Technology · module 20 of 100
Cloud or local: where does your AI run?
Understand where your data goes and what trade-offs you must make.
After this lesson
You map the data flow and compare cloud-AI, local AI, and a hybrid setup.
In view · Side by side
Where does the calculation happen?
Your device sends input to an external server and receives an answer.
The model calculates on your own computer or own server.
Also check searching, logging, storage and synchronisation.
Part 1 of 3 · Cloud route
Your preference for less movement is on. Choose each part yourself.
Cloud: relying on a server elsewhere
With cloud-AI, your input goes via a connection to a service provider's servers, where the model performs the calculation. The answer comes back to your app. You need little local computing hardware, but you are dependent on the service, connection, and agreements regarding storage and data usage. A paid service is not automatically suitable for every kind of confidential information.
Local: relying on your own device
With a locally running model, the calculation takes place on your own computer or own server. After downloading, such a model can sometimes work without internet, as long as the functions used remain local. You need sufficient memory and processing power and manage updates and security yourself. "Local" does not automatically mean private: an app may still use external search services, synchronisation, or telemetry.
Choose based on the complete data flow
A hybrid application can search locally and still send text to a cloud model, or vice versa. Therefore, look at every step: input, model, search, storage, logging, and sharing. Compare quality, total costs, latency, management, and data requirements using the same fictitious trial task. Simply placing the model locally does not prove that the entire chain is offline.
New practice example
From explanation to a usable result.
Local refers to where the model calculates. Search services, storage, and synchronisation can still be external.
Fictional practice situationFictitious app: the language model calculates on a laptop, documents are in online storage, and a search query uses an external search service.
View a possible solution
This is how it can be done
Model calculation: local. Document storage: external. Search: external. The app is therefore not a fully offline workflow, even if the model runs on the laptop.
Why this works
You assess the entire data route. When checking a real app, also verify settings, logging, updates, and permissions.
Make it slightly more difficult
In the practice lab, only enable online storage. What conclusion regarding "everything stays on my device" changes?
Extra practice, at your own pace. You do not need to submit anything. This does not count towards an automatic assessment.
Watch and understand · with narration
Where does your prompt go?
With a computer-generated voice, on-screen text and captions in your chosen language. Start and pause whenever you like. Open the video separately.
Read the text version of the video
- In the cloud, your input travels through a connection to an external server. The model runs there.
- The answer returns to your device. Check the terms for storage and the use of your data.
- Locally, the model runs on your own computer or server. This needs memory and computing power.
- Look at the whole chain. A local app can still use external search services or synchronisation.
Your workplace · learning by doing
Make something. See what changes.
Checking the AI connection… You can already view the written example.
Present your assignment to Ende.
Read the answer, change one thing, and compare.
Previously held
New version
You remain in control
What do you want to improve in your next attempt?
On this device only. AI answers remain only in this page memory. This is self-review, not a grade.
Why this approach works
You assess the entire data route. When checking a real app, also verify settings, logging, updates, and permissions.
A step further
In the practice lab, only enable online storage. What conclusion regarding "everything stays on my device" changes?
Try & discover · no obligation
Follow the route of your data
Let the model calculate locally and then switch on online search. Does the entire application remain local?
- Model calculationOwn deviceNo external step
- SearchNo external search stepNo external step
- StorageOn the deviceNo external step
In this practice model, calculation, searching and storage remain on the device. In a real application, you also check logging, updates and telemetry.
This trial works locally with fictitious data and does not count towards your lesson completion. It starts anew with a new lesson or reload.
A prompt to get you started
Help a fictitious team choose between cloud-AI and a model on their own computer. Do not use real files. Compare data flow, required hardware, connection, management, quality, and total costs. First, make clear which requirements are still missing. For both options, draw the route input → model → answer and name any external services. Do not provide a general privacy guarantee.
Apply it
An app runs the language model on your own computer. The search function sends the text of your query to an external search service.
How do you describe this setup?
- Fully offline, because the language model is local.
- A local model with an external data flow; check the search service and which text is being sent.
- Automatically suitable for every confidential file.
With JavaScript, you can check your approach and receive immediate feedback.
Test your insight
What proves that a AI application works completely offline?
- Only the word "local" on the website.
- Only the fact that you have installed an app.
- That all components used work without an external connection; including search and storage.
Extra practice
Draw the input, calculation, and storage locations for a AI tool you use. Describe one cloud advantage, one local advantage, and one control point. How can an application with a local model still send data?
Cloud requires less local computing hardware; local gives more control over where the model calculation takes place. I also check search functions, synchronisation, and logging. A local model can still send text to an external search service. Therefore, I test the complete data flow and read the settings and documentation.
