Google Assistant vs Pokobot
Said out loud, or written down somewhere you will look.
Saying “remind me to take the bins out at seven” while your hands are in the sink is the fastest way anyone has invented to set a reminder. No app, no typing, no unlocking anything. Google Assistant does this well, and on a speaker or a display it does it without a phone in the room at all.
The trouble starts after you have said it. Speech is a superb way to get something in and a poor way to keep it, because the moment it has been spoken there is nothing left: no thread to scroll back through, nothing to check, and nothing to correct except by starting another conversation with a machine.
The short answer
Google Assistant turns speech into reminders better than anything else, and if your hands are usually busy it will always beat typing into a chat. The weaknesses are all properties of voice rather than faults in the product. You have to say it out loud, which rules out anywhere you are not alone. It is announced out loud, which means it reaches a room rather than a person, and tells everyone in that room what you had planned. You cannot skim a spoken list, so reviewing what is set is genuinely tedious. And names and mixed-language phrases, the exact things Indian users type constantly, are where speech recognition is weakest. Pokobot is slower to set and leaves a written record in a thread you can scroll, search, correct and read in a meeting without saying a word.
Head to head
| Google Assistant | Pokobot | |
|---|---|---|
| Setting one with your hands fullGoogle Assistant | Say it. Nothing else on this list comes close | You have to type it, like any message |
| Around the house with no phoneGoogle Assistant | Speakers and displays in every room. Works with the phone upstairs | Needs the phone. It is a chat app |
| Calling out to another roomGoogle Assistant | Broadcast to the speakers and everyone in the house hears it | No equivalent at all |
| Free foreverGoogle Assistant | Yes, on hardware most people already own | No. 7 days free, then a small subscription (from ₹59/month in India) |
| Setting one without speakingPokobot | Not the point of it. In an office or a meeting you are stuck | Typing is silent, so anywhere works |
| Names and mixed-language phrasesPokobot | Where speech recognition is weakest, and the failure is silent | You typed it, so it says what you meant |
| Who it reachesPokobot | A room. Whoever is standing in the kitchen hears your reminder | You, in a thread nobody else is reading over your shoulder |
| Checking what is setPokobot | Ask, then sit through the answer being read to you | Send “list” and skim it in three seconds |
| Something to look back atPokobot | Once it has been spoken it is gone. There is no transcript to scroll | It is a chat. Every reminder is still there in writing |
Voice is genuinely better at getting things in
There is no argument to be had about capture, so this page will not attempt one. Speaking is faster than typing whenever your hands or eyes are busy, and that covers a large share of the moments when people remember something.
In these situations Assistant is the right tool and a chat thread is a downgrade.
- Cooking, driving, holding a baby, halfway through a repair. Anything where a phone is not in your hand.
- Anything at home where a speaker is closer than your phone.
- Broadcasting to the house, which nothing here can do.
- Reminders that are genuinely simple: bins, tablets, take the chicken out of the freezer.
- Paying nothing, on a device you already bought for something else.
A spoken reminder leaves nothing behind
Every other tool in this comparison produces an artefact. A row, a note, a card, a message. Assistant produces a sentence in the air, and then the sentence is over. That feels efficient right up until you want to check something.
It shows up in small ways that add up. You are not certain it heard the time correctly and there is no way to glance and confirm. You want to change one of six reminders and have to describe which one out loud. You want to know what is set for tomorrow and the only route is to ask and then listen to a list being read at you, which is slower than reading for anything past two items.
A chat thread is the opposite of ephemeral. The reminder you set in March is still sitting there in writing, with the exact wording you used. You can scroll it, search it, and correct one by replying to it. And the nudges themselves can run at a pace voice was never going to manage: 35.8% of the reminders set here repeat more than once a day, each with a window so it goes quiet at night.
remind me to call Bhavika about the GST filing tomorrow at 11amlistactually make it 7pmYou have to say it out loud, and that is most of the problem
Voice input has one hard requirement that nobody lists as a limitation: you must be somewhere you are willing to speak. That rules out the office, the meeting, the train, the room where someone is sleeping, and the moment during a call when you remember something.
Those are not edge cases. They are a large share of the working day, and they are exactly when obligations tend to occur to you, because that is when you are dealing with the things that generate them. The reminder you cannot set in the meeting is usually the reminder that came out of the meeting.
So voice ends up covering the domestic half of life very well and the working half barely at all. Typing has no such constraint, which is unglamorous but decisive: a message can be sent from anywhere, including places where saying anything at all would be rude.
It is announced to a room, not delivered to a person
When a smart speaker fires a reminder, it says it out loud to whoever is standing there. That is a lovely property for “take the chicken out of the freezer” and an uncomfortable one for a fair amount of ordinary life.
Reminders are more private than people assume. Medication, a clinic appointment, a difficult call you have been putting off, something you are organising for a family member's birthday. A speaker does not know the difference and will read any of them to the room at the moment they are due.
There is also the plain reliability question of whether you happen to be in that room at that minute. A reminder delivered to a place only works if you are standing in the place. A WhatsApp message follows the person, which is why it is the safer default for anything that actually matters.
Names, and everything else voice mishears
Speech recognition is very good at ordinary English sentences and noticeably worse at proper nouns, Indian names, brand names, and sentences that switch between languages mid-way. Those are not exotic inputs. They are what most reminders in this part of the world actually contain.
The failure mode is what makes it costly rather than the failure rate. A misheard reminder does not tell you it was misheard. It gets stored as something plausible and slightly wrong, and you find out at the moment it fires, if you find out at all.
Typing removes the entire category. Poko reads what you wrote, confirms it back to you in writing, and if the phrasing genuinely could mean two things it asks rather than guessing.
remind me to send Pranjal the rent receipt on the 3rd at 10amUse both, split by where you are
The division that works here is not by type of reminder but by where you happen to be standing when you think of it.
At home, with your hands busy and a speaker in earshot, speaking is faster and you should keep doing it. Anywhere you cannot or would rather not talk, and for anything you will want to check later, correct later, or keep private, a written message in a chat is the better home.
The two do not conflict and there is nothing to connect. Most people who use both stop thinking about it within a week and just use whichever their mouth or their thumbs can reach first.
Questions, answered
- Are Google Assistant reminders private?
- Not when they are delivered on a speaker or a display. The reminder is announced out loud to whoever is in the room at the time, and the assistant has no way of knowing that a medication or an appointment is something you would rather not have read aloud at dinner. On a phone it is a normal notification and about as private as any other.
- Can I set a Google Assistant reminder without talking?
- You can type to Assistant on a phone, but the whole advantage of voice disappears when you do, and you are then typing into an assistant rather than a keyboard designed for it. If most of the moments you remember things happen in meetings, offices or quiet rooms, voice is not the tool doing the work.
- Why does Assistant get names wrong in reminders?
- Speech recognition is trained hardest on common words in one language at a time, so proper nouns, Indian names and sentences that switch language partway are where it is weakest. The bigger problem is that it does not flag the failure. It stores a plausible mishearing and you find out when the reminder arrives.
- Can Pokobot work on a smart speaker?
- No. It is a WhatsApp chat and needs a phone. There is no voice interface, no speaker support and no broadcast. If most of your reminders are set and heard around the house, Assistant is doing something here that this does not attempt.
- How do I see all my reminders in one place?
- With Assistant you ask and listen, which is fine for one or two and slow for more. With Pokobot you send “list” and get a written list you can skim, then “cancel 2” to drop one of them. Reading is faster than being read to for anything past a couple of items.
- Can Google Assistant remind me every two hours?
- Not as one instruction in the way a reminder bot handles it, and there is no active window, so anything you do get running has no way to pause overnight by itself. Pokobot takes “every 2 hours between 9am and 7pm” as a single sentence and stops outside those hours.