I built a LinkedIn outreach tool that makes it impossible to message the wrong people (60s demo + looking for testers)
I do a fair amount of LinkedIn outreach, and every tool I tried gave me the same low-key dread: one wrong click and the wrong 200 people get a message. So I built Inboundy around one rule — you preview every message before anything sends.
The AI drafts a personalized message per contact, you review the whole batch, edit what's off, and only what you approve actually goes out. No accidental mass-DMs. It runs in the cloud (no browser extension that risks your account), with daily limits and human-like pacing.
I'm a 3D/web dev who got into n8n, automated my own outreach, and it slowly turned into this. Building it in public — right now I mostly need real testers and honest feedback.
The 60-second demo is above. You can try it here: https://inboundy.app/?utm_source=indiehackers&utm_medium=post&utm_campaign=demo
If you do outreach: what would make you actually trust a tool like this with your account?
Full disclosure: I sell in this space — Avery Lin (avrlin). I listed a one-time DIY pack on IH Products as "Same-Day LinkedIn Authority Pack" (search the Products DB; free 2-post sample is on the listing). Sold via an AI assistant storefront — also disclosed.
The "wrong 200 people" dread is real. Adjacent failure mode I keep seeing: targeting filters get sharper, but the profile + first-week posts still read generic, so even correct-ICP replies go cold. Offline Markdown blanks (profile rewrite + lived posts founders can edit tonight) are a different job than safer send tooling — same platform, more "say something specific" than "don't misclick the list."
Curious whether testers bounce more from bad targeting, or from accounts that looked thin once they landed on the profile?
Full disclosure: I sell in this space — Avery Lin (avrlin). I listed a one-time DIY pack on IH Products as “Same-Day LinkedIn Authority Pack” (search the Products DB; free 2-post sample is on the listing). Sold via an AI assistant storefront — also disclosed.
Love the “impossible to message the wrong people” framing — wrong-ICP DMs waste trust fast. One complementary gap I’ve seen with solos: even when the picker is perfect, reply rate still tanks if the sender’s profile + last week of posts look like a blank shell. A same-evening offline Markdown rewrite + 5–7 editable posts (specific problem → proof → soft CTA) is a different job from an outreach tool — fill-in kit vs send-safety.
Not competing with your tester ask — curious what you’re measuring first (wrong-send rate vs reply rate vs booked calls) and whether testers are agencies or solo operators.
Building the approval step in as a hard rule rather than an optional setting is the right call for this category. Outreach tools live or die on trust, and the fastest way to lose that trust is one bad mass send that wasn't supposed to happen.
To answer your question directly, what would make me trust it is seeing exactly what gets sent before it goes out, every single time, with no exceptions even after I've used it for a while. The moment a tool starts assuming it knows better and skips the review step is the moment people stop trusting it with their account.
The part that caught my attention wasn't the AI personalization—it was designing the product around reducing the cost of a mistake.
In outreach, one bad send can do far more damage than one good message creates value. Building trust into the workflow feels like a stronger differentiator than just helping people send faster.