I’m Udi Shkolnik, founder of AIB — Artificial Intelligence Bureau.
My background spans information technology, cybersecurity, information security, IT management, auditing, governance, automation, software development, and building technology for real-world business problems.
I’ve worked with technology from several perspectives: as the person responsible for operating it, securing it, auditing it, managing it, and building it.
That combination eventually led me to a problem I couldn’t stop thinking about.
We know how to investigate almost every important part of a company’s digital presence — except what the AI world knows about it.
The Question That Started AIB
As a business owner myself, I started thinking about how people discover companies today.
For years, the answer was relatively straightforward.
- Google.
- Search engines.
- Directories.
- Reviews.
- Social media.
- Word of mouth.
And businesses built entire strategies around understanding those channels.
Then AI changed the interface.
Instead of searching through ten websites, someone can now ask:
- What company should I use?
- Who are the best providers for this?
- Is this company trustworthy?
- Compare these three companies.
- What are the alternatives?
And an AI system can provide the answer directly.
That made me ask a surprisingly difficult question:
- What does AI actually know about my business?
Then came another:
- What does it know about my competitors?
And another:
- Where is it getting that information from?
Then it became a rabbit hole.
One Question Became an Investigation
I realized there wasn’t one AI world.
There was an ecosystem.
ChatGPT could produce one answer.
Gemini could produce another.
Claude could understand the company differently.
Perplexity might discover different sources.
Grok could surface something else.
Different models, retrieval systems, languages, locations and prompts could create completely different representations of the same organization.
And simply asking each system:
- What do you know about Company X?
wasn’t enough. We needed a way to investigate this systematically. That became the idea behind the Artificial Intelligence Bureau.
Why a Bureau?
The name isn’t accidental.
I didn’t want to build another dashboard that sends ten prompts to an AI model and produces a mysterious percentage.
The concept is closer to an intelligence operation.
AIB sends structured investigations across the AI ecosystem.
- It collects observations.
- It discovers sources.
- It identifies competitors.
- It extracts claims.
- It compares answers.
- It looks for contradictions.
- It measures patterns.
- It evaluates confidence.
- It tracks changes.
And then it converts all of that information into something a business can actually understand and act upon. That’s why we call it the Bureau.
Built With an Auditor’s Mindset
My background in information security and auditing strongly influenced how I wanted AIB to work.
AI has an interesting characteristic:
It can sound incredibly confident while being completely wrong.
So an AI answer alone isn’t evidence.
If one model says something about a company, I want to know:
- What exactly did it say?
- Which question produced that answer?
- Was live information used?
- Which sources supported it?
- Do other models agree?
- Does the company’s verified information agree?
- How consistently can we reproduce the observation?
- How fresh is the information?
- How confident should we actually be?
That thinking became one of the core principles behind AIB:
Observe. Verify. Compare. Preserve the evidence.
AIB is designed to distinguish between what an AI system said, what available evidence supports, and what we can reasonably conclude.
Those are not the same thing.
Built for Businesses That Don’t Have Unlimited Resources
There’s another reason I wanted to build AIB.
I’m familiar with the reality of running and growing a business without an enormous corporate budget behind you.
Small and medium-sized businesses can’t hire a team of analysts to continuously investigate ChatGPT, Gemini, Claude, Perplexity, Grok and every new AI system that appears.
They shouldn’t need to.
The investigation can be automated.
The machines can investigate the machines.
AIB can collect the data, normalize it, compare it, score it and turn it into useful intelligence.
That makes capabilities that would otherwise require significant manual research available to a much smaller business.
And I think that matters.
AI Visibility Isn’t Just Marketing
I originally approached the problem through visibility.
But the deeper I went, the bigger the problem became.
Suppose an AI knows your company exists.
Good.
But what if it describes your services incorrectly?
What if it associates your competitor with a problem you solve better?
What if it believes outdated information about your company?
What if different AI systems disagree about where you operate?
What if an influential source contains incorrect information?
What if your company dominates Google but almost disappears when someone asks AI for recommendations?
Those aren’t simply SEO questions.
They’re questions about your organization’s representation inside an emerging information ecosystem.
That’s the larger problem I want AIB to investigate.
What I’m Building
My goal for AIB is straightforward:
Give organizations visibility into how the AI world sees them.
Not just one score.
Not just one model.
Not just one prompt.
I want businesses to be able to understand:
- what AI knows about them,
- what AI doesn’t know,
- what AI gets wrong,
- when AI recommends them,
- when competitors appear instead,
- which sources influence those answers,
- where different AI systems disagree,
- how confident the evidence is,
- and ultimately: what they can realistically do about it.
What AIB Will Never Promise
There is no button that forces independent AI systems to recommend a company.
AIB cannot control ChatGPT.
We cannot control Gemini.
We cannot guarantee that an organization will appear in a particular answer.
And I don’t want to build a company around pretending otherwise.
What we can do is measure.
We can investigate.
We can identify patterns.
We can identify information gaps.
We can find inconsistencies.
We can discover important sources.
We can measure competitors.
We can provide evidence-based recommendations.
And then we can measure again.
That distinction is important to me.
Building AIB in Public
AIB itself is also part of the experiment.
We’re building our own public identity using the same principles we expect the Bureau to investigate for customers.
As AIB grows, we’ll monitor:
- how different AI systems understand AIB,
- which sources discover us,
- which questions cause AIB to appear,
- which competitors appear beside us,
- what information models get wrong,
- and how that representation changes over time.
In other words: AIB investigates AIB. If our methodology works, we should be willing to point it at ourselves.
The Mission
I believe we’re entering a period where businesses will need to understand two versions of their digital identity.
The identity they publish.
Their website, content, products, documentation and messaging.
The identity machines construct.
The information AI systems retrieve, remember, infer, summarize and ultimately present to people.
Sometimes those two identities will match.
Sometimes they won’t.
AIB exists to measure the difference.
Welcome to the Bureau
We’re still early in this shift.
The terminology will change.
The models will change.
The technology will definitely change.
But I don’t think the underlying question is going away:
What does AI know about you?
That’s the question that started this project.
And that’s the question the Artificial Intelligence Bureau exists to investigate.
Udi Shkolnik
Founder, Artificial Intelligence Bureau
Intelligence from the AI world.