Flow by Velory: can you prove your AI compliance?
Can you prove your company's AI compliance?
Velory gives leadership one screen showing how AI is used across the company, and an audit-ready record for when an authority, auditor or board asks. Start from your own AI policy, or from the EU baseline.
Take the quick test to see where your company stands before you run a company-wide assessment.
Eight questions · About 3 minutes
- Early
- Developing
- Structured
- Advanced
How much of your people's work involves AI today?
Your best guess, as a share of the tasks people do in a week.
How many of your people know where to find your AI rules?
Your best guess, as a share of everyone. The rules can be a policy, a page or a short list.
How many AI tools do people use that nobody approved?
Your best guess. Count each tool once, however many people use it.
How many of your people set up AI to do parts of their job, like a saved prompt or an assistant for one task?
Your best guess, as a share of everyone.
If a regulator, auditor or board member asked tomorrow, could you show that your AI policy is followed?
A current record you can export on request.
How clearly can leadership see how work is actually done across the company?
Including core processes, and where AI is already in use.
How often does leadership review how work and AI use are changing?
As a fixed agenda item. A discussion that comes up now and then does not count.
What is the main thing holding you back from better visibility and control?
Choose the single strongest constraint. It shapes your recommendation.
- Early
- Developing
- Structured
- Advanced
Start with a clear overview of tools and practice.
Leadership does not yet share a reliable view of how work is done or where AI is in use. Policy, tool control and evidence become workable once that overview exists.
- Early
- Developing
- Structured
- Advanced
Close the gaps that matter most first.
Some structure exists, but coverage is uneven. Leadership cannot yet decide what to change first, or show its measures with confidence across the company.
- Early
- Developing
- Structured
- Advanced
Extend coverage and make the record automatic.
Visibility, policy and control are largely in place. What remains is incomplete coverage, uneven literacy, or evidence that still takes effort to produce.
- Early
- Developing
- Structured
- Advanced
Keep the overview current as tools change.
Leadership already sees how work is done and has it under control. The job now is keeping the overview current as tools change, with the compliance record in step.
If a regulator asked tomorrow, you would be putting the evidence together from scratch.
Biggest gapStrongest
Next: ask each team which tasks already use AI, so leadership starts from what is true.
Next: adopt or refresh an AI policy, name an owner, and review it every quarter.
Next: list the tools each team uses, including the ones nobody approved.
Next: record what each role has learned, and mark where support is still missing.
Next: decide what you would export tomorrow, and collect it from the overview each quarter.
Next: run a short team questionnaire, so leadership shares one current overview.
Next: put AI use and ways of working on the leadership agenda at least once a quarter.
What to do next
- Run a short team questionnaire on tools and AI use in the functions that matter most.
- Name an owner for the overview and refresh it every quarter.
- Set policy and controls against what the overview shows.
- Finish the tool and practice overview in the functions that matter most.
- Name owners for literacy measures and for following the policy.
- Refresh both every quarter, so one overview replaces scattered local efforts.
- Extend the overview to the remaining functions.
- Line up day-to-day practice with the policy where the two still differ.
- Keep leadership review on a fixed cycle, and let the audit export come from the same data.
- Keep the team assessment light and the review cycle fixed.
- Keep the evidence export running every quarter.
- Use the overview to decide where to move ahead, as well as where to reduce risk.
Ownership is the main hold-up. Settle who owns this before adding more process.
The missing overview is the main hold-up, so start by building one.
Time is the main hold-up. Keep the first assessment light.
Policy and process gaps are the main hold-up. Close them alongside the overview.
Velory builds this overview from a ten-minute team questionnaire. The same answers build a continuous compliance record.
Velory builds this overview from a ten-minute team questionnaire. Once it refreshes on a fixed cycle, the evidence record follows.
Velory builds this overview from a ten-minute team questionnaire. Use it to decide where to invest, what to retire and where people need support.
Velory builds this overview from a ten-minute team questionnaire. An outside review now and then can confirm it stays complete as the tools change.
Book a Demo
A focused session on your policy and tools. You leave with your gaps ranked and a sample audit record.
Your result
Early. Start with a clear overview of tools and practice.
Developing. Close the gaps that matter most first.
Structured. Extend coverage and make the record automatic.
Advanced. Keep the overview current as tools change.
Thanks. We have your demo request.
We will email you to find a time.
How is AI used in your company?
Tools spread faster than policy. How work really gets done sits in people's heads and across systems, so leadership decides on half the facts.
Work is invisible
Real workflows run across teams and tools. Nobody has one overview of what is done by hand, what uses AI and what is automated.
AI arrives untracked
AI assistants show up in daily work before anyone maps where they are used, and for what.
Last year's overview is already out of date
How people use AI changes every few months. A one-off snapshot can't carry decisions or compliance evidence.
Velory gives leadership one current overview
See how work is done across the company, by hand, with AI and automated, on one screen. The same data builds an audit-ready record when you need to show your measures under the EU AI Act.
Built by Velory, which brings IT, Finance and HR together around the tools your people use. Not a performance review.
To start, bring your AI policy, or begin from the EU baseline.
Velory · Operating overviewSample data
- Tasks done by hand
- 6 in 10
- An estimate, from the tasks people did last week
- Tasks done with AI
- 3 in 10
- An estimate, from the tasks people did last week
- Tasks fully automated
- 1 in 10
- An estimate, from the tasks people did last week
How work gets done, by team
- Finance
- IT
- Sales
- Support
Share of work involving AI, round by roundLater
Sample
EU AI compliance record
EU AI Act · Article 4What the record holds, card by card. Fields marked Planned are on the roadmap.
01Document control
- Organisation
- Nordhaven Logistics Group AB
- Registration
- Planned
- Reporting period
- Planned
- Round
- Round 3, sent 8 Sep 2026
- Record owner
- Planned
- Classification
- Planned
02Scope of the duty
- Status
- Planned
- People in scope
- 412 people asked this round
- Aggregation
- Team figures. Admins and managers never see a person's answers. A person's level reaches their manager and admins, and they are told before they answer.
- AI in use
- The tools people picked, under Tool landscape
03Policy in force
- Policy
- AI and Technology Use Policy, version 3
- Rules asked
- 12 rules, each turned into a question about real work
- Approved
- Planned
- Next review
- Planned
- Owner
- Planned
04Coverage by team
Team Asked Finished At L3 or above PolicyPlanned GapsPlanned Finance 48 44 (92%) 45% Planned Planned Sales 110 92 (84%) 30% Planned Planned Support 95 74 (78%) 22% Planned Planned IT 57 54 (95%) 61% Planned Planned Operations 80 60 (75%) 18% Planned Planned Other teams 22 10 (45%) 30% Planned Planned Everyone 412 334 (81%) 33% Planned Planned A team gets its own line at five answers. Teams with fewer fold into Other teams (3 smaller teams here).
05Tool landscape
Tool Status People who picked it Enterprise chat On your list 286 people Transcription app On your list 94 people Code assistant On your list 41 people Consumer chatbot Not on your list 23 people Other tools Under 5 each 7 people 06Assessment method
- Collection
- Know Your AI: role-based questions, about 10 minutes
- Cadence
- Each round when you send it. This round ran 8 to 19 Sep 2026.
- Response
- 334 of 412 asked finished (81%)
- Smallest group
- 5 answers for a team line
- System logs
- Not used. Answers are self-reported.
- Limitation
- Does not certify anyone's competence
07Findings for the period
- Finished
- 334 of 412 asked (81%)
- Strongest
- IT, with 61% at L3 or above
- Lowest
- Operations, with 18% at L3 or above
- Rule to go over first
- Keep client data out of public AI tools. 81% chose it. Lowest: Support 64%.
- Not on your list
- Consumer chatbot, picked by 23 people
- Since last round
- 46 of the 280 who answered both rounds are at a higher level
Today's export is the completion record: a CSV, by name, of who was sent the assessment and who finished it, and when. It holds no answers and no levels. About the completion export
Four views on one screen
What leadership needs to set priorities: how work gets done, where AI is involved, what teams can do, and what to change next.
Sample data
How work gets done
By hand about 6 in 10 · With AI about 3 in 10 · Automated about 1 in 10
Tasks by team, how each was done, and which take more than 4 hours by hand.
Where AI is involved
- Chat assistantOn your list
- Consumer chatbotNot on your list
- Code assistantOn your list
The tools people use, and which teams use them.
What teams can do
- Sales
- Finance
- IT
AI skills by team. A team gets its own line at five answers.
What to change next
- 1Go over the data rule with Support63 people
- 2Approve or block the consumer chatbot23 people
- 3Help with invoice matching6 people
Ranked by how many people each line reaches.
How the overview is built
About ten minutes per person. Short questions about real work. Answers roll up to team level, and admins never see a person's answers.
- 1
People answer
Role-based questions about their tasks, their tools, and where AI helps or gets in the way.
- 2
Velory builds the view
Answers become team views of work done by hand and with AI. Admins and managers never see a person's answers.
- 3
Refresh it every round
Send a round when you want a fresh read, and the overview updates with it.
- About 10 minutes
- Self-reported
- Admins never see a person's answers
- 1
Which AI tools do you use?
Chat assistant, Copilot, other
- 2
What does AI help you do each week?
Drafting, analysis, summaries
- 3
What do you check before sending AI-written work?
Facts, tone, client data
- 4
How confident are you using AI at work?
- 5
Which task would you hand to AI first?
Free text
ResultSample data
Team record
Finance92%Team view
Why this matters for EU compliance
Article 4 of the EU AI Act asks every company that uses AI to take measures that support AI literacy. Enforcement has started, and your evidence only builds from the day you start recording it.
Feb 2025
The AI literacy duty starts to apply to companies that use AI.
27 Jul 2026
Article 4 is reworded: take measures that support AI literacy. It is still binding.
Aug 2026
National authorities can enforce. Each member state sets the penalties.
Today
The earliest your record can begin. It grows with every round.
Start here
Your evidence record, round by round
You can't guarantee every person's level. You can prove what you did.
Dates follow the EU AI Act as amended by the Digital Omnibus on AI. Not legal advice.
From compliance to better decisions
Start with the record. The same overview then grows into AI literacy and workflow change.
- Compliance recordProve the policy is followedLive
- AI literacy by roleSee where support is neededNext
- Workflow opportunitiesRank what to improveLater
Decide what to change next
LaterOn the roadmap: Velory will rank the moves worth making by impact and effort. You will choose the route and track the result.
Sample data
| Opportunity | Impact | Effort | Route |
|---|---|---|---|
| Automate invoice matchingFinance | High | Low | Self-serve |
| Rework contract intakeLegal | High | High | Consultant |
| AI-assisted first draftsMarketing | Medium | Low | Self-serve |
| Retire two overlapping AI licencesIT | Medium | Low | Retire |
Then approve a tool, name an owner or bring in a consultant, and track the result on the same screen.
EU AI Act, Article 4
Article 4 asks employers to build AI literacy. Flow helps you show it.
There is no official certificate. The record of what your teams know, kept up to date, is the proof.
- February 2025
- In force across the EU.
- August 2026
- Enforced by each EU country.
Bring your policy. See if it's followed.
30 minutes. Free. We show you Flow on your own policy.
Thanks. We'll be in touch.
We will write back within one working day to find 30 minutes that suit you.
- Have your AI policy to hand, in whatever shape it is in. PDF, Word, or slides all work.
- The demo runs on that document, not on a sample company.