Open training · start here
From prompts to reusable ways of working.
Learn the foundations of how we work with AI: first how to instruct one task, then how to turn successful methods into workflows that can be used again.
5 parts · 17 chapters
The full path from first prompt to verified result
All 17 chapters are open. Follow the complete learning journey in sequence or go directly to the subject you need.
Foundational language
Chapters 1–3
Understand the building blocks and choose the right workspace.
- 01Open · start herePrompt and metapromptSeparate the task from the instruction that shapes the task.Open the chapter →
- 02OpenSkill and metaskillMake a successful workflow reusable and controllable.Open the chapter →
- 03OpenChatGPT and CodexMove from conversation and role to an agent executing in a project.Open the chapter →
From request to task
Chapters 4–7
Give AI the right context, objective, plan and working contract.
- 04OpenContext and sourcesUnderstand what AI sees, what is missing and what must be checked.Open the chapter →
- 05OpenGoals, scope and completion criteriaTurn a loose request into an outcome that can be evaluated.Open the chapter →
- 06OpenFrom request to executable planDivide the work into verifiable steps with visible progress.Open the chapter →
- 07OpenYAML contracts and living worklistsPreserve requirements, dependencies, tests and evidence throughout large jobs.Open the chapter →
Work in a real project
Chapters 8–10
Make changes in a controlled, reversible and auditable way.
- 08OpenTools, files, Git and project rulesSee how Codex discovers instructions, skills and agent roles in a folder tree.Open the chapter →
- 09OpenSmall and reversible changesBuild with an explicit preflight, backup, stopping point and rollback.Open the chapter →
- 10OpenTesting, verification and evidenceSeparate claims from checks that actually demonstrate the result.Open the chapter →
Safe and durable collaboration
Chapters 11–14
Preserve boundaries, continuity and clear ownership over time.
- 11OpenAccess, secrets, boundaries and responsibilityGrant only necessary authority and keep safety boundaries visible.Open the chapter →
- 12OpenCognitive memory and checkpointsPreserve verified project context without treating history as truth.Open the chapter →
- 13OpenOne task, one ownerCoordinate agents and subagents without unclear responsibility or collisions.Open the chapter →
- 14OpenAutomation and recurring workAutomate only when the workflow, boundaries and notification rules are clear.Open the chapter →
Build your own way of working
Chapters 15–17
Combine the components into methods you can test and reuse.
- 15OpenBuild your first skillPackage a trigger, method, resources, boundaries and test into one workflow.Open the chapter →
- 16OpenBuild a metaskill and quality controlCreate a method that builds and improves other skills.Open the chapter →
- 17OpenFrom conversation to verified resultComplete an entire task with a contract, agent, tests and evidence.Open the chapter →
How the guide works
Every chapter moves from understanding to practical use
Plain explanation
The concept is explained without assuming a technical background.
Side by side
The differences become visible before the details are introduced.
Concrete example
The same task is shown at several levels of maturity.
Try it yourself
A small exercise makes the knowledge usable.
Technical guides and demonstrations
Apply the foundations to real technical subjects
These in-depth resources are currently available in Norwegian. They are not required to complete the English foundation course.
API integrations from first principles
Contracts, authentication, validation, idempotency, retries, logging and testing, with a local API laboratory.
SIM900, Arduino and Visuino
A source-backed review of testing and verification in a published SMS-controlled output project.
Open source code and technical evidence
Explore public code projects, licenses, maturity and the exact version referenced by every description.
The principle behind Finnandre Open
The knowledge is open. You can commission the implementation.
Open methods. Verifiable evidence. Protected people. We show the calculator – not the customer’s calculation.
For people who want to understand and do the work
Use the explanations, examples, exercises and checks without purchasing a service.
For people who want the solution implemented
Finnandre can map, adapt, test and deliver within an agreed first phase, price range and stopping point.