
Wednesday, Oct 01
Presented byArcjetFeeds Made of Prompts
gm legends, happy Wednesday.
Hereās todayās lineup: OpenAIās Sora 2 comes with a TikTok-style app where every clip in the feed is AI-generated, complete with cameo slots for your face if you opt in; Verdent Deck runs multiple coding agents in parallel and gives you clean diffs so you actually see what changed; CrePal turns your ideas into short films by handling the script, scenes, edits, and audio from a single prompt.
P.S. Building something new? Tell us about it ā editorial@producthunt.co š«¶
TikTok, but All Fake

OpenAIās new drop has two parts. First, Sora 2, a video model that can spit out high-quality short clips from prompts. Second, an actual TikTok competitor where every single video in the feed is AI-generated. You can even hand over your face for cameos, which means strangersā prompts can star⦠you.
š„ Our Take
This isnāt just a model upgrade, itās OpenAI building an entertainment machine. The idea of a feed where nothing is real feels equal parts brilliant and cursed. On one hand, it kills the excuse of āI donāt have time to make content.ā On the other, do we really want infinite AI skits starring our friendsā faces? Feels like the uncanny valley just got its own social network.
Whereās the IDE Crown in 2025?

fmerian asks: Poll: Best IDE in 2025? He tees it up with the Stack Overflow survey and asks what people actually use day to day. The thread leans VS Code for stability and extensions, with strong pockets for Cursor and Claude Code, and JetBrains when language depth matters. The subtext is simple: code editors are identity as much as tooling.
Your agents stopped answering and started doing
Your agents call tools, write to databases and move money, so a wrong answer is now a wrong action, which is a considerably worse meeting. Arcjet runs policy before every LLM call, tool call, query and API request: prompt injection, sensitive data, bots, rate limits, spend quotas, and your own rules in Rego on top. It reads the OpenTelemetry you already emit, so switching it on is one environment variable, no code changes and nothing to deploy. Each check also reads the earlier steps in the same workflow, so an agent's tenth action gets judged against the first nine and not on its own. JavaScript, Python and Go SDKs all at 1.0, agent framework guards included, free tier to start on.
Parallel Agents, Real Code

Verdent Deck coordinates multiple AI agents to attack coding projects in parallel. You fire off a rough idea, it checks in with you, builds a plan, then spins up agents to generate, review, and refine code all at once. Everything is transparent, you see the diffs, you see what changed and why. It supports GPT-5, Claude Sonnet 4.5, and other models so you pick what powers your stack.
š„ Our Take: Running agents in parallel feels like sending clones to do your work. The clever part here is giving you visibility and control, so you donāt end up in āwhich agent screwed it upā hell. If youāve ever watched a coding tool spit out garbage youāre too scared to touch, this one at least gives you eyes on it.
Movies From a Prompt

CrePal is your AI director. You type in an idea, and CrePal handles the rest, script, storyboard, scene generation, edits, to turn a prompt into a short film. It blends models, tools, and sub-agents so you donāt have to mess with all the pieces.Ā
š„ Our Take: You donāt need to learn every video tool,Ā you need someone (or something) to get it done. CrePal tries to be that. Itās not about replacing directors, but giving storytellers a shortcut when they just want to see their vision on screen without wiring every part themselves.










