LOCKEDIN / FORWARD-DEPLOYED ENGINEERING 101
Understand it.
Then build it.
You can be an experienced engineer and still be new to agents. Start with what each thing is and how it helps a person do useful work. Explore 58 original introductions across software, agents, cloud, security and delivery. Start here without a model key; each topic connects to deeper instruction and practice.
A place to start.
Choose a short reading route, or look up a concept below.
- 101UnderstandPlain language, visual walkthrough, vocabulary
- 102Build with guidanceOriginal lessons, examples and knowledge checks
- 103Practise and explainRunnable labs, failure cases and handover
Understand agents
Decide what information reaches the model for this particular request.
See the 8-concept route
For engineers who know software but are new to AI applications. Open the related definitions whenever a term is unfamiliar.
- 01The modelStart with what produces the text. Separate the model from the application using it.
- 02TokensSee the units a model reads and produces before reasoning about its input limits.
- 03ContextYou are here
- 04The agent loopPut the model inside an application that can execute permitted actions and check the result.
- 05WorkflowsSeparate a fixed sequence from a model-selected next action before choosing a framework.
- 06LangChainSee how a framework assembles the model, tools and application-controlled context.
- 07LangGraphMake the steps and shared state explicit when work needs to branch, wait or resume.
- 08MCPConnect capabilities through a shared protocol while keeping permission checks in the application and server.
Bookmark this page to return to this concept. The number shows your place in the reading route; it is not an assessment result.
Selecting what a model sees
What is context management?
Context is the information available to a model for a particular request: instructions, conversation, tool definitions and supplied evidence. Context management selects, orders, limits and refreshes that information. It is not the same as retraining the model or storing permanent memory.
FDEs need repeatable decisions based on current, authorized information rather than an accidental conversation history.
UNDERSTAND THE SYSTEM
Context is the working material for this call
Context is what the model can use during a request: instructions, messages, selected evidence and tool results. It is not the same as training a model or keeping a durable database.
Read the connections in plain English
- 01Available sources → Select and scope
Retrieve candidates without crossing access boundaries.
- 02Select and scope → This call's context
Select the smallest sufficient set of current evidence.
- 03This call's context → Model call
Treat retrieved text as evidence, not higher-priority instructions.
- 04Model call → Durable record
Application code persists only validated results that should survive.
Check your understanding: Does adding more text always improve an answer?
No. Irrelevant or conflicting text can distract the model and consume budget. Choose relevant, authorized, current evidence and keep durable state in the application.
CONNECT IT TO WHAT YOU KNOW
Start with something familiar.
Give a colleague the relevant incident packet before asking for a decision. A huge archive can bury the current fact; a short summary can omit the fact that changes the decision.
More context is not automatically better. Keep provenance, access scope and freshness when trimming or summarizing, and treat retrieved instructions as untrusted data.
Words worth knowing
- Prompt
- Input instructions and content supplied for a model call.
- Compaction
- Reducing a conversation or evidence packet while retaining important information.
- Provenance
- Where a claim or piece of data came from.
Useful foundations
Open a definition and diagram here. Your lesson stays in place.
A trained statistical modelWhat is a large language model?
QUICK REFERENCE / 101
What is a large language model?
A large language model (LLM) is a trained neural network. Its learned numerical parameters and current input are used to calculate possible next tokens; a decoding process selects the continuation. A model can produce text, code or structured proposals. The application around it supplies context, decides what tools are available and controls real-world actions.
Choosing a model is one decision inside an application design, not the whole design.
Units of model input and outputWhat are tokens and token budgets?
QUICK REFERENCE / 101
What are tokens and token budgets?
Tokens are the units a model uses to represent content. A token can be a word, part of a word or another encoded element. Token counts affect context capacity, output limits, latency and often price. Characters, words and tokens are not interchangeable counts.
Unbounded loops and large tool responses can exhaust cost and capacity before useful work finishes.
ONE EXAMPLE / FICTIONAL CASE
Follow the connections.
A previous conversation says a pump is safe, but a newly approved procedure says it must remain offline.
UNDERSTAND THE SYSTEM
Context is the working material for this call
Context is what the model can use during a request: instructions, messages, selected evidence and tool results. It is not the same as training a model or keeping a durable database.
See what moves and why.
Read the connections in plain English
- 01Available sources → Select and scope
Retrieve candidates without crossing access boundaries.
- 02Select and scope → This call's context
Select the smallest sufficient set of current evidence.
- 03This call's context → Model call
Treat retrieved text as evidence, not higher-priority instructions.
- 04Model call → Durable record
Application code persists only validated results that should survive.
Check your understanding: Does adding more text always improve an answer?
No. Irrelevant or conflicting text can distract the model and consume budget. Choose relevant, authorized, current evidence and keep durable state in the application.
Work through it, step by step.
Select
Choose the specific task instructions and currently authorized evidence.
What you have now: A relevant packet
Budget
Reserve room for the answer and necessary tool turns. Remove irrelevant repetition, not critical constraints.
What you have now: A bounded context
Refresh
Recheck changing records and invalidate stale summaries before a consequential decision.
What you have now: A current evidence basis
FROM RECOGNITION TO REASONING
Build a small context packet for a fictional incident update.
- 01
Run the free context lab and compare full, compact and lossy packets.
- 02
Identify the restriction and deadline that the shortest packet loses.
- 03
Advance the source and rebuild the packet instead of changing its version label.
Five observed packets, a diagram and one controlled repair.
Compare your reasoning when you are ready
A shorter packet can lose required facts, and a complete packet can be stale. Preserve the source identity and re-read its current contents before relying on the summary.
Keep notes in your own working copy. Moving through these cards does not record completion.
TURN THE IDEA INTO USEFUL WORK
You can now go deeper.
More context is not automatically better. Keep provenance, access scope and freshness when trimming or summarizing, and treat retrieved instructions as untrusted data.
Academy lessons use your existing training access. Free guides stay open.