What Should I Eat Tonight? How to Stop Overthinking It
If the question "what should I eat tonight" has ever turned into a 30-minute loop that ends with you ordering the same thing you always get, you are not alone. Dinner decisions are genuinely hard, and they tend to arrive at the worst possible moment: late in the day, when decision bandwidth is nearly spent.
The fastest exit from the loop is to stop generating options mentally and start reacting to concrete ones. Open NomBot, tap spin, and three real nearby restaurants appear immediately. Every result is confirmed open right now. Pick whichever looks best, head out, and the loop is over.
Reacting to three specific picks takes fifteen seconds. Staring at an open question from scratch takes much longer.
Why is deciding what to eat at the end of the day so hard?
Decision fatigue is real. By dinnertime, most people have already made hundreds of small choices during the day. Most people know from experience that choices feel harder at the end of a long day than at the beginning. Restaurant decisions land at exactly the wrong end of that curve.
A second problem compounds the first: an unlimited option set. When you can choose from every restaurant you have ever heard of, each candidate needs its own assessment. Is it open? Is it nearby? Does it match my current mood? That invisible mental work adds up fast, and it happens before you have even committed to a direction.
Limiting the field to three specific, vetted options short-circuits the paralysis. The question shifts from "pick something from the entire universe of restaurants" to "which of these three." That is a completely different and far simpler cognitive task.
Three picks is the right number because it gives you real options without recreating the overwhelm. One pick feels like an ultimatum. Twenty picks is just a smaller version of the original problem. Three gives you something to react to.
How does NomBot decide which restaurants to show?
Every spin starts with two hard filters before anything appears.
The first is the open-now filter. Every result is a place confirmed open at the exact moment you tap spin. Not an estimate based on typical hours, and not "usually open on weeknights." Actual live hours data eliminates closed places entirely. This matters because the most common way a restaurant suggestion fails is recommending somewhere that has already shut down for the night.
The second filter is the radius gate. Every result falls within the distance you have configured. Nothing farther appears in the three picks, period.
After those two gates, NomBot applies a tiered recency penalty to restaurants in my spin history. If a place showed up in a recent spin, its score is reduced so it is less likely to surface again right away. The history window runs 90 days, and the more recently a place appeared, the steeper the penalty. This structural rotation means spinning on Monday, Tuesday, and Wednesday will naturally return different results each time without any manual input from me.
The last time I was stuck in the loop, I opened NomBot, chose ramen in Be Picky mode, and had three open spots in under ten seconds. One of them was a place I had walked past for months without thinking to try. That is the kind of variety the recency penalty produces: not random, but genuinely fresh.
NomBot also learns your taste the more you use it, getting better over time at surfacing what you are actually in the mood for.
What if none of the three picks feel right tonight?
Spinning again is always an option. The free tier includes three spins per day, resetting at UTC midnight. If the first round misses, a second spin pulls from a different part of the available pool.
If I have a general direction but not a specific craving, Be Picky mode lets me choose a cuisine type before spinning. Every result will match that category. More than 40 cuisine types are available, including breakfast and dessert modes for off-hour searches. When the mood is "something with broth" or "nothing too heavy," Be Picky turns that rough signal into three concrete open options.
The spin works at any hour. Late-night hunger uses the same flow as a 7pm dinner search. The open-now filter simply evaluates against whatever time it is when you tap.
If the free three spins per day are not enough for how often you eat out, the Pro plan at $4.99 a month removes the limit entirely. For frequent users, the Founders plan offers lifetime unlimited access for a one-time fee.
The loop stops when you have something specific to react to
The painful part of the "what should I eat tonight" question is not picking from three options. It is the time spent before anything concrete is on the table, mentally cycling through possibilities without committing to a direction.
Putting three real picks in front of yourself ends that phase of the decision. One of them usually stands out immediately. On the nights when none of the first three land, a second spin resolves it quickly.
Before NomBot, I would sometimes spend more time deciding what to eat than actually eating. These days I tap spin and go with whatever catches my eye in the first set. The 90-day recency window means the picks keep rotating, so even in familiar neighborhoods the results stay fresh.
For more on how each spin is assembled from start to finish, how NomBot picks three restaurants covers the full pipeline in detail.
When the tonight question turns into a standoff with someone else, what to do when neither person can settle on where to eat covers that scenario.
And if the back-and-forth with another person is the actual problem, breaking the where-to-eat loop when dining out with someone else has specific approaches for that.
NomBot is free on iOS. Three picks per spin, all confirmed open, and the recency window keeps results fresh night after night.
