Welcome to the new year!

I am so ready for AI in 2026, because this is the year AI stops sitting in a separate chat window and really takes over some tasks and does them for us.

We are this much closer to AI employees that can take over entire tasks.

In 2025 we laid the foundations. We selected our tools, created policies, trained our teams, and built an AI‑literate workforce.

Now those people are asking a new question:

When will it do my tax return for me? When will it annotate the contract for me? When will it prepare the case?

Why you need to understand AI Agents and AI Employees?

First, so you can implement what is ready.

Second, so that when people try to sell you AI agents or AI employees, you can tell whether they are offering something genuinely valuable or something you could build yourself in one prompt.

There are three gaps between today’s technology and AI Agents that can actually do the work for you.

We need to understand where each one is at, and what we can do today:

Gap 1: Playbooks

If I want AI to do a tax return or review a contract for me, “general tax knowledge” is not enough.

It needs my way of doing things. My firm, my processes, my preferences.

It needs a playbook.

We can already create AI agents with solid system prompts that spell out exactly how we want a task done. They can:

run in the background and triage the inbox
categorise the calendar
monitor a spreadsheet or folder and act when something changes
or wait for us to say, “Go and do the thing”

We have been creating these agents in tools like Microsoft Copilot, ChatGPT, Gemini, Zapier and other tools. If you have not been doing that yet, this is the time to get on top of it.

This is a mature area. We built these throughout 2025 and will build many more in 2026.

I am running a workshop on agents in 2026. You might want to join. www.inbal.com.au/events

Gap 2: Deep Context

Next problem.

Even if AI knows everything on the internet, it does not know this particular client, this file, this matter.

For AI to do that tax return or contract well, it needs access to:

your practice management system
your files and working papers
your policies and preferences
similar work you’ve done in the past

The technology to connect these is being built right now.

You will hear terms like “MCP server”, often described as the USB of AI.

Think of this as the universal connector that lets your AI tool of choice, Copilot, ChatGPT, whoever you use, look into your other systems, read the data there and, increasingly, take action there.

This gap is halfway closed.

The standards have been established, the technology is there, now app vendors are releasing early versions of their servers so AI can look into their data.

This technology is still a bit hit and miss. Innovators and early adopters are playing with it.

If you are not working with it yet and you are not clear on what an MCP server is right now, don’t worry. You will probably need to know in a few months as it matures.

This is an area to watch and stay aware of, I’ll let you know when it’s time to jump in.

Gap 3: Hands

The final gap is hands.

We need to give AI hands so it can touch our systems.

It needs to be able to go into your workpaper in Excel and type in it. It needs to be able to go to a website and complete a form for you, upload or download documents, copy here and paste there, and enter data where it belongs.

That is where AI-powered browsers come in.

Tools like Comet from Perplexity, Atlas from OpenAI, similar capabilities in Microsoft Edge, and in the Google ecosystem are all being worked on as we speak.

This is where you can ask the browser not only to find things for you, but also to do things in the browser. You can say:

“Complete this form.”
“Place this order.”
“Fill this workpaper.”

This technology is still young and has a lot of risks.

We are giving AI the ability to act, so we need to keep a close eye on what it does.

For most firms, the right move today is to have one or two trusted innovators experimenting with these tools and letting you know when they are ready. I will also let you know as they mature.

This is probably the most open and risky area right now. And if you jump in, you will spend much time before you get any returns.

Where we will be by the end of 2026

By the end of 2026, we are heading towards a world where your chosen AI tool:

has your playbooks
has deep context from your systems
can drive your browser and other key tools

At that point, “AI employees” in accounting and law are no longer just a slide in a keynote. They sit alongside your human team and get real tasks done.

What you should do now

Here is how I recommend you approach this, without the hype.

1. Start with simple playbook agents

Whatever platform you use, the tools are already there, Copilot, ChatGPT, Gemini, Google, Zapier and others.

Start by creating small, reliable agents that follow a clear playbook. That is where most firms should be right now. This builds capability and trust inside your firm.

I am running workshops on this. Join me and we will build them together. www.inbal.com.au/events

2. Stay aware of context connectors and hands

Keep an eye on MCP servers, agents and agentic browsers.

If you are an innovator or early adopter, you might already be playing with them.

If you are in the early or late majority, this is mostly watch and wait. The technology is there in theory, but in practice it is still being rolled out across your systems.

I will let you know when these are ready for mainstream firms.

Want to go deeper with me?

I have booked in webinars and workshops on exactly these topics so we can work on them together.

Have a look at what is coming up and join me: inbal.com.au/events

Let’s make 2026 the year AI moves from a separate tab to a real part of your team.

—

Inbal Rodnay

Guiding Firms in Adopting AI and Automation

Keynote speaker | AI Workshops | Executive briefings | Consulting CIO

Want to receive these updates straight to your inbox? Click here: www.inbal.com.au/join

When you are ready, here is how Inbal can help:

Transform your firm in 30 Days with the 30days to AI Program — Bring your entire team on the AI journey in just 30 days. This program is designed to give your team a solid foundation in using generative AI in responsible and impactful ways. Inbal helps you choose your AI tools, create an AI policy and train your team.