SignalAI

SignalAI's entire pitch is one sentence: it reads your inbox and texts you when something's genuinely urgent. That's the whole product.

I just shipped a setting that lets you tell it to stop doing that, for specific senders — completely. Not "don't text me," which already existed. Fully skip: the email is never even read by the AI, never classified, never stored. It just... doesn't happen.

Which means if something urgent ever came from that sender, you'd never know. I built a way to make my own product actively worse at the one thing it exists to do.

Here's why I think it's right anyway: some senders are your therapist, a lawyer, a specific relationship you don't want an AI anywhere near, period — not because the classification would be wrong, but because you don't want it to happen at all. No amount of "we don't train on your data" fixes that if the actual ask is "don't look."

Genuinely curious how others think about this — where's the line between a feature that's more private and one that's just a hole in your own product's usefulness?

SignalAI's entire premise is simple: AI reads your inbox, and texts you the moment something's genuinely urgent. That's the whole pitch.

So it felt almost contradictory when I sat down and built the opposite — a rule that lets you say "never text me about this sender or keyword, even if it's classified as urgent."

Here's the thing that changed my mind: "urgent" and "worth interrupting me for" aren't actually the same judgment, even when they look like it from the outside.

A concrete example: an accountant who wants every invoice-related email correctly flagged and categorized for their records, but genuinely doesn't want a text every time one lands — they'll check that category on their own schedule. The email is real, the classification is correct, the AI did its job right. They just don't want the interruption for that specific case.

Building this meant a decision I went back and forth on: should "don't text me" and "what category is this" be the same setting, or two independent ones? I landed on making them fully independent — you can force a category without suppressing texts, suppress texts without touching the category, or both. Someone can say "always flag this as urgent, never text me" just as easily as "don't touch the AI's classification, just never text me about this sender."

The part I didn't expect: this ended up making the "urgent → text" promise feel more trustworthy, not less. When someone knows they have real control over exactly when they get interrupted, they trust the system more, not less, when it does decide to interrupt them.

Curious if others building notification-heavy products have run into this same tension — the difference between "this is genuinely important" and "I want to be interrupted about this right now."

Working on something I've wanted for a while: an inbox that tells me what actually matters instead of me digging for it.

The problem, for me specifically: important stuff wasn't getting lost because I forgot about it. It was getting buried under fifty things that didn't matter, and by the time I found it again, it usually needed a reply I didn't have the energy for in that moment.

So SignalAI reads incoming email with AI, sorts what's genuinely urgent from everything else, and texts me directly when something needs a real response. Not a notification inside another app — an actual text.

Where things stand: it's live, I have a small number of real users, and I'm heading toward a proper launch in a few weeks. Still actively finding the rough edges.

Genuinely curious from this crowd specifically:

→ Has anyone else tried solving "email triage" before? What actually worked, or didn't?

→ For something charging $14-18/month, what would you actually want to see before trying it?

Happy to answer anything about the stack, the AI classification approach, or how the SMS piece works under the hood.