The BEST ESP32 project based on ChatGPT | IOT Projects | NextPCB

Подписаться 130 тыс.

Просмотров 69 тыс.

50% 1

Ready to take your ESP32 projects to the next level? In this video, we'll show you how to create your own ChatGPT voice assistant using an ESP32 board. This DIY tutorial is perfect for anyone looking to incorporate artificial intelligence into their IOT projects. Watch now to learn how to make your own ChatGPT voice assistant with ESP32!
⚡️⚡️ You can also start learning IOT from scratch with my Course "Master of IOT". Enroll Now - techiesms.graphy.com/
⚡️⚡️ Order all components required for making this project using this single link - techiesms.com/product/chatgpt...
You can support me🙏🏻 and also get benefitted on Patreon :- / techiesms
__________________________________________________________
👨🏻‍💻 NextPCB Links 👨🏻‍💻
This video is sponsored by NextPCB - the Reliable Multilayer PCB Manufacturer.
www.nextpcb.com
Free international shipping for Authentic Electronic Components:
www.hqonline.com/
Enjoy free PCB assembly for 5 boards:
www.nextpcb.com/pcb-assembly-...
DFM free online PCB Gerber viewer:
www.nextpcb.com/free-online-g...
__________________________________________________________
👨🏻‍💻 Important Links 👨🏻‍💻
For OpenAI API - platform.openai.com/
For Google Cloud API - cloud.google.com/
Postman App Download - www.postman.com/downloads/
Codes & Sch Used in this project - github.com/techiesms/ESP32-Ch...
ESP32 ChatGPT Video Series - • Running ChatGPT on ESP...
__________________________________________________________
Components used in this project 💡🔌
⚡️⚡️ Order all components required for making this project using this single link - techiesms.com/product/chatgpt...
Or if you want to order only specific components then here is the list and link,
ESP32 Dev board - www.techiesms.com/product/esp...
I2S Mic - www.techiesms.com/product/inm...
I2S Audio Amplifier - www.techiesms.com/product/max...
Speaker - www.techiesms.com/product/rec...
IR Sensor Module - www.techiesms.com/product/inf...
TP4056 Battery Charging Module - www.techiesms.com/product/tp4...
Battery - www.techiesms.com/product/kp-...
__________________________________________________________
⏱ Time Stamps ⏱
__________________________________________________________
🎥 🎙 My Video Gear 🎙🎥
Camera - Canon 750d
Lens - 50mm f/1.8 STM Lens (amzn.to/32gqnsi), 24mm f/2.6
Key Light - Godox SL60W (amzn.to/3522mHe)
Fill Light/Back Light - Osaka Lightning Setup (amzn.to/2I5cdDi)
Tripod - amzn.to/3uNhC4H
Microphone - Zoom H1N(amzn.to/3Osh12A)
Editing - Final Cut Pro X on MacBook Pro ( amzn.to/3vHcSyJ )
__________________________________________________________
Get connected with techiesms everywhere..
Subscribe Main Channel :- bit.ly/techiesms_youtube
Subscribe Shorts Channel :- bit.ly/techiesms_shorts
Facebook :- / techiesms
Twitter :- / imtechiesms
Instagram :- / techiesms
Website :- www.techiesms.com
Telegram - @techiesms
__________________________________________________________
techiesms
explore | learn | share

Наука

Опубликовано:

19 янв 2024

Ссылка:

Скачать:

Готовим ссылку...

Добавить в:

Мой плейлист

Посмотреть позже

Комментарии : 219

@techiesms 4 месяца назад

If we ever make version 3 of this project, what will be your suggestions about adding/removing things in this project? Share your thoughts..

@ghostandinfinite9874 4 месяца назад

Make it offline. Even tho it seems a little hard

@ghostandinfinite9874 4 месяца назад

Great work anyways. As always I will keep buying products from you

@ghostandinfinite9874 4 месяца назад

The thing is ultimately it isn't free

@AmmarNegm 4 месяца назад

you can do the same thing (generative language) for Gemini Pro and I have a good idea, you can make the project more beautiful by using OLED screen (SSD1306) and Wowki Animator.

@AmmarNegm 4 месяца назад

Because Gemini API is free and I have another idea:2. ESP (I mean the ESP32 module that controls the speaker) should be in the box, but 1. Let the esp (I mean the esp32 module that controls the microphone) be on a remote control and record the sound when you click the button.

@techieviral9741 4 месяца назад

Thank you so much for this latest project. Hats off for your contribution to the hobbyist community. ❤❤

@user-yr2qe6sg4h 3 месяца назад

Hello Techiesms! The project is working great - great job! The problem is low processor memory. only 2 seconds for the Questions task is too little. Longer responses are skipped because the receiving buffer is too small. I have a question whether it is possible to modify the program for ESP32 VROVER to extend the Voice Recorder (5 sec) and the receiving buffer for Text 2 Speech (Max Tokens 100). Then the device would be perfect! Warm greetings from Bavaria! Roman

@marcelverhoeven3514 4 месяца назад

As suggested below: use a wakeup word and create the possibility to 1. run an Internet radio 2. use ChatGPT for question/answer, 3 Perform an action with Servomotor. Integrate also the board DY-HV20 for predefined responses. Then this project will become a very good foundation for a ROBOT. Anyway: Sachin congrats: version 2 works!!

@kaloprojects 3 месяца назад

just seen i made your 123.000's youtube abo lol :D, congratulation techiesms :) .. go on with this great work !, locking forward v.3

@techiesms 3 месяца назад

Thank you ✌️

@vuaxumu0lin3 3 месяца назад

Great, thanks for share. Love u.

@mibrahim4245 4 месяца назад

so did you solve the problem of slow response from Google Cloud Speech to Text ? did you provide it with the optimal audio resolution ?

@lifescience4all 4 месяца назад

its look amazing ... this time your talking clearly and slowly thanks for this video last video i got so many errors. hope this time its work... i will update once its works...Please keep doing this kind of videos...

@techiesms 4 месяца назад

I try my best to minimise the errors So do try it out and let me know if you still get any error

@amalnp3934 2 месяца назад

Bro did that work

@lifescience4all 2 месяца назад

@@amalnp3934 not working bro

@scienceoftheuniverse9155 4 месяца назад

Great as always. Really appreciate your great effort

@techiesms 4 месяца назад

Thank you so much 😀

@LooksmaxingYT 4 месяца назад

@@techiesmsBro help me about Google api is there alternative

@naitik9534 4 месяца назад

bro techsims can I make it work without pcb

@naitik9534 4 месяца назад

through jumper wires

@sparshsinghg 4 месяца назад

Is there a specific reason you chose to have two different boards for two different functions? Can’t both of them be uploaded into one?

@kaloprojects 3 месяца назад

Hi techiesms, your project is awesome .. can't wait for next steps. I will rebuild it for sure, doing some adjustments and enjoy ;). Any ideas for battery mod would be best anyhow. To your questions on eof videe (ideas for next steps), here my personal coins: 1. stay with ESP32, not using Raspberry Pi, reasons e.g: (a) we can utilize the existing code, no Python needed, (b) no Pi (Unix OS) overhead needed, keep it small and beauty, and (c) much easier for battery powering .. just my coin (i jumped once to Pi, did not like it .. then went back to my sweet home ESP32 & Teensy ;) 2. solving the input length issue would be most important (i feel 2 or 3 seconds are too short for nice chats) 3. yes, solving the charging issue, would be awesome to have this beauty box totally portable ! 4. later on maybe a 'wake-up-key' phrase (just optionally, i will stay with a simple capacitive touch button for now) just my coins, I'm exited next steps ! .. cool work !

@Monsieur.Nobody. Месяц назад

G'day I am wondering if it can do Speech to Text, and how small can it really get? I saw the XIAO Esp32 being tiny. I have a project that requires wireless, real-time Speech to Text capabilities, and 4 programmable buttons on it... in the smallest form factor possible... Max one index finger size. Do you reckon it can be done with Esp32's or am I better off looking somewhere else? what would you suggest?

@leeronlacson813 3 месяца назад

Hello! Can you provide the design of the PCB that you used?

@singhlaprit2117 3 месяца назад

use DC to DC converter and for the trigger use an accelerometer like lis2dw12 which has a power consumption of around 1 microamp, put deep sleep both esp32, and when someone taps on your project or shakes it it gives a trigger, same as your IR sensor, and both esp32 wakeup and ready for normal operations there same delay when wakeups like connect to the router and other process so indicate with light when the system is ready after no voice listening for certain period, go back to sleep overall power consumption should be around 20 uam ,if Load Switch IC use where ever needed and your low power consumption device is ready is may able to run 1 month on 1000 mah battery

@xeuangammer9364 3 месяца назад

Custom PCB design isn't there on the description Can u please upload that

@pushparajahpushpandan6268 4 месяца назад

Great project and why can not we use single DevKit for this?

@sanchaykumar9903 4 месяца назад

Can we use esp home and add wakeup word and keep rest code same?

@rselvarajanMBA 4 месяца назад

Unfortunately, ChatGPT API isnt cheap. You incur cost for both input and output , so a device running chatgpt as the interface will quickly run out of credits. A better alternative is to use Hugging chat which is free but it only has python library.

@omkarbansode6305 4 месяца назад

Thanks for the suggestion

@LooksmaxingYT 4 месяца назад

Bro help me about Google api is there alternative

@snopz 4 месяца назад

You can just scrape the library api and use it in c code instead of the openai api

@omkarbansode6305 4 месяца назад

@@snopz can you please explain it in little more detail if possible or provide with the some links please

@jaysalomon5298 3 месяца назад

I've tried multiple options , from llama based models , home start etc and a few local LLMs all native on my GPU but none of them compare to GPT APi, it is expensive but I've set my spend to £6 a month and use the cheapest 3.5 API and it's just unrivaled, I used it integrated into home assistant and no other LLM comes close , which is unfortunate for the time being

@amalnp3934 3 месяца назад

The audio.h Library shows an error like In file included from c:\Users\sreerama\Documents\Arduino\libraries\ESP32-audioI2S-master\src\Audio.cpp:15:0: c:\Users\sreerama\Documents\Arduino\libraries\ESP32-audioI2S-master\src\opus_decoder/opus_decoder.h:7:20: fatal error: memory.h: No such file or directory compilation terminated. exit status 1 Compilation error: exit status 1

@eyesontheshow1376 3 месяца назад

Same error did you find the solution...? Please help 🙏

@eyesontheshow1376 3 месяца назад

@amalnp3934 did you find the solution of this error, please help

@saidevdhal 2 месяца назад

Download that directory or library

@rohitpujari2710 2 месяца назад

@@saidevdhalfrom where?

@DUBIX_INT 3 месяца назад

What changes do i need to make in the code to be able to use Google Gemini api instead of chatgpt api

@ArjunKohli Месяц назад

How much google cloud account charge initial amount when we verify. Can we direct pay by netbanking that amount ?

@AmmarNegm 4 месяца назад

You are great, you can do the same thing (generative language) for Gemini Pro and I have a good idea, you can make the project more beautiful by using OLED screen (SSD1306) and Wowki Animator.

@AmmarNegm 4 месяца назад

For example, when receiving a response, let the OLED screen perform a certain animation, let a logo appear when it is started, etc.

@mr.whitetech3887 3 месяца назад

If Possible can We Change The Voice note Of The Assistant with The Help Of Other AI Softwares Like Eleven's lab To Change The Voice Note Of The Output?

@suvojitkumbhakar1724 2 месяца назад

@techiesms can i use Gemini api instead of chat gpt api ?

@5013Lakshminarasimman.S 3 месяца назад

how to get the pcb ??

@kaloprojects Месяц назад

Hi techiesms .. let us know: could you solve the 2 sec record length issue meanwhile ? I would love to record e.g.5-10 secs. ANY idea so much welcome - thank you!

@TheBestClips221 4 месяца назад

Great video, can you make a a video about esp32c6 and his zigbee ability inside arduino ide? There is no one video on that topic....

@snopz 4 месяца назад

You could use one single esp32 and link it to esphome and make a pipeline for the chat gpt and the tts and stt (tts and stt are free and open source in home assistsnt) And also you will be able to use wake word instead of the distance sensor.

@LooksmaxingYT 4 месяца назад

Bro help me about Google api is there alternative

@LooksmaxingYT 4 месяца назад

Bro help me about Google api is there alternative

@antonpegov2745 4 месяца назад

What free tts and stt you mean? 😮

@snopz 4 месяца назад

@@antonpegov2745piper(tts) and whisper(stt) they are both free and open source and included in home assistant as integrations

@FunnyGaming-up7gl Месяц назад

Where do I get the gerber file of the PCB?

@Thuliolima2008 4 месяца назад

Very good 😮

@techiesms 4 месяца назад

Thanks 😋

@prachidevram120 4 месяца назад

What about proximity sensor ? How it should be connected in circuit? Also, how is working the way its changing color ?

@eyesontheshow1376 3 месяца назад

same question , Have you found a solution to this issue?

@Adrian-yz7oe 4 месяца назад

Great video! When its finished and polished it will bean awesome unit to buy

@HassanButt-eo9fp 29 дней назад

sir when i compile the code error: architecture or board not supported sd.h file ka sath is there any solutionn??????

@mailoctho.success 18 дней назад

hello, i want this project to be able to combine voice to control multiple devices

@kalalakinto6830 Месяц назад

How could I order the PCB board of the project

@LooksmaxingYT 4 месяца назад

Bro help me about Google api is there alternative credit card issues 😭

@20shubhampatil77 Месяц назад

I am getting error into speech to text section actually I am not getting the text results what should I do

@26shalinilande56 2 месяца назад

Project is amazing but I need help that speech to text works but text to speech doesn't work Plz give me suggestions. .??what should I do

@sanjayraju3084 2 месяца назад

Bro can you send the link of the PCB design you have build it

@anishpatil9843 Месяц назад

sir where can i find ur pcb design so i can order it

@surflaweb 4 месяца назад

Good job

@techiesms 4 месяца назад

Thanks

@pdlRSD 3 месяца назад

How many words does it respond at a time?

@snopz 4 месяца назад

Does anyone found anyway to increase recording time?

@niwanthanaweendrajith6555 4 месяца назад

Hi your painless mesh library is not working pls check it😢

@Mithusanth-xg8ys 2 месяца назад

How to connect a display with it for see pictures

@mr.whitetech3887 3 месяца назад

Waiting for the Next Version video !!!!

@user-ve4lw6hg6s 2 месяца назад

can you post pcb documents? Thanks

@VDVR-le4wx 28 дней назад

Hi sir very useful this video And then I am Studying college I have submit my Minor project In this Chatgpt based voice assistant project is ok or not It was successful completed and ahow the output clearly pls clear my doubt and pls reply

@marcelverhoeven3514 4 месяца назад

WiFiManager with SPIFFS data memory for the API keys

@TheAvidtech 4 месяца назад

sir please upload fast and make a video on how to change the voice of ai and work like robot

@TomS699 4 месяца назад

Why are you using two esp32's ??? It has dual core

@amoghrao9260 2 месяца назад

Can we replace google cloud ai with chtgpt api

@vijayendragoud119 4 месяца назад

sir can you make evive - The Best DIY Prototyping Tool with in custam pcb desion and in limited buget it can fill all the dream who are intreasted in electroice plz help

@ZenHulk 3 месяца назад

I wish i could solder, maybe if i can get the cornea transplant someday.

@technowiz10 Месяц назад

Can we make this project by capturing photos using esp32 cam according to our prompt and processing it using open ai and get the result through a speaker

@techiesms Месяц назад

@waynesouth 2 месяца назад

can you make ESP32 project based on ollama Stand alone GPT

@yourbrotherdeepak7699 Месяц назад

Agar es assistance ko bna banaya kharide to kitne ka pad jayga

@jalmince 4 месяца назад

I had done that 6 month ago. With the following differences: used only one esp32, and amazon for tts and stt. It works fine, BUT : The sequence stt - > chatgpt -> tts always takes all together at least total 30s....honestly, did you not make it seem shorter when editing your vid ? Cheers, jp

@_plsubscribe_ 3 месяца назад

found any solution to make it faster?

@jalmince 3 месяца назад

@@_plsubscribe_ nope. I gave up!

@_plsubscribe_ 3 месяца назад

@@jalmince am i better off with raspberry pi zero 2w then?

@jalmince 3 месяца назад

@_plsubscribe_ sorry I don t get your question. To me the issue has to do with how fast the api' s reply to your queries. No matter which hardware you use.

@_plsubscribe_ 3 месяца назад

@@jalmince ohh thought it due to hardware capabilties.

@PavolFilek 4 месяца назад

Nice

@techiesms 4 месяца назад

Thank you

@sanjeebkumar8997 3 месяца назад

What about rpi small version

@vijayendragoud119 4 месяца назад

sir can you make a evive starter kit at low cost

@MANITDANGAL1125 2 месяца назад

Why didn’t you use esp32 module only

@usmanumer9871 4 месяца назад

maza a gaya

@user-ns5iz1kk2p 2 месяца назад

How can I get the gerber file

@amoghrao9260 2 месяца назад

How can i get youyr PCB

@byakuyalight9007 4 месяца назад

Exist code for esp32 voice recognition offline ?

@user-rm4gy6ir5e 3 месяца назад

Can I get the PCB design

@gamingwithncsvsrg2416 4 месяца назад

can i use esp8266

@joshuaolaiya 4 месяца назад

How much does it cost to buy one from you including the schematics and code?

@joshuaolaiya 4 месяца назад

Isn't there free api instead of using Google cloud and OpenAI chatgpy

@saidevdhal 2 месяца назад

Code is free idk about schematics and iots are 1500 inr

@tarundandekar6179 4 месяца назад

Sir I am getting error while uploading the code plz help

@Drzafar2438 2 месяца назад

Esa PCB kahan sy milay ga?

@arsalankhan5974 4 месяца назад

WHICH SOFTWARE YOU USE TO MAKE PCB DESIGN AND ALSO CAN A BEGINNER USE THAT .....???

@techiesms 4 месяца назад

Easyeda And yes it’s beginner friendly

@reni_christian 4 месяца назад

first view ❤🎉

@techiesms 4 месяца назад

You're the best!

@amoghrao9260 2 месяца назад

Wher is the github reposiory

@jhansidurgasreechapala1237 Месяц назад

I'm having an error "[HTTPS] begin... [HTTPS] Connection established![HTTPS] GET... [HTTPS] GET... failed, error: connection refused" like this if i running .can anyone suggest what to do

@ENVYxMajestics Месяц назад

I'm also facing same problem, pls someone solve these questions:)

@Pxxmprimegame 4 месяца назад

"Compilation error: 'i2s_read_bytes' was not declared in this scope " anyone help me please!!

@techiesms 4 месяца назад

Share the screenshot of the error on our WhatsApp 8200079034

@winn8888 4 месяца назад

Pretty sure this happens if you're using an esp32 board package version newer than 1.0.6. 'i2s_read_bytes' has been superseded in newer versions. If you downgrade to 1.0.6 using the board manager (see instructions in the video), that might help.

@antonpegov2745 4 месяца назад

Why we need two ESP? We have multicore esp’s now, RTOS for multitasking, so why still use this two big guys?

@SA-oj3bo 3 месяца назад

They even not work at the same time, I also wonder why he used 2 ESP's..

@shivendrapratapsingh5188 4 месяца назад

Sir also make a video on google bard AI

@techiesms 4 месяца назад

Do you have any links about their API?

@paulinbiju2611 3 месяца назад

Can we train the chatgpt api to answer our questions in a particular way

@saidevdhal 2 месяца назад

Ofc

@joshuaolaiya 4 месяца назад

Isn't there free api instead of using Google cloud and OpenAI chatgpy

@saidevdhal 2 месяца назад

I can make one 😉

@clips4399 4 месяца назад

How many times can we use chat gpt api with free credits, is that sufficient for my college project?

@saidevdhal 2 месяца назад

Matters on you how much you use it

@aliher6281 4 месяца назад

Is there any other alternative for google cloud?

@saidevdhal 2 месяца назад

Cloudflare

@katspro-techug7687 4 месяца назад

Good work am replicating it but got issue with Speech_to_text. I followed video all steps but you didn't talk about network_param.h file. About "getting the certificate for your region". Both set of code uploadd fine for both esp32 but for the case is speechToText my esp32 crash (keeps on restarting when trying to communicate with Google STT api) Error: Processing your Audio File Json daata--- Guru Meditation Error: Core 1 panic'ed (LoadProhibited). Exception was unhandled. Core 1 register dump: PC : 0x4000c677 PS : 0x00060a30 A0 : 0x800d1d11 A1 : 0x3ffb1c90 A2 : 0x00000000 A3 : 0x3f4002cb A4 : 0x3ffb1d08 A5 : 0x000001d5 A6 : 0x00000000 A7 : 0x3ffb86e0 A8 : 0x3ff44000 A9 : 0x00000000 A10 : 0x00002a00 A11 : 0x3ff49040 A12 : 0x08000000 A13 : 0x00000007 A14 : 0xffffffff A15 : 0x0000000a SAR : 0x0000001e EXCCAUSE: 0x0000001c EXCVADDR: 0x00000000 LBEG : 0x400014fd LEND : 0x4000150d LCOUNT : 0xffffffff ELF file SHA256: 0000000000000000 Backtrace: 0x4000c677:0x3ffb1c90 0x400d1d0e:0x3ffb1cb0 0x400d2893:0x3ffb1f90 0x400d8054:0x3ffb1fb0 0x400899f6:0x3ffb1fd0 Rebooting...

@ronniepavoni319 Месяц назад

Did you fix this error?

@nizaralshabibi3372 4 месяца назад

How much the amount does google cloud take for verification ?

@techiesms 4 месяца назад

INR 2

@nizaralshabibi3372 4 месяца назад

@@techiesms Do they take money from my credit card after registration ,or do I have to pay money after the 2 INR?

@saidevdhal 2 месяца назад

@@nizaralshabibi3372that's automatic the will take it until the verification completes then they will refund

@akshanshmohan2175 4 месяца назад

Unfortunately, it is using ir sensor to wake it up. You have to add wake up word in the project so that the assistant can wake up without ir sensor. _try to add it fast and provide us_

@saidevdhal 2 месяца назад

Easy 🙌🏻

@pen_master Месяц назад

@@saidevdhal hi can i get the code for text to speech and speech to text ,iam getting error from github code

@saidevdhal Месяц назад

@@pen_master Are you working on this project ? currently

@trinankurbera 4 месяца назад

I could bot find the gerber file so please upload it.

@saidevdhal 2 месяца назад

Same issues

@SelvaKumarD-bu9uk 3 месяца назад

In text to speech code i am getting an error in runtime which is memory.h no such file director is found plz help me out

@jatinpal2945 3 месяца назад

how did you solve this ?

@saidevdhal 2 месяца назад

Download the directory

@SelvaKumarD-bu9uk 2 месяца назад

can you plz help me to this project plz give me your contact @@saidevdhal

@jhansidurgasreechapala1237 Месяц назад

It's a great project so i tried it but i don't know what i have done wrong can you suggest me please.i am getting this error "[HTTPS] begin... [HTTPS] Connection established![HTTPS] GET... [HTTPS] GET... failed, error: connection refused". please help me

@ArjunSingh21bee049 Месяц назад

Have found the solution to this problem?? I am facing the same issue the speech to text part is working but the chatgpt response is indicating connection refused

@jhansidurgasreechapala1237 Месяц назад

@@ArjunSingh21bee049 no ,not yet

@RohitKumar-cz5rt Месяц назад

same error if you find yet let me know

@dsavaratkar4557 28 дней назад

Ya same error, plz if any one has solution for ir

@Robotroincs 16 дней назад

sir its api is free or paid

@Novi_1011 3 месяца назад

How much did you spend on this?

@sathiraperera6634 2 месяца назад

Record start! Processing your Audio File Json daata--what is the capital of India- Asking Chat GPT [HTTPS] begin... [HTTPS] Connection established![HTTPS] GET... -1 [HTTPS] GET... failed, error: connection refused To ask againPress button I am getting this error. i tried many ways to retify it , but fail. if anyone can provide a solution it will be vary helpful.

@ArjunSingh21bee049 Месяц назад

Have found the solution to this problem?? I am facing the same issue the speech to text part is working but the chatgpt response is indicating connection refused

@ENVYxMajestics Месяц назад

Yeah I am also facing the same problem.Can anyone solve these problems.

@fahimi.saiyed8939 4 месяца назад

sir raspberry pi 4 pe karona

@hillerhilsrangmarak6267 4 месяца назад

Sir how to get this sencor

@techiesms 4 месяца назад

You can purchase everything required to make this project from our website www.techiesms.com

@tenlittleindians 4 месяца назад

When they are able to make an offline version I'll be interested. Until then it's just more spyware monitoring by every move.

@saidevdhal 2 месяца назад

For now offline is not possible cause to make api request network is needed

@tenlittleindians 2 месяца назад

@@saidevdhal Lots of people have home networks and even home servers. You can make api requests without being on the Internet. Memory bank size limits and computer power to access huge chunks of memory are getting more affordable every day. That's what I see as the limiting factor currently.