Track
Builders
You build the thing that has to survive production.
The augmented LLM
Guide · 3 minA raw prompt in a playground creates an illusion. In production the working unit is an augmented model — wrapped with memory, tools, and context.How RAG actually works
9 minPrompt engineering from zero: write a formal letter, not a text message
6 minZero-shot, few-shot, or chain-of-thought: picking the right technique
5 minFrom prompt engineering to context engineering
4 minWorking with LLM APIs: first calls, tokens, and structured output
6 minHow agents remember: memory and knowledge representation
6 minStep 3 of 5
It acts
Tools and function calls, and where letting it act goes wrong.
6 pieces · See all 6 →
The agent that acts (and when not to)
Guide · 3 minAn LLM answers questions; an agent takes action. Knowing when to hand over control — and when to hold it back — decides whether the system stays stable.How an AI agent thinks
5 minOne agent or many? When each one fits
7 minThree lessons the human mind offers agent architecture
5 minFrom worker to specialist: an agent that owns a domain
4 minYou might not need an agent
8 minWhen one agent isn't enough
Guide · 4 minWhen a single agent buckles under the weight of too many responsibilities, you have to split the work — without losing the shared context.The Coordinator, the Worker, and the Delegator
8 minEvent-driven by design: agent teams that don't lose messages
6 minTracing a request through a multi-agent system
5 minSave first, then publish: a simple rule for not losing work
5 minCognitive Least Privilege: your agent should know only what it needs
4 minDefending against prompt injection and jailbreaks — and reducing hallucination
7 minHow agent teams coordinate: a guide to choosing the right shape
10 minStep 5 of 5
You can trust it
Evaluation, guardrails, and the cost of being wrong.
20 pieces · See all 20 →
Run it for real
Guide · 3 minMoving from local prototype to production serving — how to select, adapt, and run models once cost, latency, and reliability actually matter.Fine-tune, RAG, or prompt: which one, and what each costs
8 minPagedAttention and continuous batching: how one server answers more users
5 minStop ranking LLMs, start profiling them
6 minWhy LoRA works: the intrinsic-dimensionality story
4 minHow LLM inference actually works: prefill vs decode
4 minA practical checklist for picking an LLM for your feature
4 minSee all 20 pieces
Step 5 · You can trust it →Deciding, not building? Open the Business hub →