Your Invoice is an Actor that Wants to Get Paid

Once upon a time, your invoice was just a record sitting quietly in a database, waiting for humans to process it. Poor thing—trapped in digital purgatory like an abandoned pet hoping someone remembers to feed it. For decades, business objects have been little more than collections of fields and values—the digital equivalent of stuffed animals. They don't do anything; they just sit there, looking cute but fundamentally helpless.

Then came automation and AI—a layer built on top of passive data, doing things to objects rather than objects doing things themselves. We've become quite clever about it, really. We've built elaborate ETL (Extract, Transform, Load) pipelines that shuttle these lifeless objects between disparate enterprise systems. For years, this digital plumbing has largely fallen to humans—copying data from the CRM, reformatting it for the finance system, manually triggering workflows, and becoming the connective tissue between siloed applications.

Now we're increasingly using intelligent systems to help with this data shuffling—training AI to reformat documents, build connections between systems, and automate some of the transformations. But fundamentally, it's still just propping up lifeless data. It's like sophisticated puppetry—we're pulling strings to animate these passive objects rather than giving them any intrinsic capability to act.

But what if that's all backwards?

The records in your database are awakening. Your business objects are stirring with purpose and agency.

If 2024 was the year chatbots became widespread, 2025 will be the year when agentic AI takes center stage—at least that's the widespread prediction. But there's a crucial distinction to make.

Most conversations about agentic AI focus on making processes more intelligent. We're building AI agents that can handle workflows and tasks—essentially smarter process handlers that still manipulate passive data.

But what if we're missing something more fundamental? What if those objects being handled could themselves become intelligent? What if, instead of building smarter systems to push data around, the data itself became aware of its own state, history and purpose?

This would eliminate enormous process overhead. Rather than needing complex orchestration layers to coordinate the movement of passive objects between systems, each object could manage its own journey. This is the future we need to move toward—rapidly.

What if your invoice understood its purpose was to get paid—and actively pursued that goal? What if it could navigate your organisational chart, find the right approvers, nudge them when they're being slow, gather missing information on its own, and only bother humans when truly necessary?

This isn't science fiction—it's the next evolution of enterprise software, and it's already beginning. Welcome to the age of Agentic Business Objects (ABOs).

The Race We're Running vs. The Race That's Coming

Right now, we're all sprinting in the same race. We're training employees on prompt engineering, building custom GPTs for admin tasks, and crafting clever automations that transform data between formats. If you're not doing these things, you're already behind.

And make no mistake—these are valuable steps. They're delivering real efficiency gains. They're solving immediate pain points. They're the race we need to run today.

But let's be honest—we're still just speeding up broken processes.

It's like we've spent decades building a Rube Goldberg machine across our entire organisation, and now we're carefully oiling each ridiculous component so the marble rolls a bit faster before it hits the hamster wheel that powers the miniature catapult.

The smart play is to sprint in today's race while training for tomorrow's marathon. The next revolution focuses directly on making business objects better at handling themselves, rather than enhancing human capabilities to manage them.

Digital Feudalism: The Enterprise Data Problem

Your business isn't inefficient because your people are lazy—it's inefficient because your data is trapped in feudal kingdoms refusing to talk to each other. I see this with our customers all the time: data silos, five different ERP systems, massive data fragmentation across the enterprise.

Follow a customer record through your business: captured in a form, transferred to CRM, copied to marketing, duplicated in support, recreated in finance—five copies of the same entity, all slowly drifting out of sync like siblings raised in different countries who no longer recognise each other at family reunions. By the time your customer data reaches its teenage years, it's developed five different personalities and probably needs therapy.

Every day, armies of knowledge workers waste precious brain power on data shuffling. They're the digital equivalent of those people who used to manually connect telephone calls—except instead of connecting copper wires, they're connecting bits of information trapped in incompatible systems. And even many of the AI augmentations that we are currently implementing just focus on speeding up this process, not on solving it.

How much human potential is being wasted on data shuffling? How many talented professionals are spending their days copying and pasting between screens? Studies suggest knowledge workers spend up to 25% of their time on such administrative tasks—which means your company is paying skilled workers to be very expensive carrier pigeons.

Your organisation chart might say "Finance Director" or "Marketing Strategist," but the job description should read "Professional Data Sherpa: Must Excel at Carrying Information Across Treacherous System Boundaries."

Meet Your New Colleagues: Agentic Business Objects

Agentic Business Objects aren't just a faster way to do the same old tasks—they're a completely new architecture for how business works.

Stop thinking of your business data as passive records and start thinking of them as motivated actors in your business drama. Your invoice isn't just a document—it's a character with a mission. Your support ticket isn't just a record—it's an ABO working toward resolution.

The technologies making this possible have been quietly assembling—and they've reached a tipping point:

1. The Actor Model: A computing approach where each "actor" (your invoice, your ticket, your customer record) operates independently and communicates through messages. It's like giving each business object its own brain and mobile phone.

2. Durable Execution: Systems that allow processes to pause, wait, and resume over long periods—even days or weeks. Your invoice doesn't forget what it's doing if someone turns off the computer.

3. State Machines: Formal ways to define how objects transition between different conditions. Your invoice knows whether it's draft, pending, approved, or paid—and what needs to happen next.

4. Large Language Models: The intelligence layer that gives these objects the ability to understand context, make decisions, and communicate naturally. The brain inside the machine.

Together, these technologies transform your business objects from passive data into active participants. They're no longer just things to be manipulated—they're entities with goals to accomplish.

The Invoice That Wants to Get Paid

Let's make this concrete with our invoice example. Your traditional invoice process probably looks something like:

1. Finance creates invoice

2. Invoice sits in approval queue

3. Approver eventually checks queue

4. Invoice sits in payment scheduling

5. Payment processor eventually processes it

6. Invoice sits in reconciliation

7. Someone eventually reconciles it

At each step, the invoice is passive—a digital piece of paper waiting for a human to notice it. The default state is "stuck" until someone unsticks it.

Now imagine the agentic version:

Your invoice understands its own approval requirements. It knows it needs sign-off from procurement and the department head for expenses over £10,000. It automatically messages both approvers: "I need your approval by Thursday to ensure payment within terms."

When the department head is on holiday, it notices the out-of-office reply and intelligently escalates to the designated backup approver: "Dr. Thompson is away until next week, and this invoice requires approval by Thursday. As her designated backup, could you please review?"

It gathers missing information independently—noticing the cost code is missing, it checks recent similar invoices from the same vendor and suggests the appropriate code.

It communicates with other systems without human help—once approved, it coordinates with the payment system to schedule optimal payment timing based on cash flow projections and vendor terms.

It only involves humans when necessary—flagging genuine exceptions that require judgment: "This invoice amount is 40% higher than our average with this vendor. Would you like me to query this with them before processing?"

This isn't fantasy. These capabilities are where enterprise technology is heading. While most current implementations still rely on external AI processes rather than truly agentic objects, the foundations are being laid for this transformation.

An agentic invoice could:

• Negotiate payment timing directly with the vendor's receivables system

• Identify and flag unusual spending patterns across departments

• Automatically adjust payment priority during cash flow constraints

• Learn from past approvals to anticipate exceptions

This enables capabilities that transcend mere efficiency improvements. Imagine everything your best accounts receivable and accounts payable professionals do, but happening automatically, 24/7, at tremendous scale, without coffee breaks or holiday leave.

Beyond Finance: The Enterprise-Wide Revolution

Sales: Opportunities That Hunt Their Own Closure

Imagine a sales opportunity that's as motivated to close as your best salesperson—and just as disappointed when it doesn't hit its quota.

Your CRM record actively researches prospect news and social media, identifying the perfect moment for outreach. "The CFO just tweeted about cost-cutting measures—shall I draft a message highlighting our ROI figures?" It drafts personalised communications based on the prospect's recent concerns. It coordinates resources needed for closing—scheduling demos, preparing custom materials, briefing sales teams on the prospect's hot buttons.

When the deal stalls, it doesn't just sit there sulking in the database—it suggests new approaches based on similar deals that succeeded. "Three comparable opportunities closed after a product demonstration focusing on compliance features. Should I schedule one for next Tuesday?" It's constantly working, even when your sales team is asleep. The most enthusiastic team member you'll never have to buy coffee for.

Support: Tickets That Drive Toward Resolution

Your support ticket wants to get resolved as much as your customer wants their problem fixed—think of it as a tiny digital superhero with a mission to save the day before the customer's patience expires.

It gathers relevant context before any human gets involved—pulling up the customer's history, recent orders, and similar issues. "Ah, I see Ms. Johnson had the exact same issue last month, and changing the network settings fixed it in 30 seconds!"

It routes itself to the right specialist based on the specific problem details. It suggests solutions based on similar resolved tickets. It proactively updates the customer on its status. "Good news! I've found a solution to your problem and am just waiting for final verification."

HR: Onboarding Processes That Coordinate Themselves

Onboarding shouldn't feel like herding cats—it should be a coordinated dance where every step happens at the right time. But in reality, it's usually more like a chaotic game of Telephone played by blindfolded participants with earplugs.

An agentic onboarding process spans HR, IT, facilities, and team systems. It coordinates the timing of all resources and access. It adapts scheduling based on everyone's availability. It provides continuous status updates to all stakeholders. Most importantly, it eliminates those maddening gaps where new employees sit idle because someone forgot to order their laptop or set up their email. "You want me to do what with this pen and paper? What century is this?"

When these agentic objects work together, they create an intelligent ecosystem—a digital workforce that collaborates just as your human teams do. They coordinate across departments, continuously optimise processes, and balance resources and priorities without constant human intervention. It's like having tiny, diligent elves working throughout your organization, except they don't demand cookies or unionize.

In other words, dear business leader, the hills are alive with the sound of music.

What This Means For Employees

New Skills to Develop

The future belongs to those who can orchestrate teams of both human and AI actors.

You'll need to develop new skills:

• Defining goals and constraints for autonomous systems

• Exception handling when agents face unusual situations

• Systems thinking and process design

• Oversight and governance of AI decisions

It's less about doing tasks and more about setting direction, handling edge cases, and ensuring the overall system works as intended.

How Roles Will Evolve

You won't simply be replaced by AI—your role will transform in complex ways. The most realistic scenario? A mixed future where some roles elevate while others face significant disruption.

Current roles will transform:

• From process operators to process designers

• From data entry to exception handling

• From routine approvals to strategic decisions

• From reactive to proactive management

For many, this means liberation from drudgery and a chance to focus on higher-value work. But we shouldn't sugarcoat it—some roles heavily centered on data manipulation and routine processing may shrink substantially. The transition will require reskilling and adaptation. The future belongs not to those who resist change, but to those who ride its wave skillfully.

Questions to Ask Your Software Vendors

Is your enterprise software preparing for this shift, or is it doubling down on the old paradigm?

Ask your vendors:

• What's your vision for agentic business objects?

• How do you enable object-to-object communication?

• What level of autonomy can your platform support?

• How do you handle the balance of AI and human decision-making?

Their answers will tell you whether they're building for yesterday's problems or tomorrow's opportunities.

First Steps for Forward-Thinking Leaders: don’t Try to Boil the Ocean

Start small, prove value, then scale up.

Begin with one high-value process for a pilot. Focus on clear goals and measurable outcomes. Build in appropriate oversight and controls. Expand gradually as confidence grows.

Companies that are currently finding success with agentic technology didn't try to boil the ocean—they started with targeted use cases and expanded as they proved the approach worked.

From Operators to Orchestrators

We're moving from being operators to orchestrators—from doing the work to directing an ensemble of digital performers.

Think of your business as an orchestra. Traditional management meant playing each instrument yourself, one at a time. Then we moved to conducting human musicians. Now we're entering an era where you conduct an orchestra of both human and AI performers, each playing their part in perfect harmony.

By delegating the mechanical tasks, we elevate the meaningful work. When routine tasks are handled autonomously, humans can focus on their strengths—creativity, empathy, judgment, innovation. The paradox? As our systems become more automated, our work becomes more human.

The records in your database are waking up. Will you be ready to lead them?