IT Let’s talk

Two tools to decide with numbers

A calculator to estimate how much time and value an automation frees up, and a ten-question check-up on the security of AI systems, based on the OWASP Top 10 for LLM applications. Free, no sign-up: what you enter stays in your browser.

ROI calculator AI security check-up

What is automating a process worth?

Think of a repetitive task: data entry, answering the same questions, reports, moving data between systems. Enter your numbers and see how much time comes back.

Rarely 100%: checks and edge cases stay with people.
Full cost to the company, not net pay.
Optional. Used to estimate the payback time.

The starting values are only an example: replace them with yours.

414hours freed every year

€12,420yearly value of the time freed

  • Time that comes back
  • Time that stays with people
Today
690 hours a year
After automation
276 hours a year

With an investment of up to €12,420 the project pays for itself within twelve months.

Let’s talk about this process

Hours freed = people × hours per week × weeks × automatable share. Value = hours freed × hourly cost.

This is a rough estimate: it leaves out running and maintenance costs, and indirect benefits such as fewer errors and faster replies.

Is your AI system protected against the most common risks?

Ten questions, one for each item of the OWASP Top 10 for LLM applications. It applies to chatbots, assistants and AI agents, live or planned.

  1. Does the system treat text it reads from emails, documents and web pages as data, not as instructions? LLM01, Prompt Injection
  2. Do you know exactly which personal or confidential data goes into prompts and where it is stored? LLM02, Sensitive Information Disclosure
  3. Do you keep an up-to-date list of the models, libraries, plugins and providers in use, with their data terms? LLM03, Supply Chain
  4. Do the documents and data feeding the AI come only from controlled, approved sources? LLM04, Data and Model Poisoning
  5. Are model outputs validated before they reach web pages, queries, commands or other systems? LLM05, Improper Output Handling
  6. For important actions such as payments, sending and deleting, does the agent use minimal permissions and ask a person to confirm? LLM06, Excessive Agency
  7. Is the system prompt free of passwords, keys and rules that must stay secret? LLM07, System Prompt Leakage
  8. If you use a knowledge base (RAG), does each user get answers only from the documents they are allowed to see? LLM08, Vector and Embedding Weaknesses
  9. On critical topics, do answers cite their sources and get a human check before they become decisions? LLM09, Misinformation
  10. Do you have usage and spending limits, with alerts, to prevent abuse and runaway costs? LLM10, Unbounded Consumption
  • In place
  • To check
  • Unanswered

0controls in place out of 10

0 of 10 answered

Answer the questions: the result updates with every answer.

Let’s talk about your AI system’s security

This is a rough self-assessment: it does not replace a penetration test or a risk assessment. Your answers stay in your browser.

Source: OWASP Top 10 for LLM Applications 2025

Frequently asked questions

How do you calculate the return on an automation?

Start from the time the task takes today: people involved times hours per week times working weeks. Multiply by the share the software can take over and by the cost of one hour of work to the company: the result is the yearly value of the time freed. Dividing the investment by the monthly value gives the payback time. It is a cautious estimate, because it leaves out indirect benefits such as fewer errors.

Does the AI security check-up replace a penetration test?

No. The check-up is a self-assessment that shows where to look, based on the ten items of the OWASP Top 10 for LLM applications. A penetration test verifies in practice, with controlled attacks, whether those controls really hold.

Is the data I enter saved or sent anywhere?

No. The calculations run in your browser: numbers and answers are neither sent nor saved. We only receive data if you choose to write to us through the contact form.

Tell us what you want to build.

Write a few lines about your project: we will reply personally to set up a first call.

What happens next

  1. We read your request and reply personally.
  2. We set up a first call to understand goals, processes and constraints.
  3. You get a written proposal with phases, timeline and costs.
Prefer to write directly? info@softboom.it
Enter your name.
Enter a valid email address, for example name@company.com.
What do you need? (optional)
When would you like to start? (optional)
Goal, timeline, tools you already use. Write a few lines about your project.
We need your consent to process your data before sending.
Let’s talk