I built an AI data extraction engine and E2E encrypted SMS router in Rust
TL;DR
SendStackr is a self-built Rust backend for email ingestion and SMS routing, born from managing ~50 physical Android devices at a digital agency.
Key Points
- Custom SMTP and IMAP servers replace standard APIs, avoiding timeouts and no-code wrapper limitations when handling heavy unstructured data.
- Graph RAG extracts specific variables from unstructured PDFs, raw HTML emails, and webhooks, returning strictly mapped JSON.
- An Android Node app handles the physical devices and routes inbound SMS (mostly 2FA/OTPs) with end-to-end encryption.
- The core problem solved: centralized inbox and SMS forwarding for multi-account operations without expensive or webhook-heavy SaaS solutions.
Nauti's Take
Choosing Rust to build custom SMTP and IMAP servers from scratch is an opinionated call – but a defensible one when throughput and memory safety matter at scale. Using Graph RAG instead of brittle regex pipelines for unstructured email and PDF extraction shows genuine engineering judgement.
The missing piece: concrete accuracy benchmarks – 'strictly mapped JSON' sounds clean, but real-world agency email is messy and adversarial. Still, this is exactly the kind of infrastructure tool that fills a gap most SaaS vendors either ignore or monetize aggressively, and it deserves more visibility in the indie dev space.
Context
Anyone running multi-account agency operations knows the pain: 2FA SMS scattered across devices, PDFs that resist automation, and no-code tools that fall apart under load. SendStackr targets exactly this gap by combining Rust performance, Graph RAG, and physical device management in one stack. This is not a side-project toy – it was built under production pressure to replace expensive, fragmented tooling.
The pairing of LLM-driven extraction with E2E-encrypted SMS routing in a single system is architecturally notable and practically rare.
Sources