TRACK 01 / OSWORK

Fundamentals

Understand the system before choosing the tool.

0% 0 of 0
01 / OSWORKModels: choose by task02 / OSWORKChat, Work and Desktop03 / OSWORKPRACTICE04 / OSWORKCHECK
Read, apply to a small task and check the result before moving on.
2 modules
12 topics
60 min with practice
Guided step by step

Track map

1.1 ~30 min

🧭 Models: choose by task

Compare models with a real task and a quality criterion.

1.2 ~30 min

🧭 Chat, Work and Desktop

Write a work order with inputs, output, and review.

Detailed content

MODULE 1.1

Models: choose by task

Compare models with a real task and a quality criterion.

0% 0 of 0

What it is

A model generates responses; a system organizes how those responses become work. OSWork combines model, interface, files, instructions, tools, memory, and automations. Think of a small workshop: the professional’s skill matters, but tools, materials, and quality criteria also determine the outcome. We are not installing a new computer operating system: we use that expression as a metaphor for organization.

Why learn

Without this distinction, every failure turns into an attempt to write a larger prompt. Sometimes only the input file, a permission, or the way to verify the output is missing. Separating the pieces allows you to fix the right point.

Key concepts

Model reasons; interface receives the goal; files provide evidence; tools execute; you verify.

What it is

In the documentation consulted on 20/09/2026, the GPT-5.6 family includes Sol, Terra, and Luna. Sol offers the highest capacity in this family; Terra balances everyday work and cost; Luna favors clear and repeatable tasks. GPT-6 Astra appears as an option for complex work. Availability varies by account, client, authentication, and release.

Why learn

A model’s name does not guarantee it appears in your account. Choosing based on what is available avoids turning the lesson into a promise of access. No model eliminates the need to verify results.

Key concepts

Model family; capacity; speed; cost; availability.

What it is

The model is the chosen mechanism. Reasoning effort is a setting of that mechanism. Higher levels may consume more time and tokens, units of text processing. The selector names change between Chat, Work, and Codex; there is no single list of Instant, Medium, High, and Pro that represents all products.

Why learn

Requesting the maximum on every task can increase consumption without improving the result. Increasing effort also does not provide a missing document. First complete the inputs, then assess whether the problem requires more analysis.

Key concepts

Model and effort are different axes; start with the default; compare under equal conditions.

What it is

A rubric turns opinion into observation. Define before the request which facts must appear, which errors are unacceptable, and how the output will be used. Use three small criteria: fidelity to the data, combined format, and verifiability of the conclusions.

Why learn

Without criteria, you choose the prettiest answer. A report can sound convincing and alter values. The test should reflect the work you need to deliver, not a demonstration made to impress.

Key concepts

Acceptance; evidence; representative sample; controlled comparison.

What it is

Login with ChatGPT uses the rights and limits associated with the account and workspace. An API key uses consumption‑based billing on the platform. Subscribing to ChatGPT does not mean you receive free credit for any program that calls the API. Check the active method before a long execution.

Why learn

This precaution avoids discovering later that an experiment is consuming a different account. Usage limits, model access, and data rules can change; the source of truth is the current configuration of your account.

Key concepts

Subscription; authentication; API; consumption; spending limit.

What it is

Autonomy is an authorization with scope, not an invitation to do anything. Combine objective, permitted files, authorized actions, time limit, and stop condition. The output must include what was done and how it was verified.

Why learn

An agent can execute more steps than a chat, including making multiple mistakes. A simple limit protects time, budget, and files without preventing useful work. Technical permission and written instruction complement each other.

Key concepts

Scope; approval of external actions; execution ceiling; reviewable result.

View full →

MODULE 1.2

Chat, Work and Desktop

Write a work order with inputs, output, and review.

0% 0 of 0

What it is

Chat is a conversational interface. You present a question, receive an answer, and can adjust the request. It works well for clarifying a concept or drafting a message. Additional features vary: a chat can also work with files and tools when available. The pedagogical distinction is the task’s intent, not a technical prohibition.

Why learn

Recognizing a small need prevents building an automation for something that only requires an answer. The best environment is the one that lets you verify delivery with minimal friction.

Key concepts

Conversation; clarification; draft; human review.

What it is

Work allows delegating a task with a reviewable result, such as an analysis or a document. It can use approved files and tools. Instead of monitoring every sentence, you define what must exist at the end and track the relevant steps. Availability depends on the account and environment.

Why learn

Jobs with multiple inputs need a definition of done. Without it, the agent may keep researching when you only needed a comparison of three options.

Key concepts

Goal; sources; delivery; limits; stop condition.

What it is

Desktop means an application installed on the computer. In compatible environments, it can access folders and applications with permission. The presence of the application does not grant universal access to your documents. If a tool is unavailable, provide the files via a supported path.

Why learn

Many task failures are access failures: the learner assumes the agent sees a folder, but it wasn’t shared. Checking the context before execution prevents conclusions about documents the model never read.

Key concepts

Authorized folder; local access; tool available; read confirmation.

What it is

Local execution runs on your machine. Cloud execution runs on remote infrastructure. A local job depends on power, network, and computer permissions. Some environments offer cloud execution that continues without the machine being on; this depends on the functionality, not just the Work name.

Why learn

An automation isn’t permanent because it was started in a modern interface. You need to know where the process lives, where the files are, and which connections it depends on.

Key concepts

Execution location; persistence; file access; continuity.

What it is

A good request contains objective, context, inputs, constraints, result, and verification. These are fields of an order, not magic words. The easier it is to inspect the output, the easier it will be to detect a deviation before it becomes rework.

Why learn

The agent doesn’t need to guess whether you want an explanation, an editable file, or a publication. Naming the artifact and the ready condition shortens the distance between intention and execution.

Key concepts

Artifact; format; authorized sources; observable criterion.

What it is

A delivery only ends after the review. Open the file, compare numbers with the sources, and test the relevant links or formulas. Ask the agent for evidence of what it verified, distinguishing executed tests from suggested tests.

Why learn

The word “ready” does not demonstrate that the file opens or that all data were preserved. A small independent verification often finds more problems than a new generic improvement request.

Key concepts

Open; compare; test; record limits.

View full →
Full module