community & growth systems for AI/web3 | content, incentives, automation
Phuket, Thailand
Joined April 2021
- Tweets2.5K
- Following704
- Followers1.5K
- Likes38.8K
Pinned Tweet
One of my @discord apps has crossed 10k users.
For me, it’s another proof point for a simple model:
- content creates attention
- community creates trust
- product utility turns both into adoption.
Without the last part, even strong distribution ends as a temporary spike.
The best community-led growth doesn’t feel like marketing.
It feels like “we need this here”
I got tired of choosing between data privacy and AI speed when working with sensitive docs.
So I switched to @usewisp_io
Then I did something I never expected to do as a non-developer: I built an iPhone app for it.
Full story 🧵👇
I built this for my own workflow. Now it's open source.
Try Wisp, there's a free plan: usewisp.io
Want the iPhone 👉 Mac setup from the video?
Code + install guide: github.com/web3illy/unoffici…
1. sign up here: platform.claude.com/offers/8…
2. submit the request form
3. get your $100 in Claude API credits within 1-2 days
Most CTA tests are fake tests.
Changing “Try now” to “Get started” will not fix a journey that asks users to leave the feed, trust the product, complete setup, and wait for value.
The bottleneck is rarely the wording.
It is the cost around the action.
Map that cost before testing the copy👇
For an AI product, this is not a useful test:
A: "Try now"
B: "Get started"
This is:
A: CTA 👉 signup 👉 empty chat
B: CTA 👉 pre-filled use case 👉 one-click run 👉 signup after the first result
Primary metric:
users who reach their first useful output.
Now the test changes the journey, not the vocabulary.
The best sybil farmers don't break your campaign rules.
They optimize them.
They find the cheapest action that qualifies, automate it, and scale it across accounts.
If that path creates no value, better detection won't fix the campaign.
The reward rule is the exploit 👇
The app itself is simple. It automates a few community actions and adds small mechanics that make participation more engaging.
Nothing revolutionary. Just useful community infrastructure doing its job.
The easiest sybil farm to miss is the one where every account looks legitimate.
The evidence sits between accounts:
- shared funding
- repeated sequences
- referral overlap
- reused content
One match proves little.
Recurrence across independent layers is what matters 👇
