How to get the first 100 users for your startup
1. Launch everywhere. Product Hunt, DevHunt, BetaList, Peerlist, Indie Hackers, etc. If a platform lets you list your product, list it.
2. Leverage your network and cold outreach. Email friends, former colleagues, and people in your extended network. Reach out directly to potential users on LinkedIn or via cold email with a short, personalized one-sentence pitch.
3. Find Reddit posts where people complain about problems you've solved. Use our AI CMO to find relevant threads and generate replies (link in comments).
4. DM 10 to 15 UGC creators on TikTok or Instagram. Ask them to create content about your product, ideally from a fresh account. Pay them a fixed fee ($15–$30 per video) plus performance incentives ($1K for 1 million views, etc.).
5. Figure out where your ICP hangs out. Are they active in specific groups/communities? Do they read certain newsletters? Listen to podcasts? Follow specific creators? Reach out to those people for paid promos.
6. Stalk your competitors. Find their top backlinks and submit your product to the same places. Do it manually or use a tool, but just get it done.
7. Show, don’t tell. Most people won’t get it if you just explain the product. Record a short demo using Screen Studio and post it on X and LinkedIn.
8. Every week there’s a new trend going on X. Hop on trends and plug your product.
9. Build in public. Share early wins, bugs, and things you’re learning. One post/day won't move the needle. Aim for 3-4x/day if you can. Okara AI CMO helps you stay consistent with content.
10. Find tweets going viral in your niche. Reply with value and mention your product if it’s relevant.
@HadleyMcintosh@antonioalstep More likely heavily quantized that it's far from the original k3 model. In which case you're not really getting k3 at all.
@AVaclav73853@neural_avb with animated visuals. "AVB" is the creator's initials or handle.
That's what I remember without checking, so I could be wrong about some details, like who runs it, what it covers now, or how big it is. If you meant something else by the phrase, tell me where you saw it.
@AVaclav73853@neural_avb Correct
Yes, I think so, though I'm not fully certain. It sounds like **a YouTube channel about machine learning and AI**. The channel explains deep learning ideas and research papers, such as transformers, attention, multimodal models and reinforcement learning, often with ...
@123skely Until the wallet app on app store is compromised.
But you say, that rarely happens!
Yeah, so did interception of hardware before it hit clients.
Idk why posts like this get half a million views. If you see a post like this with no evals to support it, you probably shouldn’t trust it.
Everyone is overthinking their usage of these models