

Responsibility scope

Built this on my own under Twindevs as a personal side project, not a paid client product.
Explored ingestion, retrieval, and workflow patterns for large mailbox archives.
Used it to keep my automation and AI skills sharp while focused on my job search.

Stakeholders

Me, as the builder learning and experimenting
Anyone curious about how I approach AI and automation outside work

Decision points

Keep ingestion tied to source archives rather than opaque derived-only stores.
Keep AI features grounded so search and triage stay traceable to source messages.
Treat it as a learning project, kept secondary to my IT job search.

Problem and constraints

Personal side project to explore how an AI-assisted email tool could process large offline mailbox archives while keeping retrieval grounded and trustworthy.

Architecture approach

Background processing for categorization, prioritization, summarization, and follow-up preparation.
Offline .mbox ingestion for large mailbox archives feeding semantic retrieval.
Hybrid search and feature-gated capabilities across Google and Microsoft mailbox patterns.

Outcomes

A working personal exploration of grounded AI email workflows.
Hands-on practice with ingestion and retrieval patterns for large datasets.
A way to keep my automation and AI skills current between IT roles.

Next iteration

Keep iterating as time allows; it stays a side project behind my IT job search.

TwinMail (side project): AI-assisted email workflows

Need a quick answer from me?