Figure Me Out is easy to pitch: read your opponent's number and type it first. Smoke and binoculars change who can get that information. I think that's a useful product question: can you add depth without making the basic promise harder to explain?
This indie dev is making a game where you eliminate others by typing the number on their forehead
- Race to read their number first
- Throw smoke and flashbangs to blind them
- Read far away numbers with binoculars
Game is Figure Me Out
If Azlen adds segmentation to this demo, I'd want a search result to open the footage just before the object appears. Let me follow it through the scene. I think video search needs more ways to explore a recording, not just more ways to summarize it.
Continuation of my old video-as-3d-object demo
Always wanted to extend it further with segmentation so you can search for any type of object within the scene
To me, 'find people already complaining' is the useful bit here. But I'd question the blanket rule on ads: why rule out a small test that checks whether strangers understand one specific offer? The channel is secondary to what you're trying to learn.
i got to $120k revenue without doing half of this.
no product hunt. no 50 chatgpt blog posts. no facebook ads.
what actually got me my first 100 users (no bs):
1) dm'ing people who had already complained about the exact problem i was trying to solve on LinkedIn and X. keep it short and simple.
2) replying under every post in the niche for 30 days. not “check out my tool.” actual answers with insights.
3) obsessing over the first 20 users. be it calls, loom videos, or fixing their setup myself.
4) posting the $$ numbers. $0. then $10k. people trust authenticity.
ads and blogs only start working after some users are already paying you. before that 90% of the job is just talking to people who already have the problem.
are you stuck at getting users for your product?
comment the product link and i’ll tell you how i'll go about acquiring your first 100 users.
first 10 comments get a loom video.
Disclaimer - I lead growth at a YC startup
I think compatibility work is underrated. Getting FEX, Wine and a Direct3D-to-Metal runtime to cooperate means existing Windows games can run on iOS. You keep the software people already care about instead of asking them to start over.
Red Dead Redemption 2 is running at 30+ FPS on an iPhone 18 Pro.
No PC, no streaming, fully on-device via Madeira.
x86-64 → arm64 with FEX
Windows → iOS syscalls with Wine
DX12 → Metal with the madeira-d3d12 runtime
What AAA game should we get running next?
I'd rather encode what a machine can do than explain its limits in a prompt. Jevspresso uses XState to expose the available actions to Jev. The model still has to choose well, but it can't talk a forbidden transition into existence.
I mixed Jev + an espresso (state) machine.
Turns out, Jev's a pretty good barista ☕
Introducing Jevspresso: XState determines what's possible, Jev decides the next move. Order in plain language, or break stuff and watch Jev cope.
Try it: jevspresso.dev
"Today, we’re releasing two Cloudflare-trained decision models, Clef and Clef-flash"
blog.cloudflare.com/clef-dec…
Never seen an idea spreading so fast.
Cloudflare's Clef beat Jev in just two weeks
How: "frozen" Qwen does one prefill pass, a tiny schema head scores every answer option in parallel, with no text generated at all.
That's why it's 4x faster than Jev at 2x the accuracy on some benchmarks.
And it's open source!
Link to HF:
huggingface.co/Cloudflare/cl…
1/6
I think "open another dashboard" is a lousy answer to a simple indicator check.
I'd rather bring threat intel into the tools you've already chosen.
Here's how I'd pick between the isMalicious API, TAXII feeds and MCP.
ALT isMalicious brand graphic reading "THREAT INTEL. YOUR TOOLS." Blue data streams enter a central metallic block and branch toward three smaller blocks, with ismalicious.com at the bottom.
5/6
I'd choose the interface after deciding what needs to happen: a lookup, a collection import, or an agent tool call.
Then check the output format, refresh schedule and failure handling.
Those details matter more than another integration badge.
6/6
That's the point of "Threat intel. Your tools."
Keep the investigation where it belongs and choose the delivery path that fits it.
The API, feed and MCP setup are documented here:
ismalicious.com/integrations
AI release FOMO is a terrible engineering roadmap. Faster tools can amplify a messy system just as easily as a good one. Test upgrades against your own workload. This timeline's exact cadence depends on what counts as a major release.
1/6
I don't think anyone should block an IP just because a vendor painted it red.
I'd want enough context to challenge that verdict.
The isMalicious connector puts risk scores and source references into OpenCTI.
ALT isMalicious brand graphic with the words "BUILT FOR OPENCTI." above two dark metallic blocks connected by a glowing cyan bridge, with ismalicious.com at the bottom.
5/6
The connector is in the OpenCTI connectors repository and listed in Filigran's catalog as community supported.
You can read the code and configuration, including how the connector creates the enrichment.
6/6
If you use OpenCTI, try the isMalicious connector on an IP or domain you already understand.
Check the added score, labels and source references against your own findings. That's how I'd start evaluating it.
ismalicious.com