v1.0
Building with LLMs
PROGRESS
·
WORDS
writ
ding
by
with
Buil
hehe
LLMs
me
ten
REQUEST
TOKENS
MODEL
CORPUS
TRAINING
LOOP
AGENT
HARNESS
read_file
240 lines
run_tests
3 failed
UX
add dark mode
why is this slow?
write the tests
writ
ding
by
with
Buil
hehe
LLMs
me
ten
REQUEST
TOKENS
MODEL
CORPUS
TRAINING
PART 01
The Model
1.1
How LLMs work.
0 WORDS
TODO
1.2
Context window.
0 WORDS
TODO
1.3
Variance.
0 WORDS
TODO
1.4
Tokens and cost.
0 WORDS
TODO
1.5
Structured output.
0 WORDS
TODO
LOOP
AGENT
HARNESS
PART 02
The Agent
2.1
Anatomy.
1.2K WORDS
IN PROGRESS
2.2
Loop.
0 WORDS
TODO
2.3
Tools.
0 WORDS
TODO
2.4
History.
2.4K WORDS
IN PROGRESS
2.5
Context.
0 WORDS
TODO
2.6
Steering.
0 WORDS
TODO
2.7
Artifacts.
0 WORDS
TODO
2.8
Surfaces.
0 WORDS
TODO
2.9
Orchestration.
0 WORDS
TODO
2.10
Tradeoffs.
1.7K WORDS
DONE
UX
PART 03
The Application
3.1
Chat UX.
3.5K WORDS
DONE
3.2
Building evals.
0 WORDS
TODO
3.3
Tracing and logging.
0 WORDS
TODO
3.4
Safety and guardrails.
0 WORDS
TODO
3.5
Cost and latency.
0 WORDS
TODO
3.6
Deployment and rollout.
0 WORDS
TODO
3.7
Coding agent internals.
0 WORDS
TODO