The creator tooling problem nobody posts about: your laptop is the bottleneck
Here’s a thing I’ve started asking people at creator meetups: how many browser tabs, CapCut exports, and AI chat sessions were running the last time your fans spun up so loud you had to leave the room? Almost every hand goes up. We spend enormous energy optimizing the output side of the creator stack — scheduling, repurposing, analytics — and almost none on the input side, which is the physical machine sitting on your desk running four tools at once while you batch-record voiceovers. That’s why a tiny macOS utility called WhoIsCooking caught my attention this week, even though it has nothing to do with social media. The maker’s framing — “which agent is responsible for my Mac getting hot” — is the same diagnostic question social operators should be asking about their own workflow: not is my machine slow, but which specific process is eating my afternoon. The answer changes how you schedule your week.
The actual problem: attribution, not temperature
The maker, Andrew Wilson, describes running Claude Code and Codex on a MacBook and finding that the machine “kept getting hot, with no way to tell which agent was responsible.” His observation about Activity Monitor is the crux: it “shows dozens of node and compiler processes and no temperature at all.” Anyone who has run a long AI-assisted build knows this exact feeling — the fan ramps, you open Activity Monitor, and you’re staring at forty node entries with no idea which one belongs to the session you actually care about.
WhoIsCooking’s answer, per the launch page, is threefold: it shows chip temperature “in your notch or menu bar,” it “groups each agent session with everything it started into one row,” and it offers a “Cool down my Mac” button that “moves busy work to the efficiency cores.” Crucially, the maker states “nothing is killed or frozen” — the work continues, just slower. He reports that on his M5, “4 processes flat out went from about 83°C at full speed to about 48°C cooled.” That’s a maker-reported figure, not an independent benchmark, so treat it as directional rather than gospel. The honest limits are stated up front: it can’t cool GPU work or Apple’s own processes, and it requires Apple Silicon (M1 to M5) on macOS 14+. Pricing is $5 one time, signed and notarized, no account — per the launch post.
Why this matters more to a solo creator than to a dev team
A dev team has a build server. A solo creator has a laptop that is simultaneously their editor, their scheduler, their AI research assistant, and their podcast recorder. When that single machine thermally throttles mid-export, the cost isn’t abstract — it’s a re-render you have to babysit. The insight worth stealing here isn’t the temperature readout; it’s the grouping model. One row per session, with everything that session spawned nested underneath, is a UI pattern that maps almost perfectly onto how a content operator thinks about a publishing day.
What the tool actually is — and what it isn’t
Let me be precise, because the Product Hunt page is a launch post, not a spec sheet. WhoIsCooking is a macOS menu-bar/notch utility that (a) surfaces chip temperature, (b) groups agent sessions and their child processes into single rows, and © throttles by migrating work to efficiency cores rather than suspending it. It is not a task manager, not a process killer, not a GPU monitor, and not a cross-platform tool. The maker explicitly says it “can’t cool GPU work or Apple’s own processes.” If your bottleneck is a GPU-bound CapCut export or a DaVinci Resolve render, this won’t help you — that’s a real limitation, and I respect that the maker led with it.
The comparison set here is thin, which is itself informative. The nearest incumbents are general-purpose system monitors like iStat Menus and Stats, both of which show temperature and per-process CPU but do not group child processes under a parent session or offer a one-click efficiency-core migration. So the differentiation is real but narrow: it’s the session-grouping and the cool-down action, not the thermometer. My take: the $5 one-time price with no account is the right call for a utility this focused — it signals “small sharp tool,” not “platform.”
Where the math breaks
The efficiency-core trick is clever but not free. The maker says the work “just runs a bit slower,” and that’s the honest trade: you’re trading wall-clock time for thermal headroom. For a background agent doing research or a long test suite, that’s obviously correct. For anything on a deadline — a scheduled post going out at a fixed time, an export you need in ten minutes — throttling is the wrong lever, and the tool has no way to know your intent. That’s not a flaw in the product so much as a reminder that “cool down” is a decision, not a magic button, which is exactly the point one commenter made.
What social media teams can borrow from this
The most useful thing about this launch isn’t the software — it’s the mental model. Keesan’s comment nails it: “separate a healthy long compile from a runaway loop and make ‘cool down’ a decision instead of a blind throttle.” Swap “compile” for “content batch” and you have a genuinely useful framework for social operations. When I scheduled thirty posts across five platforms last month, the thing that actually slowed me down wasn’t the Buffer queue — it was three AI drafting sessions, two export jobs, and a browser with forty tabs open, all competing for the same machine while I tried to write captions. I had no visibility into which of those was the real cost.
The session-grouping principle, applied to content ops
Here’s the borrow: stop thinking in tools and start thinking in sessions with children. A single “Instagram Reel for the Q3 launch” session spawns a script draft, a CapCut project, three caption variants, a thumbnail in Canva, and a scheduled slot in Later or Metricool. If you can’t see that whole tree as one unit, you can’t tell whether the session is healthy (long, productive, expected) or runaway (you’ve been re-editing the same fifteen seconds for two hours). The same attribution logic that makes WhoIsCooking useful for agents makes a content calendar useful for humans — and most calendars, frankly, don’t do this. They show posts, not the work that produced them.
Why TikTok creators should care more than LinkedIn ones
A LinkedIn operator publishing three text posts a week barely touches their machine’s limits. A TikTok or YouTube Shorts creator lives in export, re-export, and AI-assisted iteration — the exact workload that heats a laptop and the exact workload where a throttled render costs you a posting window. If you’re batching five vertical videos in a single afternoon, knowing which session is the thermal hog is operationally meaningful in a way it simply isn’t for a text-first publisher. My bet: the creators who get real value from tools like this are the ones whose output is heavy media, not the ones whose output is a paragraph and a link.
The honest limits and open questions
I want to be balanced here, because the launch page is refreshingly candid and I’d rather reward that than pile on hype. The stated limits are real: no GPU cooling, no coverage of Apple’s own processes, Apple Silicon only, macOS 14+. The maker also asks for feedback on “what else it should show per agent,” which tells you the per-session detail is a work in progress. Ida Hansen’s question — whether you can check temperature history and see which agent puts the most strain on your Mac across a day — is the right next question, and the page doesn’t yet say whether history is supported. Not disclosed. Simon Moxon’s reaction — “one row per agent… beats staring at 80 processes” — is the value prop in one sentence, and it’s the part I’d want to verify hands-on before recommending it to a team.
The bigger open question is scope. This is a single-machine, single-user utility. It doesn’t touch the cloud, doesn’t integrate with your scheduler, and won’t tell you anything about a distributed team’s aggregate load. If you’re a social team of eight running a shared content pipeline, this is a personal productivity toy, not infrastructure. And if you’re on Windows or Intel Macs, it’s simply not for you — the maker says so plainly. That honesty is worth more than a feature list, but it also means the addressable audience is narrower than the launch-page enthusiasm suggests.
What I’d watch / test next
If you run a content operation on an Apple Silicon Mac, here’s a concrete experiment for this week: before your next batch-production session, open your normal system monitor and try to attribute CPU load to specific projects. If you can’t — and you probably can’t — that’s your signal that session-level attribution is a gap in your workflow, whether or not you buy a $5 utility to fix it. Then, during the session, note which activity actually correlates with the fan ramping: is it the AI drafting, the export, or the forty tabs? That single data point will change how you sequence your week more than any new scheduler will.
I’d also watch whether the maker ships temperature history per session, because that turns the tool from a live thermometer into a genuine diagnostic — the difference between “my Mac is hot right now” and “this specific agent has been the thermal cost all day.” And I’d test the efficiency-core migration on a non-deadline task first, precisely because the trade is wall-clock time for heat, and you want to feel that trade before you rely on it. For the broader creator stack, the lesson stands on its own: the most under-optimized part of most content operations isn’t the publishing tool — it’s the machine and the sessions running on it. Attribution beats brute force, on a laptop and in a calendar alike.
