Field Notes

Field Notes and Perspectives

Perspectives, analysis, and commentary on the shifting landscape of customer experience, automation, and the evolving role of humans in the AI era.

Latest From The Field
November 11, 2026LinkedIn Article

The AI Operations Gap

Infographic titled 'The AI Operations Gap: Deploying AI is easy. Operating it at scale is the challenge.' Three tiers flow downward: AI Models & Agents, then AI Operations (governance, human oversight, escalation, quality assurance, compliance), then Business Outcomes (customer experience, productivity, growth, risk reduction).

For years, digital transformation was about implementing technology.

Today, the challenge is different.

As enterprises deploy AI across customer service, operations, compliance, and internal workflows, a new layer of work is emerging:

Oversight. Intervention. Escalation. Governance. Continuous improvement.

AI doesn't operate itself. And increasingly, that's where success or failure is determined.

The future isn't just AI.

It's AI operations.

What does "AI Operations" mean inside your organization today?

Read & discuss on LinkedIn
November 3, 2026LinkedIn Article

The Hidden Cost of AI Containment

Illustrated infographic titled 'The Hidden Cost of AI Containment,' in three panels: why organizations invest in AI, a current-state service workflow where AI assists isolated tasks between human handoffs, and expected versus actual outcomes, concluding that AI adoption increases while business outcomes lag.

Organizations invest in AI to improve productivity, reduce costs, and deliver better customer experiences. Yet many initiatives fall short of expectations.

The challenge is often not the AI itself. It's where AI is being applied.

In many organizations, AI is confined to individual tasks such as summarizing conversations, retrieving information, or generating recommendations. While these capabilities create efficiencies, the broader workflow often remains unchanged. Manual handoffs persist; systems remain disconnected, and decision-making continues to rely on fragmented processes.

The result is a gap between AI adoption and measurable business outcomes.

Real value emerges when AI is integrated into how work gets done, while maintaining appropriate human oversight and accountability. Technology alone does not transform operations. People, process, and technology must work together.

Organizations that focus on outcomes rather than AI activity are more likely to achieve faster resolution times, lower costs to serve, better customer experiences, and more scalable operations.

The hidden cost of AI containment is not the investment made in AI. It's the business value left unrealized when AI remains isolated from the processes it was meant to improve.

Is your AI improving your operations, or just sitting next to them?

At Malasakit.ai, we help enterprise teams move AI out of isolated tasks and into end-to-end workflows, with human oversight built in.

DM us to book a 30-minute discovery call. We'll map where AI is contained in your operation and where the fastest wins are.

Read & discuss on LinkedIn
October 27, 2026LinkedIn Article

Why We Built Malasakit.AI

For years, the conversation around AI has focused on what can be automated.

We became interested in a different question:

What happens when automation reaches its limits?

Not because AI is failing. Quite the opposite.

As AI handles more interactions, transactions, and decisions, the remaining exceptions become more important, not less.

They're the moments where judgment, accountability, business context, and customer outcomes intersect.

What we kept seeing was a growing gap between successful automation and successful resolution.

That's why we built Malasakit.AI.

We believe the future isn't AI versus humans.

It's creating the right operating model for AI and humans to work together, so organizations can scale automation without losing ownership of outcomes.

If that challenge sounds familiar, we'd love to connect.

Read & discuss on LinkedIn
October 20, 2026LinkedIn Article

Most Companies Don't Need Another AI Agent

Our last post was about measuring the wrong things.

Too many organizations celebrate AI activity instead of business results. More chatbots. More automation. More dashboards.

But here's the bigger problem:

Many AI initiatives focus on making AI smarter. Very few focus on making humans and AI work better together. Because regardless of how advanced the technology becomes, there will always be moments when human judgment, expertise, or empathy matter.

The question isn't: "How much can we automate?"

The better question is: "How do we create the best possible outcome when AI and people collaborate?"

That's where many organizations struggle.

AI handles part of the interaction.

Your people handle another part.

The customer gets passed through multiple systems. Context gets lost. Knowledge gets fragmented. And productivity gains never fully materialize.

At Malasakit.AI we take a different approach. We don't replace your existing ticketing systems, CRM platforms, or AI investments. We complement them.

Whether your organization uses ServiceNow, Zendesk, Salesforce, Dynamics, Intercom Fin, Microsoft's AI solutions, OpenAI-powered agents, or internally developed AI systems, Malasakit.ai is designed to work alongside your existing tech stack rather than replace them.

The future is organizations that know when to automate, when to involve people, and how to make both work together effectively. That's the metric worth measuring.

Want to see how Malasakit.AI works with the systems you already have? Let's talk.

Read & discuss on LinkedIn
October 16, 2026LinkedIn Article

AI Can Optimize Every Metric. It Still Can't Fake Customer Value.

Comic of a man pointing proudly to a dashboard showing likes, impressions, clicks, followers, and reach all climbing, while actual customers are down 12%, captioned 'Great news, team! Our engagement is through the roof. Too bad no one's actually buying.'

AI can optimize every metric on your dashboard. It still can't fake customer value.

The companies that win won't be the ones with the most content, clicks, or automation.

They'll be the ones using AI to solve real problems for real people.

What's one AI metric you're tired of hearing about?

Read & discuss on LinkedIn
October 13, 2026LinkedIn Article

You're Measuring the Wrong Thing

Many organizations track AI success using operational metrics:

  • Cost reduction
  • Ticket deflection
  • Response times
  • Productivity gains

These metrics matter. But they don't answer the most important question: Did the customer achieve the outcome they came for?

An AI agent may resolve a ticket in 30 seconds. A chatbot may answer a question instantly. A workflow may complete without human intervention. The dashboard turns green. The metrics look fantastic.

Yet customers still leave. Why?

Because efficiency and outcomes are not the same thing. A fast interaction that fails to solve the underlying problem is still a failure. A closed ticket is not necessarily a successful customer experience. A completed workflow is not always a completed outcome.

As organizations deploy more autonomous systems, leaders need to evolve what they measure.

Instead of asking: "How many tasks did AI complete?"

Start asking: "How many customer outcomes were successfully achieved?"

The difference is subtle. But it's transformational. Teams focused on tasks optimize for speed. Teams focused on outcomes optimize for success. And customers rarely remember how fast a process was. They remember whether their problem was solved.

In the AI era, the organizations that win won't be those with the most automation. They'll be the ones that can prove automation delivered the outcome that mattered.

Efficiency measures activity. Outcomes measure value. Your dashboard may be green. Are your customers actually winning?

Read & discuss on LinkedIn
October 2, 2026LinkedIn Article

Quick Question: Who Owns the Issue Between AI, Support, Ops, and Product?

Two-panel comic: panel one, 'Automation looks like a success,' panel two, 'but someone still has to own the outcome,' as an AI agent, billing system, and support queue all point fingers at a customer-success rep buried in tickets — closing with 'Automation scales work. Ownership scales trust.'

Quick question: When a customer issue falls between AI, Support, Operations, and Product... Who actually owns it?

That's the conversation most organizations haven't had yet. Automation scales work. Ownership scales trust.

We're spending a lot of time helping enterprises solve this exact problem. If you're navigating AI adoption and accountability, let's talk.

Read & discuss on LinkedIn
September 30, 2026LinkedIn Article

The Outcome Operations Model

Most organizations are redesigning work around automation. Few are redesigning ownership.

The conversation around AI usually focuses on three questions:

  • What can AI automate?
  • How much cost can it remove?
  • How many interactions can it handle?

Those are important questions. But they miss the operational challenge that appears after deployment. When an exception occurs...

Who notices it? Who intervenes? Who coordinates resolution? Who owns the outcome?

Historically, organizations embedded ownership directly into human workflows. Managers reviewed edge cases, specialists resolved exceptions, and teams coordinated seamlessly across functions. Accountability existed because humans naturally carried it.

AI changes that equation. The workflow is becoming faster. The organization becomes leaner. But the ownership requirement hasn't disappeared. It has become more important.

This is why the next generation of customer operations won't be built around automation layers. It will be built around outcome ownership.

An effective AI operating model requires a function responsible for:

  • Detecting risk before failure occurs
  • Coordinating across systems and stakeholders
  • Resolving exceptions when automation reaches its limits
  • Continuously improving both workflows and models

Not because the AI is broken. Because outcomes still need owners.

The companies that scale AI successfully won't be the ones that automate the most processes. They'll be the ones that maintain accountability across all of them.

Automation scales work. Ownership scales trust.

Read & discuss on LinkedIn
September 29, 2026LinkedIn Article

The Automation Myth, Illustrated

Four-panel comic contrasting a dashboard showing 'exceptional' resolution metrics with a customer still waiting on an unresolved order, captioned 'Automation handles the routine. Ownership handles the exception.'

The automation conversation often focuses on what AI can do. The more important question is what happens when something goes wrong.

The exception doesn't care about your automation rate. The customer only cares whether the problem was solved.

That's the automation myth: reducing workload is not the same as eliminating responsibility.

Read & discuss on LinkedIn
September 25, 2026LinkedIn Article

The Automation Myth

Every week, another company announces AI-driven efficiency. Hiring freezes. Flatter organizations. Fewer frontline roles. In many cases, the economics is real.

Organizations are discovering that AI can handle an enormous amount of routine work. Some leaders are even requiring teams to prove AI cannot do a task before approving additional headcount. In a widely discussed example, Tobi Lütke, CEO of Shopify, stated that teams must show why they "cannot get what they want done using AI" before requesting more people or resources.

But there is an important assumption hiding underneath many of these conversations...

That automating work also eliminates responsibility. It doesn't.

When an AI successfully handles 90% of customer interactions, the focus naturally shifts to the cost savings. But customers don't experience percentages.

They experience exceptions. The flight that wasn't rebooked correctly. The payment that disappeared between systems. The order that shows "Delivered" but never arrived. The policy that made sense in the workflow but failed in the real world.

Ironically, the more successful AI becomes, the more visible these moments become. Because scale magnifies exceptions.

A million successful interactions are invisible... the unresolved one is memorable.

That's why the future isn't AI versus humans. And it isn't humans versus efficiency.

The organizations that win will be the ones that understand a simple truth: Automation reduces workload. It does not eliminate ownership.

In fact, ownership becomes more important than automation scales. Many organizations are redesigning their workforce around AI. Some are reducing roles, while others are repositioning human responsibilities toward higher-value judgment and exception management.

The question isn't: "How many people can AI replace?"

The more important question is: "Who owns the outcome when the workflow reaches its limits?"

Because customers don't care whether a failure was human or AI. They care whether their flight gets rebooked, their payment is found, or the package gets delivered. They care that someone owns the resolution.

Reference: Palmer, Annie. "Shopify CEO says staffers need to prove jobs can't be done by AI before asking for more headcount." CNBC, April 7, 2025

Read & discuss on LinkedIn
September 22, 2026LinkedIn Article

CX Has Metrics for Everything. Did It Actually Work?

Comic of a boardroom reviewing a slide of Q3 AI success metrics — workflow completion, tasks automated, cases closed — with one executive asking, 'That's excellent. Do we have a metric for whether the customer's problem is actually gone?'

CX has metrics for everything. Workflow completion. Automation. Handle time. Sometimes the hardest thing to measure is the only thing customers care about: Did it work?

At Malasakit.AI, we're increasingly asking a different question: "Was the customer's problem actually solved?"

How does your organization measure true resolution?

Read & discuss on LinkedIn
September 15, 2026LinkedIn Article

The Resolution Gap

Organizations often measure AI success by completion. Customers measure success by resolution. Those sound similar but they are not.

An AI agent can:

  • Provide an answer
  • Execute a workflow
  • Retrieve information
  • Complete a transaction

And still leaves a customer frustrated.

Why?

Because customers aren't trying to complete workflows. They're trying to solve problems.

A billing credit applied incorrectly.

The reservation changed unexpectedly.

A delivery that never arrived.

A refund that was technically processed but never appeared in the account.

From a systems perspective, the workflow may be complete. From a customer perspective, the problem still exists. This creates what many organizations are now encountering: The Resolution Gap.

The space between operational completion and customer resolution is important. The challenge isn't that AI failed. In many cases, it performed exactly as designed. The challenge is that customer outcomes often span:

  • Multiple systems
  • Multiple decisions
  • Multiple stakeholders
  • Multiple moments in time

No single workflow always contains the full answer.

As organizations scale agentic AI, the question becomes less:

"Can the AI complete the task?"

and more:

"How do we know the customer's problem is actually resolved?"

The next generation of customer operations won't be defined by automation rates alone. It will be defined by an organization's ability to close the resolution gap.

Because customers don't measure effort.

They measure outcomes.

Read & discuss on LinkedIn
September 7, 2026LinkedIn Article

Why the Ownership Layer Becomes a Competitive Advantage

In the last post, we introduced the concept of the Ownership Layer: the function responsible for maintaining accountability when automation reaches its limits.

At first glance, that sounds like risk management.

It is. But that is not where its greatest value comes from.

Imagine two airlines facing the same weather disruption. Both operate similar aircraft, fly similar routes, and use similar technology. Yet one recovers faster, communicates more clearly, and restores normal operations sooner. The difference is not aircraft. The difference is what happens when the workflow breaks.

AI is creating a similar divide.

Many organizations assume competitive advantages will come from better models, higher automation rates, or lower support costs. Those things matter. But as AI scales, a more important question emerges:

Who owns the outcome when the exception occurs?

Organizations without clear ownership tend to slow down. Every new AI initiative introduces uncertainty around accountability, decision-making, and escalation. As a result, automation expands cautiously, not because the technology is not ready, but because the organization is not.

Organizations with a strong ownership model operate differently. Because accountability remains intact, teams move faster, leaders are more comfortable expanding automation, and exceptions become manageable events rather than organizational bottlenecks.

This is where the Ownership Layer stops being a risk-management function and becomes a competitive advantage.

The companies that realize the greatest value from AI will not necessarily be the ones with the most sophisticated models. They will be the ones that maintain ownership of outcomes as automation scales.

Because customers do not experience AI.

They experience outcomes. And the ability to consistently own those outcomes is where competitive advantage begins.

Read & discuss on LinkedIn
September 1, 2026LinkedIn Article

The Ownership Layer

Most organizations think they need better AI. What they actually need is better ownership.

The first wave of AI transformation focused on capability: Can the model answer questions? Can it resolve tickets? Can it automate workflows?

The second wave is something much harder: Who owns the outcome when the workflow breaks?

Customers don't experience organizations as systems. They experience outcomes. They don't care whether the issue originated from:

  • An AI model
  • A policy engine
  • A CRM workflow
  • A third-party vendor
  • Or an internal process

They care whether their problem is solved.

This creates a new operational requirement that didn't exist in traditional automation. Not a support layer. Not an escalation layer. An ownership layer.

A function responsible for:

  • Detecting exceptions before they become failures
  • Coordinating resolution across systems and stakeholders
  • Maintaining accountability when automation reaches its limits
  • Continuously feeding lessons back into the AI ecosystem

This is why the future of customer experience isn't AI versus humans. It's AI supported by the right ownership model.

The organizations that succeed with AI won't be the ones that automate the most. They'll be the ones that maintain ownership of outcomes at scale.

Read & discuss on LinkedIn
August 25, 2026LinkedIn Article

When My Food Delivery Goes Wrong, Who's Accountable?

You order dinner through a delivery app.

The food arrives cold and is missing an item.

You contact support.

“The restaurant marked the order complete.”

So, you call the restaurant.

“We packed everything that was on the ticket.”

You contact the driver.

“I delivered the bag exactly as I received it.”

The app says it simply connected you to the service. At that point, you don't care who is technically correct. You just know your dinner is wrong.

Everyone can explain what happened, but nobody seems responsible for fixing it.

That's what an accountability vacuum feels like. When a human makes a bad decision, someone owns it. When AI makes a bad decision, ownership gets surprisingly blurry.

A customer doesn't care whether a failure came from a policy, a workflow, a vendor, or an AI model. They care that the outcome was wrong. Yet many AI deployment discussions focus on capability while ignoring accountability.

The AI vendor says the model performed as designed.

The enterprise says the workflow followed the policy.

The customer is left with a problem nobody seems to own.

As AI adoption increases, this becomes more than a customer experience issue. It has become a business risk issue. The challenge is not determining who caused the problem. The challenge is determining who owns the outcome. Historically, organizations built accountability into human processes. As more decisions move into automated systems, the accountability layer becomes increasingly important.

Not because AI is failing. But because AI is succeeding at scale. And scale magnifies exceptions.

The organizations that navigate this successfully won't be the ones asking:

“Can AI do this?”

They'll be asking:

“When something important falls outside the workflow, who owns the outcome?”

Read & discuss on LinkedIn
August 18, 2026LinkedIn Article

The Escalation Fallacy

One of the most common assumptions in AI is also one of the least examined.

If AI can't handle something, escalate it.

At first glance, this sounds reasonable. But it assumes failure is a moment. In reality, failure is usually a process.

Conversations rarely go from “everything is fine” to “escalate immediately”.

They drift.

Context gets lost.

The customer repeats themselves.

The interaction becomes increasingly misaligned with what the customer is actually trying to accomplish. By the time an escalation occurs, the problem has often existed for quite some time.

By the time an escalation occurs, the actual problem has usually existed for much longer than the escalation itself.

The escalation is simply the moment when the organization finally notices.

But notice what this reveals: the real value isn't on responding faster once an issue is escalated; it's recognizing risk before it reaches that point. Yet most of the conversation around AI remains focused on efficiency. We talk about automating repetitive work, reducing handling costs, improving response times, and scaling operations without scaling headcount. All those things matter. They're measurable. They're easy to justify.

But they also represent the easiest part of the problem.

What receives far less attention is what happens when an interaction stops being routine. This matters because the most important conversations are rarely the highest-volume ones.

They're the exceptions. The edge cases. The situations where policy, judgment, accountability, and business outcomes collide.

As organizations race toward AI-driven operations, a question is emerging, who owns the conversation when it no longer fits the workflow?

The answer is more important than most people realize.

Stay tuned for the next post.

Read & discuss on LinkedIn
August 6, 2026LinkedIn Article

The Dress

Customer: “Hi, I really need help! My bridesmaid’s dress hasn’t arrived, and the wedding is on Saturday!”

AI: “I’m so sorry for the inconvenience. I can see your order is delayed. It’s currently scheduled to arrive on Monday morning. I’ve processed a full refund for you.”

Customer: “I don’t want a refund! I need the dress by Saturday! Can you overnight it, call the warehouse, anything? Please, I just need it here.”

AI: “I completely understand how frustrating this is. As an apology for the extreme inconvenience, I’ve added a $25 coupon to your account for a future purchase.”

The line goes quiet. The problem isn’t solved. The customer is beyond frustrated.

Most AI deployment strategies assume that removing humans from a workflow is a net efficiency gain. Malasakit.AI understands this assumption breaks down at the exact moments where judgment, accountability, and escalation become necessary. A frustrated customer gets routed through the same script as everyone else. A confused, upset, or vulnerable person gets a technically correct answer that solves nothing, because the situation called for judgment that a script cannot anticipate, and an AI system did not escalate.

These moments are rare on a percentage basis, and high stakes on every other basis, for the customer, for the brand, and increasingly, for regulators. As AI adoption expands, the absolute number of these exceptions grows alongside it. More automation creates more total interactions, which creates more edge cases requiring human judgment.

To AI, it was never about the dress. It was about mitigation. To the customer, it’s only about the dress.

This is the structural gap that Malasakit.AI bridges, translating AI adoption into business outcomes.

Read & discuss on LinkedIn
More From Our Feed
July 29, 2026LinkedIn Article

Why Solving the Tier-2 Agentic AI Gap is Urgent

When enterprises deploy an AI frontline, the justification is pretty straightforward: automated interactions cost pennies compared to human conversations, so total support costs should decrease substantially.

This is true, as long as this doesn't mean replacing all humans in the support chain. As discussed in prior posts, someone has to pick up the slack for the layer that is required between frontline AI and enterprise high-value subject matter experts.

Let's discuss how large this gap is.

When enterprises make it effortless and instant for a customer to interact with their company, they will do it significantly more often. Total interaction volume doesn't stay flat; it explodes.

This is a long-standing somewhat-non-intuitive phenomenon called Jevons' Paradox and was first coined when more efficient steam engines paradoxically caused coal consumption to skyrocket, rather than decrease. In the AI era, frontline automation is the steam engine, and customer interactions are the coal.

Even if AI raises the bar from where it is today to 90+% resolution via frontline Tier-1 resolution, the remaining ~10% becomes massive due to Jevons' paradox.

This tidal wave of Tier-2 work has to fall somewhere. And legacy providers for such services lack an efficient way to address this new work without spooling up multiple teams, which eats back into cost savings by replacing frontline support with AI.

At Malasakit.AI, we are enabling this gap to be mitigated via a proprietary, efficient, and canonical solution.

Please stay tuned.

Read & discuss on LinkedIn
July 22, 2026LinkedIn Article

Tier-2 for Agentic AI Doesn’t Have to Mean More Headcount

Once enterprises realize that someone needs to be between AI and their high-value subject matter experts (e.g. doctors, lawyers, engineers), a new concern arises.

Enterprises assume they will need to build out multiple, separate, specialized teams: one to watch live conversations and intervene, another to be on constant stand-by to process back-office exceptions, and yet another to provide feedback back into the AI models.

Adding multiple new teams is obviously untenable. The efficient solution is to enable, via software, a significantly upgraded version of the same Tier-2 team that exists today, to intervene when AI reaches its limits, resolve operational exceptions, and continuously improve system performance.

Read & discuss on LinkedIn
July 8, 2026LinkedIn Article

The CX Hot Potato

Your hotel marks you a no‑show and when you arrive, you have no hotel room. The airline says the delay isn’t their responsibility and the hotel informs you that there’s nothing they can do because you used a third party booking platform.

You call the booking platform and their “Tier-1” customer support team picks up. Tier-1 is designed for high-speed, predictable requests. Because the script instructs the agent to “inform the customer they must take it up with the hotel directly,” they cannot deviate. No matter how much you argue, Tier-1 cannot rewrite a policy or change the script.

This forces a handoff to a dedicated exception handler (“Tier-2”) because brokering resolution now involves cross‑entity negotiation, ambiguous policy interpretation, and potentially real financial or legal accountability.

The prevailing belief for Tier-2 is that agentic AI will extend their Tier-1 support to “cover this soon”, or that customers will just absorb Tier‑2 alongside their Tier‑3 SME (subject matter expert) support layer.

Both assumptions lead to the same outcome: no one owns Tier‑2.

The work doesn’t disappear, the lack of ownership becomes a bottleneck. High stakes transactions stall, customer satisfaction degrades, and compliance risk increases.

That blocker is the Tier‑2 ownership gap: agentic AI providers intentionally and deliberately don’t provide Tier-2 support, and enterprise customers don’t want their SMEs (doctors, lawyers, engineers, etc.) handling administrative edge cases.

This is a gap that neither agentic AI providers nor end customers are structurally designed (or incentivized) to absorb.

Read & discuss on LinkedIn

Join the Discussion on LinkedIn

Our latest thought pieces are published weekly. Follow our page to weigh in on the debate and read the full discussions.