Building Browser-Native AI Runtimes:
A Hands-On Workshop with WebLLM, WASM
a Berlin-based engineering leader focused on distributed systems, serverless architectures, and AI-native applications. Iβve led cross-functional teams at Intel, Delivery Hero, Blacklane, and WΓΌrth Cloud Services shipping scalable platforms and shaping long-term technical strategy. My current interests revolve around Rust, WebAssembly, Synthetic Data, and in-browser AI.
Who am I?ππΌ
Where Does the Loop Close?
Portability, Isolation, and the Locality of Intelligence
γ
Feedback is older than AI.
β The Fictional History of Machinic ToolsThermostat
Measure β Compare β Adjust
CI/CD
Deploy β Observe β Roll Back
Compiler Optimization
Rewrite β Benchmark β Rewrite
Recommendation Systems
Behavior β Update β Suggest
Feedback Loops at their best
Observe β adjust β repeat.
An Agent Is a Loop With Memory
Input
The loop begins with a signal: user input, system state, or environmental data. entering the runtime boundary.
1.
Interpret
The system contextualizes the signal using its current memory and predictive model to construct a meaningful internal representation.
2.
Decide
Based on memory and prediction, the loop selects the next action either deterministic, probabilistic, or a combination of both.
3.
Act
The system executes the action, updates its memory with the outcome, and prepares for the next iteration of the loop.
4.
γ
Where does the loop run today?
β The History of Machinic Tools1st
Client
2nd
API
3rd
Cloud
5th
LLM
7th
Loop
4th
Tools
6th
Memory
The loop closes remotely today
βοΈ Cloud
π₯ Desktop
π± Mobile
π Edge
π§ Browser Tab
A loop can run in:
β’ Mozilla AI blueprints β Can the browser be a serious runtime?
β’ wasm-agents blueprint β Pyodide in-browser notebooks
β’ Built a PoC β wasm-browser-agents-blueprint
Thanks Davide from Mozilla.ai β€οΈ
How I Ended Up Testing This Idea
# Build and run with Docker (Recommended)
docker build -t wasm-browser-agents-app .
# For development with hot-reload
docker run -p 5173:5173 \
-v $(pwd)/demos:/app/demos \
-v $(pwd)/src:/app/src \
wasm-browser-agents-app
Demo π
Loop Component β Browser Primitive
| Loop Component | Browser Primitive |
|---|---|
| Memory (short-term) | JS Heap |
| Memory (long-term) | IndexedDB |
| Decision (prediction) | WebLLM |
| Deterministic Logic | WebAssembly |
| Parallel Execution | WebWorkers |
| Tool Access | Web APIs |
| Acceleration | WebGPU |
| Isolation Boundary | Workers + WASM sandbox |
Isolation is
Structural
Isolation is Structural
Determinism Inside the Loop
If the loop is real, parts of it must be deterministic, too.
γ
Loops Shouldn't Care About Language
β An unwritten book on WASM Agents The browser used to host one language
SM made it portable.
Demo π
Conceptual
Reconstruction
Same loop,
different placement.
The Six Taxes of Local Loops
When You Close the Loop Locally, You Inherit New Taxes
Tax #1
Memory
Local loops are bounded by device memory.
Tax #2
Distribution
Portability increases distribution cost.
Tax #3
Variance
Portable loops behave differently on every machine.
Tax #4
Energy
Energy is part of loop economics.
Tax #5
Heat
Feedback exists at the hardware layer too.
Tax #6
Context
Context window defines the visible past of the loop.
γ
Constraints are not reasons to avoid locality entirely. They're the cost of moving closer to the user.
β The History of Machinic ToolsStrategic Placement of Loops
Local Loop
private, low latency
1.π§π±π₯
Edge Loop
Shared
2.π
Cloud Loop
heavy compute
3.βοΈ
Where Does the Decision Become Real?
Intelligence is not where the model is
It's where the loop closes.
The browser is a serious place for loops to close.
So for agents, too.
Key Takeaways
-
Agents are feedback loops with memory not personalities. π
-
Loop placement matters more than model size. π
-
The browser already contains the primitives to host a loop. π
-
Isolation (WebWorkers) enables stability. π§±
-
Portability (WASM) enables polyglot architectures. π§³
-
Prediction should be paired with deterministic control. βοΈ
-
Energy, latency, and memory are part of loop economics. β‘
-
Moving the loop changes privacy, cost, and control. π
Thank you!
Contact me via email or social if you have any questions,always happy to help!
