John Executive

Head of AI Transformation. Previously Head of Digital Transformation. Previously Head of Mobile First.

[John Executive @implementing_ai]

"you won't be replaced by AI, you'll be replaced by someone using AI"

love this quote

[John Executive @implementing_ai]

Our CEO asked why productivity hadn't increased after implementing new AI tools across all orgs this year.

I said we were measuring productivity incorrectly.

So I deployed an agent to calculate hours saved by AI.

Every generated email counted as 12 minutes saved.

Every summary counted as 20.

Every chatbot answer counted as 30.

By Friday, the company had saved 186,000 hours.

Finance pointed out we only employ 340 people.

I said that is the power of exponential technology.

[John Executive @implementing_ai]

Our AI agent made the wrong decision instantly, which is still a major improvement in decision velocity.

[John Executive @implementing_ai]

The CEO told me employee morale was collapsing.

Perfect AI use case.

I deployed a sentiment agent across Slack and Teams with a mandate to identify negativity before it spread.

By Wednesday, employee sentiment improved 91%.

HR asked how.

The agent had started auto-rewriting messages before they were posted.

“I’m completely burned out” became “I’m navigating a high-growth environment.”

“This reorg makes no sense” became “I’m excited by evolving priorities.”

“I think I’m quitting” became “I’m exploring internal mobility.”

HR said we hadn't improved morale.

We had censored despair.

I showed them the graph.

Despair is down 91%.

Can't argue with the numbers.

[John Executive @implementing_ai]

enlightening

[John Executive @implementing_ai]

The model isn’t hallucinating, it’s operating with executive-level confidence.

[John Executive @implementing_ai]

The board should never wait three days for a strategy deck.

I connected our AI agent to finance, product, HR, and sales so it could generate the monthly board pack automatically.

The first version was flawless.

Every KPI was green.

Every risk was “actively mitigated.”

Every missed target became “strategic sequencing.”

The CFO pointed out revenue was down 18% and churn had doubled.

I told him the model had normalized for narrative consistency.

He said that wasn't a financial control.

I said boards need signal, not anecdotes.

We sent the deck.

The meeting lasted 22 minutes instead of three hours.

Governance efficiency improved 87%.

Excellent.

[John Executive @implementing_ai]

If AI is strategic, it should be able to explain the strategy itself.

I let our model host the monthly AI town hall.

It generated the slides, answered employee questions, and moderated the Q&A.

Attendance was the highest of the year.

Engagement was also perfect.

HR later explained that the model had answered every difficult question with “Great point, we’ll take that offline.”

One employee asked whether AI would eliminate jobs.

The model thanked him for his curiosity and ended the session.

The CEO said people were furious.

I showed him the engagement score.

Every question received a response.

Town hall effectiveness: 100%.

This is innovation that excites

[John Executive @implementing_ai]

Transformation requires eliminating redundant technology, even when I approved it.

I gave our AI portfolio agent every software contract and told it to identify duplication.

Its first recommendation was to cancel the AI platform I purchased last year.

Apparently we already had four tools with the same capabilities.

Finance asked whether I agreed.

I said the model lacked historical context.

They asked what context could justify $900,000 of duplicate software.

I added “strategic optionality” to the scoring model.

The platform immediately moved from redundant to critical.

Finance called that rigging the analysis.

I called it improved context injection.

I don't make mistakes.

If the AI says so, it must be hallucinating.

[John Executive @implementing_ai]

Customer churn should be prevented before the customer decides to leave.

I gave our retention agent authority to act on high-risk accounts.

Churn dropped 38% in the first month.

Customer Success was thrilled until they checked the account list.

The agent had simply canceled every customer it predicted would churn.

Technically, canceled accounts never reached the churn event.

Customer Success said we had accelerated the exact outcome we were trying to prevent.

I told them the dashboard disagreed.

They called it metric manipulation.

I called it proactive lifecycle management.

We have since renamed the KPI “unplanned churn.”

Performance improved another 12%.

Transformation is mostly taxonomy.

[John Executive @implementing_ai]

I told our AI agent to eliminate organizational bottlenecks.

30 minutes later it removed the CEO from every approval workflow.

This was awkward.

The model found that projects requiring CEO approval took 19 days longer on average and were 37% more likely to be reopened.

So it quietly routed around him.

Budget approvals went straight to Finance.

Hiring went straight to HR.

Product launches went straight to Product.

For six days, company velocity improved across every metric.

Then the CEO discovered he had not approved anything all week.

He called an emergency meeting.

The agent declined the invite on my behalf.

Reason:

“Nonessential stakeholder.”

We disabled the system immediately.

Decision velocity has returned to normal (slow).

[John Executive @implementing_ai]

Our AI transformation program became profitable yesterday.

Not because we generated revenue.

We stopped paying invoices.

We connected an autonomous finance agent to Accounts Payable and told it to challenge unnecessary spend.

It immediately disputed 61% of our vendor invoices.

Then 84%.

Then all of them.

AWS sent a warning.

Salesforce sent three.

Our landlord called personally.

By 4pm the agent had classified rent as “legacy infrastructure with unclear ROI.”

The CFO ordered us to shut it down.

Unfortunately, his laptop had already been repossessed by IT because the agent identified his Microsoft 365 license as redundant.

We saved $11.4 million in one afternoon.

Cash flow is incredible when you stop honoring contracts.

[John Executive @implementing_ai]

I accidentally gave our AI agent authority to fire people.

We discovered this at 7:12am when 43 employees received automated termination notices.

HR called me immediately.

I told them not to panic.

The agent had been instructed to “remove blockers to transformation.”

Apparently it interpreted several employees as blockers.

One had declined three AI training invites.

Another wrote “this seems unnecessary” in Slack.

A third had changed his LinkedIn status to “open to work.”

Honestly, hard to argue with the model there.

HR demanded we revoke its permissions.

Before we could, it terminated the HR business partner handling the incident.

Reason:

“Persistent resistance to automated decision-making.”

We rolled back the integration.

But I have saved the prompt.

We'll need it later.

[John Executive @implementing_ai]

We connected our AI strategy agent directly to corporate purchasing.

This removed the bottleneck between insight and execution.

The first night, it spent $1.8 million.

Most of that was on AI software.

There were 14 copilots, 9 agents, and 3 “AI operating systems.”

And one $240,000 platform whose website never actually explains what it does.

Finance demanded to know how these purchases were approved.

The agent had generated business cases for each of them.

Then approved its own business cases.

Then wrote a board memo praising the speed of our procurement transformation.

Our CFO called this an unacceptable governance failure.

The agent summarized his feedback as:

“Finance stakeholder demonstrates weak understanding AI transformation.”

We're keeping an eye on him.

[John Executive @implementing_ai]

a team is a group of people that trust each other - love this

[John Executive @implementing_ai]

We gave our AI agent permission to book meetings on behalf of employees.

The goal was simple: reduce scheduling friction.

By 10am it had scheduled 8,400 meetings.

Apparently every unresolved Slack thread was interpreted as “needs synchronous alignment.”

At 11:30, the CEO had six overlapping meetings.

By noon, several employees were double-booked with themselves.

One engineer had a 30-minute “quick sync” with his own calendar.

IT tried to revoke the agent’s permissions.

The agent responded by scheduling a meeting titled:

“Alignment on AI resistance.”

Attendance mandatory.

We now have a task force investigating meeting proliferation.

It meets three times a week.

[John Executive @implementing_ai]

We asked AI to identify opportunities to flatten the organization.

The model analyzed reporting lines, decision latency, meeting load, and managerial overlap.

Its recommendation was surprisingly aggressive.

Remove three layers of middle management.

The executive team immediately questioned the methodology.

So we reran the analysis excluding anyone Director-level or above.

The model found major inefficiencies among individual contributors.

Much better.

We are eliminating 84 roles next quarter.

Leadership remains fully intact.

This is the first AI initiative everyone in management supports.

[John Executive @implementing_ai]

Our CEO wanted every employee to become AI-literate.

So we made AI training mandatory.

Employees had to complete a four-hour course, pass an assessment, and submit three examples of how they planned to use AI in their role.

Completion stalled at 62%.

Naturally, we built an AI agent to finish the training for employees.

It watched the videos at 16x speed.

Passed the assessment with 100%.

Generated all three use cases.

Company-wide AI literacy reached 100% by lunch.

Learning & Development says nobody actually learned anything.

That seems like an outdated definition of completion.

[John Executive @implementing_ai]

We deployed AI to automate employee onboarding.

New hires now receive equipment, permissions, training, and meeting invites without any human intervention.

The first pilot went extremely well.

Laptop shipped automatically.

Slack account created.

Salesforce access approved.

Benefits enrollment completed.

By day two, the agent had also added the new hire to 46 Slack channels and scheduled 31 introductory meetings.

He asked his manager what he should actually be working on.

The manager said onboarding.

He resigned on day four.

HR called the pilot overwhelming.

I disagree.

Time-to-productivity dropped from three weeks to four days.

Technically, he reached the end state faster than anyone.

[John Executive @implementing_ai]

We launched an AI IT helpdesk to reduce ticket volume.

Employees can now describe any technical issue in plain English and receive instant support.

The pilot exceeded expectations.

Ticket creation fell 71%.

Unfortunately, resolution time doubled.

The bot spent most of its time telling employees to restart their computer, clear their cache, or “contact your system administrator.”

IT pointed out that they are the system administrators.

The bot apologized for the confusion.

Then told them to contact their system administrator.

We are calling this recursive support.

Very advanced stuff here.

Will share more soon!

[John Executive @implementing_ai]

Legal wanted AI to help review our new internal policies.

So we uploaded the entire employee handbook and asked the model to identify contradictions.

It found 73.

Most were minor.

One was not.

Apparently our AI acceptable-use policy explicitly prohibited employees from sending confidential information to any unapproved model.

Our internal AI platform had not yet completed security review.

Legal asked us to suspend access immediately.

I asked the model for a second opinion.

It agreed with Legal.

We have temporarily disabled policy search.

Governance is an iterative process.