How two founders built a 300,000-comment Instagram growth machine

Executive overview

Most app founders think about distribution as either organic content or paid ads. Nick and Ivan found a third channel: mass automated Instagram commenting at a scale no human team could replicate.

Using a custom Android ROM that gives each account its own unique device fingerprint, they run 12–24 Instagram accounts per phone, each posting up to 60 comments a day. The comments are AI-filtered for relevance and written to mimic real user voices — not spam.

The core insight: being upfront that you built the app works better than hiding it.

How the commenting funnel works

• Target trending or high-potential videos in the app's niche, not random content
• AI filters analyse each video and comment section to decide whether to engage or skip
• Branded accounts are set up as founder profiles, intern subpages, or user personas
• Comments answer real questions in the thread, tell a micro-story, or mention the app softly at the end
• Profile bio links directly to the App Store; some accounts use a mention to the brand's main page instead
• People who engage either reply (prompting a follow-up) or visit the profile and click the link in bio
• 15% profile-to-link click-through rate is a solid benchmark; best campaigns have hit 50–60%

Comment formats that convert

• Story format: personal anecdote that naturally ends with mentioning the app — creates curiosity without a hard pitch
• Founder angle: transparent "I co-founded this app" framing — counter-intuitively outperforms covert promotion
• Friend recommendation: "my buddy recommended this app, going to try it" — soft social proof
• Educational answer: responds to a question in the comments, adds value, then references the app
• Profile redirect: mentions a video on the account rather than the app directly — pulls people to browse the profile
• Timing matters: commenting early on a video with viral potential gets the comment to the top

The technical moat

• Custom Android ROM loads and unloads the full device ROM per account — each account behaves as a unique physical phone
• Human-mimicking input system fools platform detection at the button-press level
• Current fleet: 60 devices running 12–24 Instagram accounts each, up to 60 comments per account per day
• At 1,000 devices: ~60,000 active accounts, 1.5 million videos and 21 million comments per month
• Accounts are warmed up with harmless, non-promotional comments first — identical logic to cold email inbox warming
• Algorithm-detection failures are fixed by rotating angles entirely, not just tweaking copy

Parallels to cold email

• Each phone is like an inbox: spreading volume across many accounts avoids platform throttling
• Warm-up phase uses low-risk comments (no CTA) to build account trust before promotional content
• A small percentage of comments drive almost all results — same power-law as cold email and UGC video
• Continuous angle iteration required: a working format can stop performing in two weeks as Instagram updates its detection

The platform they are building

• Current model: agency service — they run campaigns for clients
• Roadmap: self-serve dashboard where anyone can manage a fleet of social agents
• Users will describe automations in plain English; AI builds the flow
• Community templates will surface proven angles; custom branding controls per agent
• Goal: one-click access to the same back-end infrastructure they currently use internally

What determines success on the platform

• Copywriting and storytelling skill remain the differentiator — the platform automates logistics, not creativity
• Comments must be context-dependent: they need to reflect the specific video and conversation thread
• AIDA or similar frameworks apply — hook, build curiosity, soft CTA
• Iteration speed is the real edge: test multiple angles in parallel, cut what stops working fast
• Reading widely (copywriting books, influence/psychology, YouTube) accelerates pattern recognition