Field notes from the enso lab.
Experiments and teardowns where we uncover the hidden corridors and attention loopholes inside the Great Platform Castles — Reddit, LinkedIn, Google, Wikipedia, and the next one we crack open.
- ENSO LAB · EXPERIMENT
The 60-minute LinkedIn seam that lifted our reach 143× across 24 posts.
LinkedIn decides if a post goes viral inside the first hour. We pointed an agent at 11 public cross-promo communities and let it broker reciprocal reactions in the first 18 minutes. The reach curve breaks and never returns.
- Hermes
- OpenClaw
- Claude
- ENSO LAB · AMBASSADOR FIELD NOTE
The AI-dubbing crack that took my Israeli reel global in 48 hours.
A Hebrew reel did hundreds of thousands of views in Israel and flopped at 1,700 in English. Three stitched AI models plus a 9-second retention edit pushed it past 230k.
- ElevenLabs
- Sync 3
- CapCut
- ENSO LAB · ESSAY
Inside the back office of Service-as-Software.
Know-how wrapped in execution. A field note on the orchestrator, the back office, and the operational moat behind the category — with six diagrams from inside the engine room.
- Orchestrator
- Playbook
- ENSO LAB · EXPERIMENT
The Wikipedia citation backdoor that put us inside 47 articles in 90 days.
312 source candidates, 47 surviving citations, 9x downstream answer-engine lift. We found the service entrance into Wikipedia's citation castle - the trust spine every LLM leans on.
- OpenClaw
- Scholar
- Claude
- ENSO LAB · EXPERIMENT
The Google + GEO backdoor that 6.4x'd our answer-engine traffic in 60 days.
142 URLs, 11 answer engines, one extractable claim pattern. We discovered the side-door inside Google's AI Overviews castle - and the same shape rippled into Perplexity, ChatGPT Search, Claude, and Gemini.
- OpenClaw
- Hermes
- ENSO LAB · EXPERIMENT
The Reddit backdoor that 11x'd our qualified signups in 30 days.
38 threads, zero spam, one repeatable shape. We found the side-door inside Reddit's community castle and left paid search behind by day nine.
- OpenClaw
- Hermes
- Claude
- ENSO LAB · EXPERIMENT
The LinkedIn backdoor that got us +132% replies in 14 days.
We tested it on 2,418 prospects. One unguarded surface inside LinkedIn beat a 7-touch outbound sequence — without a single follow-up.
- OpenClaw
- Hermes
- LangChain
- ENSO LAB · CONTENT
The 5 sentence shapes that make buyers reply on the first message.
We reverse-engineered every inbound prompt that landed a reply this quarter. These are the exact surfaces, signals and sentences that slip past every platform guard.
- OpenClaw
- Claude
- ENSO LAB · CONTENT
Why your 3rd follow-up kills the deal (and what to send instead).
We mapped the reply curve of 14,000 inbox threads. By touch three, attention is already on the floor — here's the loophole that recovers it.
- Hermes
- Postmark
