Controlled evidence

AI projects told through what was actually delivered

We don't use logos, KPIs or testimonials without permission to publish. Every project should show readers the problem, the scope, the output, how it was measured and its limits.

Project library

Numbers that speak for every result

Each project below records the original problem, the implementation scope and the measured results. Click a project to see the details.

Illustrative example

AI training by department

Problem: the team had tried tools but had no shared way of working. Sample output: a list of use cases, practice workflows and a 30-day adoption plan.

In internal use

Operational report consolidation

Describes an internal process that has been built and is in use. No figures are published until the baseline and measurement conditions are confirmed.

Illustrative example

Enterprise knowledge assistant

Sample scope: authorised document sources, answers with citations, handling of incorrect responses and a knowledge update process.

Every project must answer seven questions

What problem does the business face and who is affected?
How did the process work before the pilot?
Within what scope was Visun AI allowed to intervene?
Which data was used and who granted access?
What output did the system or programme produce?
Which source measured the results, and over what period?
What are the limits, lessons and next steps?

Rules for using numbers

A number only appears when there is a baseline, a measurement method, a measurement period, a defined scope and permission to publish. If not, the website uses phrases such as “built”, “in use”, “being tested” or “not enough data yet”.

A good project starts by deciding how to measure before doing the work

Visun AI can help your business design the pilot scope and the acceptance criteria.

Book a consultation →