You have an idea
“I know what needs improving, but I cannot write a software brief.”
AI DISCOVERY / APPROACH IN DEVELOPMENT
From an idea to a brief
you can understand.
Tell us what you need. AI asks about the important details, shows how it understands your process, and helps you approve the brief before development begins.
- 01 You explain your work
- 02 AI maps it out
- 03 You confirm it is right
YOUR WORK, IN YOUR WORDS
You know the problem.
You do not need to know IT.
You know the daily routine
“Today we handle it through email, Excel or manual work.”
You want clarity
“I want to check that we understand each other before development.”
FROM CONVERSATION TO DECISION
Six steps. You stay informed.
No hundreds of questions at once. Just small batches, relevant documents and visual summaries as understanding develops.
Tell us what you need.
In everyday language, by text or voice. No technical vocabulary needed.
Understand how you work today.
AI asks about your actual process (AS-IS) and analyses relevant documents.
Describe the future process.
Together, you identify what should change (TO-BE) and why it matters to your business.
See progress as it happens.
AI draws diagrams, summarises findings and keeps risks separate from open questions.
Approve the direction. Try the design.
A clickable prototype follows only after you approve the AS-IS and TO-BE analysis.
Approve the brief.
You receive an understandable brief; the next project roles receive structured materials.
ILLUSTRATIVE SCENARIO / INVOICES
First understand today. Then plan change.
This illustrative invoice example shows the current process (AS-IS) and a possible future process (TO-BE). Both descriptions are fictional and explain the approach.
Today · AS-IS
- An invoice arrives by email
- An employee copies it into Excel
- Approval is handled manually
Future process · TO-BE
- Invoices enter a single process
- Details are prefilled and checked by a person
- Approval and cover arrangements are defined
Even without an application, you already have a way of doing the work. That is what we need to understand.
An example of a possible design, not a confirmed product feature or a real client project.
TRY THE PROCESS
A small question leads to a clearer brief.
Illustrative demo using fictional data. It does not use real AI.
Nothing is sent or stored. The full project prototype is created only after Discovery is complete.
ONGOING FEEDBACK
Know where things stand.
What we have learned
A short summary you can correct.
Process diagram
AI draws it. You comment on it or confirm it is correct.
Risks
What could cause a problem, such as an absent approver.
Open points
What still needs clarification or a decision, such as who covers for the approver.
YOUR DELIVERABLES
A foundation for the next step.
With the details made clear.
Current process · AS-IS
A structured description and sequence diagram. What works and what does not.
Future process · TO-BE
A structured description, sequence diagram, business goals and expected changes.
Solution options
Usually 1–2 options, at most 3. Pros, cons, related risks and open points.
High-level estimate
Effort in person-days (MD), with calendar duration stated separately in weeks or months. A non-binding high-level estimate, not a quote or an exact calculation.
Clickable prototype
4–5 main scenarios selected by business value. Test data and partly simulated features. Not a production application or a complete implementation.
Decision record
Risks, open points and approval records you can read without technical expertise.
EXPLORE THE DELIVERABLES
This is what your brief could look like.
An illustrative deliverable package for a fictional invoice approval process. It shows the format and structure of the documents, not a real client delivery. Expand each example to explore it.
01 · Current process (AS-IS) and sequence diagram
SAMPLE DOCUMENT / CURRENT WORK
How invoices are approved today
An invoice arrives in a shared inbox. The accountant copies the details into Excel and forwards it to the manager. After receiving a reply, the accountant records the result.
- Supplier → AccountantSends the invoice by email.
- Accountant → RegisterManually records the details in Excel.
- Accountant → ManagerForwards the approval request.
- Manager → AccountantReplies by email. The accountant records the result.
What works
People know email and spreadsheets. They can find the conversation about the latest invoice.
What does not work
Status is spread across different places. Cover is unclear when the manager is absent.
02 · Future process (TO-BE) and sequence diagram
SAMPLE DESIGN / FUTURE WORK
One overview, clear responsibilities
Business goal: the accountant can see who owns the invoice and its status without searching through emails.
- Intake → RegisterAn invoice record is created with prefilled details.
- Register → AccountantThe accountant checks the details and assigns an approver.
- Accountant → Manager / deputyRoutes the invoice for approval under the agreed rules.
- Approver → RegisterApproves or returns the invoice with a reason. The status is visible.
Expected change: a shared overview replaces manual status searches. Prefilling and cover rules still need to be verified.
03 · Solution options and trade-offs
Option A · Improve the existing process
A shared register, consistent rules and an assigned deputy.
Pro: people keep using familiar tools.
Con: some work remains manual.
Risk: the register may not match the email conversation.
Open point: who will keep the register up to date?
Option B · A simple approval tool
A shared invoice queue with assigned owners and statuses.
Pro: status and decisions in one place.
Con: a new tool must be introduced.
Risk: an unverified accounting integration.
Open point: what integration options does the accounting system support?
04 · High-level effort and calendar duration
Two different measures, clear assumptions
- Effort · person-days (MD)
- The amount of team work. A specific range is added only after the analysis.
- Calendar duration · weeks / months
- Time from start to handover, allowing for availability and dependencies.
- Estimate assumptions
- A selected option, a verified accounting integration and agreed approval rules.
A non-binding high-level estimate, not a quote or an exact calculation. This example deliberately contains no invented figures.
05 · Clickable prototype brief
Four main scenarios to try
- The accountant opens an incoming invoice and checks its details.
- The manager approves the invoice.
- The manager returns it for clarification with a reason.
- A deputy takes over when the manager is absent.
What the customer checks: they understand the steps, can see the status and can correct a mistake.
The prototype is created only after Discovery is complete and the analysis is approved. It uses test data and simulates some features. It does not connect to real accounting systems or make payments.
Try a simplified approval scenario ↑06 · Risks, open points and approval record
Risk · a possible problem
The invoice remains undecided when the approver is absent.
Proposed mitigation: assign a deputy and define a handover rule.
Open point · a decision still needed
Who covers for the manager, and for which invoices?
Next step: the customer defines responsibility and any limits.
Sample approval record
- Current process: confirmed by the accountant in the fictional scenario.
- Target direction: a shared overview and human review of details.
- To resolve: cover rules and accounting integration.
- Final brief approval: pending resolution of open points and prototype validation.
All roles, statuses and decisions shown are fictional. This is not a real customer approval.
You approve a clear deliverable. The sales and analysis roles that follow receive structured artifacts.
COLLABORATION, NOT AUTOPILOT
AI prepares. You decide.
AI prepares
Targeted questions, document analysis, diagrams, summaries, proposals, risk identification and a demo prototype.
You decide
Whether AI understands the reality, which future process you want, which option you prefer and what you ultimately approve.
Initial confirmation can be a simple “yes” by text or voice. It checks understanding; it is not a legal signature.
FREQUENTLY ASKED QUESTIONS
Before you begin.
Do I need to understand IT?
No. Just describe your work and what you need.
What if we do not have an application today?
We map the manual or informal process. An AS-IS process exists even without software.
Will AI ask me hundreds of questions?
Questions are focused and gradual, in small batches. AI also uses the documents you provide; you see summaries and diagrams along the way.
Do I have to draw diagrams?
No. AI creates them; you comment on or approve them.
Will I receive a finished application?
No. Discovery delivers an approved brief and a partly functional demo prototype using non-production data.
What if I do not want to talk to AI?
If you prefer not to talk to AI, we will send a sales representative to meet you, discuss your needs and help finalise your brief. You can also provide the initial information in a structured document.
FIRST UNDERSTAND. THEN DEVELOP.
Give your idea
a clear direction.
AI Discovery is an approach in development. This page demonstrates the process, not an available production application.