Prototyping a business workflow

How can you turn an internal request workflow into a verifiable prototype with Lovable?

  • Lovable
  • Claude Code
  • Codex

1960 € per session, excl. VAT

In short

This 7-hour day, excluding breaks, takes you from a business need to a testable prototype in Lovable. You formalise the rules, structure the data, evolve the code through instructions and check the user journeys. Claude Code or Codex can support a targeted review, depending on authorised access and the official documentation.

Why this training?

A convincing interface can conceal inconsistent lending rules, unnecessary fields or poorly defined permissions. Formalising states and transitions before generation avoids confusing a visual demonstration with business functionality. Generating and modifying code from instructions speeds up experimentation, but replaces neither human review nor testing. The day links each change to an explicit need and evidence of verification. Running case study: The fictional organisation Avelune Équipements manages equipment loan requests through scattered messages. Its team wants to centralise requests, approval and returns. The prototype must prevent approval without consistent dates and distinguish the requester’s actions from those of the manager. Transfer to the workplace: Apply the rule specification to an authorised internal process, organise a review with the technical lead and plan a limited pilot after resolving blocking issues, without assuming deployment to production.

Objectives of this training

  • Define the scope of a lending workflow that can be implemented as a prototype.
  • Model the data, states and expected permissions.
  • Write precise generation and modification instructions.
  • Test standard journeys, errors and access restrictions.
  • Measure rule coverage and prioritise corrections.

Who is it for and what are the prerequisites?

Roles and tasks concerned

  • Operational managers responsible for simplifying an internal request workflow.
  • Business project managers and digital leads supporting AI-assisted prototypes.

Prior skills

  • Have experience using an AI assistant and know how to break down a complex request.
  • Be familiar with a business process and its common exceptions.
  • Be able to read a simple code snippet and understand form validation.

What you will work on

  1. Define the lending workflow

    Objective: Establish the priority journey and its exclusions. Actors, events, pain points and acceptance criteria; distinction between a prototype and an operational application.

  2. Structure data and rules

    Objective: Make the requirements verifiable. Equipment and request entities, mandatory fields, statuses, transitions and permissions matrix.

  3. Generate a first version

    Objective: Obtain a coherent demonstration journey. Contextualised instructions in Lovable, small increments and verification of generated elements.

  4. Review and modify the code

    Objective: Correct without unintentionally expanding the scope. Human review, targeted changes with Lovable or Claude Code or Codex; secrets management according to the official documentation.

  5. Run business acceptance tests

    Objective: Measure the observed quality of the prototype. Expected results, negative tests, authorisation checks and regression testing; limitations of purely visual restrictions.

  6. Prepare the pilot decision

    Objective: Present the evidence and reservations. Rule coverage, defect severity, technical approvals and conditions for a limited trial.

What you take away

  • Scope and rules specification with a role matrix.
  • Demonstration prototype and fictional dataset.
  • Log of instructions, changes and review decisions.
  • Acceptance test specification with results and prioritised issues.
  • Pilot plan specifying approvals and remaining limitations.

Method and assessment

  • Short explanations followed by immediate practice on the running case study.
  • Individual work followed by pair work, with peer review of the outputs.
  • Guided group discussion: what works, what needs correcting and what remains a human decision.
  • Preparation: Check access to Lovable and, if selected, Claude Code or Codex in the environments authorised by the organisation. Prepare only fictional or anonymised cases and consult the official documentation for capabilities, limitations and access conditions.

Observable criteria: Each status transition has an explicit rule and an associated test. ; The request-approval-return journey is demonstrated using fictional data. ; Negative tests cover inconsistent dates and prohibited actions. ; No secrets appear in the instructions, data or shared code. ; The summary reports tests passed against tests executed, rules covered and blocking issues..

Frequently asked questions

Do we need to use all three tools?

No. Lovable is the main tool; Claude Code or Codex remain optional for a targeted review.

Can we work on our real requests?

Only after anonymisation and authorisation; the supplied case uses entirely fictional data.

Will the prototype be production-ready?

This is not promised. Security, hosting, operations and compliance require additional approvals.

How is the result assessed?

Through reproducible tests linked to business rules, rather than the appearance of the screens alone.

Last updated: