Things I've Learned Experimenting With AI on Christa Burger

My AI Maturity Ladder: From Chat to Choreography

Wed, 24 Jun 2026 00:00:00 +0000

Here is how I currently think about AI maturity:

1. Chat

Ask questions, get answers.

2. Templates

Reuse prompts for common work.

3. Workflows

Connect steps into repeatable processes.

4. Roles

Different AI functions do different jobs.

5. Governance

Add review, evidence, escalation, and accountability.

6. Choreography

Intelligence moves through the organization in designed patterns.

Most people are somewhere between 1 and 3.

That is not bad.

But it explains why results still feel inconsistent.

Prompting Is Not the Endgame of AI

Mon, 22 Jun 2026 00:00:00 +0000

Prompting is useful.

For your phone AI app.

Prompting is not the endgame.

After a few years of experimenting with AI, I think the real maturity curve is choreography.

How does work enter the system?
Who gathers context?
Who summarizes?
Who checks risk?
Who drafts?
Who reviews?
Who approves?
Where does the evidence go?
What happens next time?

That is a different conversation.

Prompting is asking better questions.

Choreography is designing how intelligence moves.

AI Loves a Pancake: Why Good AI Should Not Flatten Complex Work

Fri, 19 Jun 2026 00:00:00 +0000

AI has a tendency to turn complex situations into pancakes.

Flat. Neat. Easy to consume. Missing several load-bearing ingredients.

Sometimes that is fine.

Sometimes you need the pancake.

But serious work often has layers:

Operational
Financial
Human
Strategic
Ethical
Historical
Political
Timing-related

Good AI should help hold those layers without making the next action impossible.

That is the balance.

Clarity without flattening. Structure without pretending the situation is smaller than it is.

AI Productivity Is Too Vague. Show the Burden Removed.

Wed, 17 Jun 2026 00:00:00 +0000

“AI improves productivity” is true and also not very useful.

Productivity where?

Which burden got lighter?

What changed?

Did we reduce rework?
Did we prevent something from being forgotten?
Did we make decisions cleaner?
Did we improve evidence?
Did we reduce manual reconstruction?
Did we make the next step obvious?

That is what I care about.

Not vague productivity.

Specific relief with receipts.

This is how AI use cases should be explained.

Turn AI Output Into Infrastructure

Mon, 15 Jun 2026 00:00:00 +0000

The best AI outputs are not just deliverables.

They become infrastructure.

A meeting summary becomes a decision log.

A transcript becomes a content workflow.

A one-off checklist becomes a reusable control.

A draft becomes a communication template.

A messy brainstorm becomes a playbook.

That is where AI starts to compound.

Not when it helps create the thing.

When it helps create the thing that helps create future things.

That is leverage.

Chat Is the Workshop, Not the Warehouse

Fri, 12 Jun 2026 00:00:00 +0000

A lot of AI value dies in chat windows.

Good idea. Useful framework. Strong draft. Solid analysis.

Then it disappears into the scroll swamp, where all productive conversations go to become vaguely remembered.

This is a problem.

If AI creates something valuable, it should often become an artifact outside the chat:

A document
A template
A report
A workflow
A decision log
A playbook

Chat is a great workshop. It is a terrible warehouse.

Good AI Should Leave You With Something Usable

Wed, 10 Jun 2026 00:00:00 +0000

A good AI interaction should often leave behind something usable.

Not just an answer.

An artifact.

A checklist. A template. A decision log. A draft. A table. A playbook. A workflow. A reusable prompt. A meeting summary with owners.

The answer may help once.

The artifact helps again.

That is the difference between convenience and leverage.

I have become much more interested in asking:

“What should this become?”

A document? A template? A control? A process? A decision aid?

Unnamed Work Becomes Weather

Mon, 08 Jun 2026 00:00:00 +0000

Unnamed work becomes weather.

It just happens.

People feel it. People complain about it. People plan around it. Nobody owns it.

This is true in teams, households, operations, governance, and AI workflows.

The first act of control is often naming.

Name the recurring burden
Name the decision point
Name the review
Name the handoff
Name the risk
Name the evidence trail

Once something has a name, it can have an owner.

AI Ecosystems Beat AI Assistants

Fri, 05 Jun 2026 00:00:00 +0000

I think the shift in AI is from assistant to ecosystem.

An assistant helps with a task.

An ecosystem supports a pattern of work.

That is a very different thing.

An ecosystem has:

Roles
Handoffs
Review points
Memory
Escalation paths
Operating rhythms
Evidence capture

This is where AI starts becoming infrastructure.

But only if we design it.

Otherwise, we just end up with several disconnected tools producing several disconnected outputs while everyone calls it transformation and quietly keeps using spreadsheets.

Give AI a Job Description Before Giving It Access

Wed, 03 Jun 2026 00:00:00 +0000

Before giving AI more access, give it a job description.

What is it responsible for?
What should it ignore?
What does good output look like?
What tools can it use?
When should it escalate?
What should it never decide?

This sounds obvious.

It is apparently not.

People will give AI access to meaningful workflows with less role clarity than they would give an intern named Brayden who starts Monday and says “for sure” too much.

Not Every AI Should Be the Same AI

Mon, 01 Jun 2026 00:00:00 +0000

I used to put too much into one AI conversation.

Strategy. Drafting. Summarizing. Planning. Debugging. Content ideas. Household systems. Possibly plant identification.

A perfectly normal range of human concerns.

Eventually, I realized the issue was not the model.

It was the lack of role clarity.

Human teams have roles for a reason:

Strategist
Analyst
Operator
Reviewer
Auditor
Assistant

AI needs the same distinction.

Some AI functions should explore. Some should execute. Some should check. Some should preserve context. Some should challenge assumptions. Some should produce clean artifacts.

Margin Is the Metric for Useful AI

Fri, 29 May 2026 00:00:00 +0000

I want to see more AI conversations include the word margin.

Not just time saved. Not just cost reduced. Not just output increased.

Margin.

Did this reduce decision fatigue?
Did it make the next right action clearer?
Did it prevent rework?
Did it improve trust?
Did it give someone enough capacity back to think?

That matters.

Speed without margin is just burnout with better tooling.

AI can absolutely help us move faster.

Automation Should Not Eat the Human

Wed, 27 May 2026 00:00:00 +0000

Automation has a suspicious habit of becoming another thing to manage.

You automate the workflow.

Then you monitor the automation. Then troubleshoot the automation. Then explain the automation. Then build a dashboard for the automation. Then attend a meeting about the automation.

At this point, the automation has not saved time.

It has developed a small administrative ecosystem and started charging rent.

AI needs a better standard.

It should reduce load, not relocate it.

AI Should Protect Aliveness, Not Just Increase Output

Mon, 25 May 2026 00:00:00 +0000

A lot of AI productivity advice seems to assume the goal is to become a more efficient toaster.

More output. More speed. More content. More emails. More dashboards. More lightly formatted urgency.

I am not interested.

The point of AI should not be to cram more sludge into the same calendar.

The point should be to create margin.

For better judgment. Better work. Better thinking. Better relationships. Better rest. Better stewardship.

Invisible Work Is Still Work, and AI Can Help Reveal It

Fri, 22 May 2026 00:00:00 +0000

A lot of work does not look like work.

Remembering is work. Noticing is work. Checking is work. Sequencing is work. Following up is work. Reconciling is work.

Holding the whole thing in your head so nothing explodes is definitely work.

It is just badly marketed.

AI gives us a way to make that labor visible.

Once visible, it can be structured.

Once structured, it can be assigned.

Once assigned, it can be improved.

AI Is Best at the Work Before the Work

Wed, 20 May 2026 00:00:00 +0000

Most of the burden is not the task.

It is the work before the task.

Finding the context. Reading the history. Checking the numbers. Remembering what happened last time. Figuring out who needs to care. Deciding what decision is actually needed.

By the time you start the “real” work, half your energy has already been mugged in the parking lot.

AI is very good at this layer.

It can:

Gather context
Summarize history
Compare versions
Surface changes
Draft the starting point
Identify the decision

This is not glamorous.

Accountability Should Not Require Archaeology

Mon, 18 May 2026 00:00:00 +0000

If accountability requires archaeology, the system is under-designed.

People should not need to dig through inboxes, Slack threads, meeting notes, and the memory of whoever was least distracted at the time.

That is not a process.

That is a scavenger hunt with legal exposure.

AI should make this better.

Every meaningful AI-assisted workflow should be able to answer:

What did it do?
What inputs did it use?
What changed?
What did it recommend?
What did a human approve?
Where is the evidence?

This is not just a security or compliance concern.

"I Think We Talked About This" Is Not a Control

Fri, 15 May 2026 00:00:00 +0000

The worst time to document a decision is after something goes wrong.

Everyone is calm. Everyone remembers clearly. No one is searching email for “final_final_ACTUAL_USE_THIS_ONE_v7.”

Obviously.

AI can help us stop doing this.

It can capture decisions as they happen:

What was known?
What was decided?
Who decided it?
What changed later?
What evidence existed at the time?

That is incredibly valuable.

But only if we design for it.

Otherwise AI just helps us create more outputs with less traceability.

Reminders Are Not Systems: Building AI Rhythms That Actually Work

Wed, 13 May 2026 00:00:00 +0000

I have a deep affection for reminders. And planners. And all the other systems you can place around yourself to luxuriate in the feeling that you might be on top of things, if you just followed all the tasks and schedules you set for yourself.

I also have a profoundly consistent history of dismissing them and continuing with my life like the free bird that I am.

Two wolves live inside us all, my friends.

Set It and Forget It Is a Trap in AI Automation

Mon, 11 May 2026 00:00:00 +0000

“Set it and forget it” is not how most useful automation works.

It is how rot gets a login.

Systems without maintenance atrophy with time.

The better model is not forgetting.

It is rhythm:

What runs daily?
What runs weekly?
What runs when triggered?
What needs human review?
What gets summarized?
What gets escalated?
What gets archived?
Do you have something auditing it all?

AI systems need cadence because real life has cadence.

Heroics Are Usually a Design Problem

Fri, 08 May 2026 00:00:00 +0000

I used to be more impressed by heroic effort.

The late-night push. The last-minute save. The frantic deck rescue. The spreadsheet that emerges from the ashes wearing a tiny cape.

Now I mostly see heroics as a signal: a broken system relying on the goodwill of one person — or their insecurities — to keep functioning.

Sometimes heroics are necessary. Workloads can fluctuate wildly.

But that is also a lazy acceptance.

Most Knowledge Work Is Recurring Work Pretending to Be One-Off

Wed, 06 May 2026 00:00:00 +0000

Most work is recurring work pretending to be one-off.

The report. The summary. The review. The prep. The follow-up. The meeting notes. The risk explanation. The “can you just pull this together?”

We treat these as isolated events because nobody has time to stop and say:

“Wait. Why do we keep doing this from scratch?”

AI is very good at exposing that.

The first time, it helps.

The second time, it suggests a template.

How Small AI Requests Become Repeatable Workflows

Mon, 04 May 2026 00:00:00 +0000

In my journey of systematizing my life and reducing my decision load, I have learned to be suspicious of “small” AI requests, because many of them are secretly recurring workflows.

“Can you make this list?”

Sure.

But should this become a template? Should it recur weekly? Should it have owners? Should it pull from prior context? Should it produce a decision? Should it feed another workflow?

This is how a simple request can turn into an entire operational philosophy.

The Task Is Never the Task: AI and Hidden Workflows

Fri, 01 May 2026 00:00:00 +0000

One of the most annoying things AI has taught me is that the task is almost never the task.

In fact, much of my journey with AI has been analogous to If You Give a Mouse a Cookie.

I will ask for something simple.

A grocery list. A summary. A draft. A checklist. A content idea.

Very reasonable.

Then, unfortunately, the system reveals that this “simple task” is actually attached to a much larger creature.

Human-in-the-Loop AI Is Not Enough

Wed, 29 Apr 2026 00:00:00 +0000

“Human in the loop” has become one of those phrases we say to make everyone feel better.

Like “cross-functional alignment.” Or “quick sync.” Or “this should be straightforward.”

The issue is not whether a human is somewhere in the loop.

The issue is whether the human is in the right place.

Before the AI acts?
After it drafts?
When risk is detected?
When money is involved?
When customer trust is at stake?
When a decision needs authority?

That matters.

AI Should Prepare Decisions, Not Replace Human Judgment

Mon, 27 Apr 2026 00:00:00 +0000

Most leaders are not lacking judgment.

They are lacking clean inputs. Good context. Sometimes, a general curiosity to understand.

AI can fix two of those things.

By the time a decision reaches a senior leader, it is often wrapped in six emails, three Slack threads, a deck, two side conversations, a deadline that is now their problem, and the risk of it all dumped into their lap.

From there, what happens next is highly dependent on the leader and their risk tolerance.

Why Metaphor Matters in AI System Design

Fri, 24 Apr 2026 00:00:00 +0000

We often think metaphor is something that explains a system after the fact.

Now I think metaphor is often how you know what you are building.

This has been especially true with AI because AI systems get abstract very quickly.

You start with, “I need help managing recurring work.”

Three hours later, you are still wrestling with what that means, trying to identify recurring work in your own life for realsies this time, and maybe even falling into a therapy session with chat about why you are like this.

When to Use a Broad AI Prompt and When to Get Specific

Wed, 22 Apr 2026 00:00:00 +0000

One of the fastest ways to creativity with AI is to say:

“Just help with this.”

I have done versions of this professionally, personally, creatively, and domestically, with varying degrees of grace and unnecessary tabs.

The reason this works so well is because it harnesses a principle about AI — the need for specificity — and flips it to your advantage.

When AI has no clear role, it tries to be helpful in every direction at once.

The Problem With One Giant AI Assistant

Mon, 20 Apr 2026 00:00:00 +0000

After 5+ years of experimenting with AI, one of my clearest lessons is this:

The answer is probably not one giant assistant.

I know. It’s a bit tragic, but you’ve tried it, I’ve tried it, and it doesn’t really work.

Eventually this will improve. But as of this writing, whenever you try to make one AI do everything, it eventually becomes a very articulate junk drawer.

It can summarize. It can draft. It can analyze. It can plan. It can hallucinate a project management structure with great confidence and light jazz energy.