LBZ Advisory LLC

08/25/2026

Chipotle's restaurant turnover ran close to 200% a year. Its COO called hiring "one of the most painful processes" managers had to deal with.

Then an AI assistant took over interview scheduling and applicant questions. Time from application to first day dropped from 12 days to 4.

What's worth paying attention to: the time that came back went to restaurant managers, the people whose actual job is running a shift. Not a back-office function with room to absorb admin. The people with the least slack.

Most AI ROI cases count total hours saved. The better question is: whose constraint does this actually remove? A tool that frees up four hours for your most stretched person usually beats one that saves forty hours nobody was waiting on.

The Chipotle story, fully sourced, with the questions to ask any hiring AI vendor: getstartedwithai.liatbenzur.com

08/21/2026

Everyone from OpenAI to the University of Helsinki gives away an AI course. Far more than any team has time to sort.

I hand-checked 11 worth taking for leaders and non-technical staff. Grouped by how you learn. Paywalls named. 9 of 11 never ask for money.

Send this to your team instead of another random Slack link:
https://getstartedwithai.liatbenzur.com/training/

AI is more than a chat window. See the software and agents real companies use in finance, HR, legal, marketing, product, support, sales and operations — what it changed, and what to ask before you buy.

08/19/2026

Klarna's AI handled two-thirds of chats at peak, work they said equaled 853 employees. Support costs still rose. They rehired.

The lesson is blunt: excellent at the repetitive two-thirds, bad at the rest, and headcount got cut before anyone drew that line.

I put failure stories next to the wins on purpose. Functional leaders need both.
https://getstartedwithai.liatbenzur.com/function/support/

Deflect repetitive enquiries automatically, equip reps with instant answers, and cut resolution times.

08/17/2026

monday.com's inbound was outrunning a ~1,000-person GTM team. A demo request could sit a full day.

Amanda, an AI voice agent that qualifies every inbound lead by phone, cut demo-request response time from 24 hours to under two minutes.

Speed-to-lead shows up in pipeline before the harder, judgment-heavy parts of the funnel do.
https://getstartedwithai.liatbenzur.com/function/sales/

Research prospects instantly, automate post-call admin, and flag deals that are slipping.

08/14/2026

HSB processed 1.5M invoices a year. Staff entered every invoice twice because two systems could not talk to each other.

AI coding earned autonomy invoice by invoice (not day-one autopilot). Year one: 60,000+ hours saved, 72% of invoices with no human touch at 96% coding accuracy.

Finance leaders: this is the sequencing story.
https://getstartedwithai.liatbenzur.com/function/finance/

Close the books faster, catch invoice errors before payment, and run real-time cash flow scenarios.

08/13/2026

Chipotle cut time-from-application-to-first-day from 12 days to 4 with an AI hiring assistant, mostly by taking scheduling off restaurant managers, not recruiters.

That is the kind of specific, boring win functional leaders actually need.

More function-level AI stories (sourced), failures included:
https://getstartedwithai.liatbenzur.com/function/hr/

Source qualified candidates faster, handle routine HR tickets automatically, and free up hours every week.

08/12/2026

Every few weeks a senior leader pulls me aside after an AI talk with the same question:

I'm not a tech expert. How do I help my team use AI beyond chatbots?

HR. Finance. Legal. Sales. Twenty+ years in the seat. They want to transform the function. They don't know where to start.

I built the page I kept wishing I could send them.

How Do I Get Started with AI?
One page per department. Real company deployments (and failures), sources you can check. Vendors worth calling, and the questions to ask before you buy. 11 free courses, hand-checked. Zero sponsored listings. Refreshes daily.

Start with your department:
https://getstartedwithai.liatbenzur.com

If you want this framed for a leadership offsite or board briefing: https://liatbenzur.com/speaking

AI is more than a chat window. See the software and agents real companies use in finance, HR, legal, marketing, product, support, sales and operations — what it changed, and what to ask before you buy.

08/12/2026

Everyone is posting the same future org chart.

Thin layer of humans on top. Wide layer of agents below. Your new job? Managing context.

Unpopular opinion: That is half the job. And the less important half.

Context is being sold as neutral infrastructure. A searchable brain of playbooks, customer history, and procedures that makes agents useful.

It's not neutral.

Context is history with an interface.

Your context layer doesn't just hold facts. It holds the residue of who was considered credible, whose exceptions got approved, whose warnings were quietly dropped, and which metrics actually got rewarded.

If you make that executable without interrogating it, you don't get a smarter company. You get a very efficient machine for reproducing the past.

That viral frame misses that power didn't disappear. It migrated. Into permissions, default settings, escalation rules, evaluation criteria, and the order in which an agent reads information.

A prompt is a political document. So is a confidence threshold.

In my new essay I break down what everyone is skipping:

Why "managing context" is not enough, why the real org chart is a network of decision rights, and the 5 leadership currencies that actually matter when ex*****on gets cheap: Judgment, Power Fluency, Narrative Control, Trust Capital, and the Courage to Dissent.

If you're building the agent-native org, the question isn't what your agents can do. It's what your organization has learned to ignore.

Full essay is live on my site - link in comments.

Why managing AI context is an incomplete view of the agent-native organization: and why judgment, power fluency, trust, narrative control, and dissent will define leadership.

08/07/2026

On August 2, 2026, the EU began enforcing its rules on general-purpose AI models. Most headlines called it European pre-market approval. The reality is more interesting — and more effective.
Software lived for years by one creed: ship first, apologize later. That creed just died, and not only in Brussels. Washington is reaching for the same lever from the other side, trying to assert pre-market control over frontier labs. Two governments now get a say in the day your model is allowed to exist.
The law does not create a formal checkpoint before a lab trains or ships a general-purpose AI model. No one in Brussels stamps GPT-5 or Claude 4 before the servers turn on. What the EU built is different. Under Articles 91 and 92, the Commission can demand your documents, inspect how the model was built, and order changes after the fact.
If you ignore that demand, they can fine you up to three percent of global annual revenue. That is enough to make any serious lab pay attention. So no, Europe did not put a guard at the front door. It put a trapdoor in the room.
Read the full essay here…

The EU AI Act is now in full force, and it's reshaping the landscape of AI compliance. From hidden compliance triggers to compute thresholds, there are many nuances to consider. This is a game-changer for AI labs and developers worldwide.

08/04/2026

AI governance for boards: who owns the stop button?

A practical guide to AI governance for boards: decision rights, named kill-switch ownership, director questions, and a 90-day start. For directors and CEOs who need usable oversight....

A practical guide to AI governance for boards: decision rights, named kill-switch ownership, director questions, and a 90-day start. For directors and CEOs who need usable oversight.