€199 · 60-minute session · Written package afterwards
FIG. 01 / FROM REQUEST TO PLAN
THE STARTING POINT
“Can we stop remaking every customer quote?”
One process. Let's unpack it.
● ● ● PROCESS MAP 01
✦Read requestAI · extract
→
◇Review quoteYou · decide
↳
starting-brief.mdScope · first build · acceptance checks
Does this sound familiar?
You can see the friction. The next step is less obvious.
A request arrives by email. Someone copies it into a spreadsheet. Details are missing. The next person asks the same questions again.
You suspect software could help. You need to work out what to build, where AI adds value, and which decisions stay with your team.
01 / Make the work visible
First, draw the process. Then give each step a job.
This is the thinking I bring to my practical talks and the systems I build.
01
Start with a real request.
Trace one piece of work from the moment it arrives to the moment someone can use the result. Find the missing information, the repeated work and the decisions.
Our example: an enquiry becomes a reviewed customer quote.
02
Use AI where information varies.
AI can turn an unstructured request into a draft: what the customer needs, what is missing and what to ask next.
A draft is something a person can check.
03
Let code carry the rules.
Required fields, prices, permissions and delivery states need consistent behaviour. Define those rules once and make them testable.
AI can help write this code. The finished step may need no AI at all.
04
Keep judgment with a person.
Someone owns the promise made to the customer. Give them the context to review the quote, approve it and handle anything outside the normal path.
Record what was approved so the next person can continue.
CUSTOMER ENQUIRY → APPROVED QUOTEIllustrative
AI understands Code handles You decide
Enquiry received Email / form
Missing details? → Request clarification ↺
✓
Deliver the approved quotePDF + email · saved to the shared record
EXPLORE THE MAP
Select a step to see what belongs there and what happens when it goes wrong.
A readable preview. The package includes an editable BPMN process file.
One useful distinction
“Built with AI” can mean two things.
BUILD TIME
AI helps create the software.
A development assistant helps write a pricing calculator. Once built, code calculates the price.
AI + developer→Tested code
RUN TIME
AI does a step in the process.
A model reads each new customer email. Its draft is checked before anything is promised.
New request→AI draft→Review
02 / Built in practice
The thinking comes from building real systems.
Different businesses. Concrete decisions about how work should move.
PRODUCT CONFIGURATOR
A catalogue that knows what fits.
Model the components and valid combinations. From that shared model, generate a product code and the documents people need.
THE DESIGN DECISION
Compatibility belongs in a data model. People can look up the exact configuration behind a code.
Command Center makes agent activity visible. Clear task boundaries, review and shared context help the next contributor continue the work.
Anonymised activity view · handoff sequence is illustrativePart of my own working practice
Victor Lipov · business systems architect
Getting something built is one part. Deciding what deserves building comes first.
I design and build systems that help businesses run. I use AI in my own work. My practical talks explore where it helps, where ordinary code is enough, and where a person needs to make the call.
That is the perspective we apply to your process.
Workshop feedback
“What I appreciate about Victor is his practical approach to AI and automation. He focuses on understanding the business problem first and choosing the simplest solution that makes a difference. His workshop at Madeira Friends Hub reflected that mindset, with useful guidance on where AI adds value and where human judgment matters. I’d recommend Victor to anyone looking to simplify their operations and put technology to work with a clear business purpose.”
Pedro Calado · Workshop feedback, Madeira Friends Hub
03 / Something you can take forward
Your process map. Your starting brief.
A concrete package for one workflow, ready to discuss with a developer or use with your development assistant.
€199
One process. A clear starting point.
Before we meetYou share the process, an example input and the result you want.
60 minutes togetherWe map the work, choose the boundaries and agree a useful first version.
Within 2 business daysYou receive the edited map and starting brief, including assumptions and open questions. One follow-up clarification by email is included.
Designed for one bounded workflow. Software development, integration testing and a complete technical architecture are separate work.
For owners and operators who can show how the work happens today, and want a practical starting point for improving it.
Can I cancel?
You can cancel and receive a refund up to two business days before the session.
What should I bring?
One recurring workflow, an anonymised example of what comes in, and an example of the result you want. Screenshots or existing documents help. A polished specification is unnecessary.
What if AI is not useful for my process?
The map can recommend ordinary software or a simpler process change. We will make that choice explicit in the brief.
Can I build it myself with an AI assistant?
The starting prompt gives your assistant context, scope and acceptance checks. You will still need to review its work. The package is a starting point for development, not a guarantee of a finished application.
What if my idea is too large for one session?
We choose one useful workflow before the session. The package covers that slice and records the larger questions separately.
Can you build the system afterwards?
Yes, implementation can be discussed separately. You can also take the package to another developer.
