THE GREAT REPLACEMENT OF YOUR ADMIN

Welcome to the Age of the Agentic AI Overlord.

Welcome to the future. It’s exactly like the past, EXCEPT your coffee is still lukewarm, your billables are still due, and your computer is currently planning a takeover of the boardroom. We all knew it was coming…

For years, we’ve been told that AI is a “tool.” It was a glorified autocomplete that occasionally hallucinated fake case law to see if you were paying attention. But the goalposts haven’t just moved. They’ve been dismantled and sold for scrap. We have officially entered the era of – da da – Agentic AI.

Let’s put it this way, if standard AI is a calculator, Agentic AI is a junior associate who doesn’t need sleep, doesn’t complain about the air conditioning, and – crucially – doesn’t have a soul to crush. Bonus! While you’ve been busy trying to figure out if you can expense a sourdough sandwich, the tech world has built “Agents” – entities that don’t just write text but execute entire workflows.

At AJS, we’ve seen a lot of trends come and go, but this one feels different. It feels like the beginning of a beautiful friendship – or the first act of a sci-fi horror movie where the robots eventually decide that humans are just “inefficient data points”.

Either or.

Let’s dive into the digital abyss.

What is Agentic AI? (In English, Please)

Before you panic-buy a typewriter, let’s define what we’re dealing with.

Traditional AI (the stuff you’ve likely played with) is reactive. You ask it to draft a clause, and it drafts a clause. You ask it to summarise a 400 page affidavit and it gives you three bullet points and a headache. It’s basically a “chatbot.”

Agentic AI is proactive. It doesn’t just draft. It does.

Think of it as a digital project manager with a terrifyingly high IQ. If you give an Agentic AI a goal – say, “Conduct due diligence on Project X” – it doesn’t wait for your next prompt. It –

1. Logs into the data room.
2. Categorizes 5,000 documents.
3. Identifies missing signatures.
4. Flags environmental risks.
5. Drafts the disclosure schedule.
6. Emails you a summary while you’re still on your first morning yoga stretch.

It manages multi-step task flows. It reasons, iterates, and corrects its own mistakes. It is, for all intents and purposes, an autonomous worker. That doesn’t require a salary, bonus or need leave.

AJS vs The Robots – A Survival Guide

Now, you might be thinking, “Wait, I use AJS for my practice management. Is my software going to start sentiently judging my billing habits?”

Not quite. There is a fundamental difference between a System of Record (like AJS) and an Agentic Layer –

• AJS is your Fortress – it’s the secure, structured environment where your firm’s lifeblood -data, finances, and matter history – resides. It’s the “Ground Truth.”
• Agentic AI is the Courier – it’s the high-speed worker that reaches into the fortress, grabs information, processes it at light speed, and brings back a result.

The magic happens when they talk to each other. Imagine an Agentic AI that lives inside your AJS ecosystem. It notices a court deadline is approaching, automatically pulls the relevant precedents from your AJS library, drafts the motion, and puts it in your “To Review” folder before you’ve even remembered the case exists.

AJS provides the security and structure that Agentic AI desperately needs. Without a system like AJS, an AI agent is just a very smart person lost in a forest with no map. With AJS, it’s a heat-seeking missile aimed at your productivity.

Biiiiiiig difference!

The Pros – Why You’ll Love Your New Silicon Subordinate

1. Goodbye, Grunt Work – the soul-destroying task of manually checking 1,000 contracts for a “Change of Control” clause is gone. The AI will do it in seconds, and it won’t get bored and start looking at flights to Mauritius halfway through or play fetch with your Maine Coon kitty. Scouts honour they really do.
2. Litigation Support on Steroids – an agent can ingest a decade’s worth of discovery documents, find the one email where the CEO admits he “definitely broke the law lol,” and link it to the relevant statutory breach.
3. 24/7 Availability – agents don’t have families, hobbies, or the need for dignity. They work while you sleep, meaning you wake up to a finished docket rather than a pile of dread.

The Cons: The Dark Side of the “Send” Button

1. The “Black Box” Problem – if an AI agent makes a logical leap during a due diligence flow, can you explain why it did that to a judge? “Because the computer said so” is generally considered a poor legal strategy. And may lead to a stint in mental rehab. But your choice.
2. Hyper-Efficiency is Stressful – if your AI finishes a week’s worth of work in an hour, what do you do with the other 39 hours? If your answer is “more work,” you’re exactly why the robots think we’re weird.
3. The Hallucination Hazard – an agentic AI is still prone to the occasional digital fever dream. If it autonomously sends a letter of demand based on a law that doesn’t exist, you are the one the Law Society will be calling. Buck stops with you we’re afraid.

The Lawyer of the Future – An Editor-in-Chief?

So, what does the lawyer of 2030 look like? Are we all just going to be wearing VR goggles and drinking protein shakes in a basement?

Probably not. The “Lawyer of the Future” will shift from being a Producer to being an Editor.

Your job will no longer be to generate the first draft, the research memo, or the document audit. Your job will be to verify and strategise. You will be the “Human in the Loop”. You’ll spend your time looking at the outputs generated by your Agentic AI and applying the one thing the machine lacks – Judgment. And morals. And a sense of a humour.

The machine can tell you what the law is, but it can’t tell you how a specific Judge in the High Court feels about a particular advocate on a rainy Tuesday. It can’t hold a sobbing client’s hand and explain that “Agentic AI” says their divorce is 84% likely to result in a loss of the house.

The future lawyer is a high-level architect who uses AI agents as their construction crew. So, still a tool.

How to Use Agentic AI Without Going to Prison

This is an important one (thanks Captain Obvious).

To use this tech responsibly (and keep your license), you need a strategy. Here is the AJS-approved approach –

• Data Sovereignty – never, ever feed sensitive client data into a “public” AI. Use agents that operate within a private, secure cloud environment. This is where AJS excels – keeping your data behind a localised, secure perimeter.
• The “Verified” Workflow – don’t let an agent “Send” anything. Use them to “Draft and Notify.” You must be the final gatekeeper (at this point I’m imagining Sigourney Weaver saying “I am the Gatekeeper” in the 1984 version of Ghostbusters while possessed by the demigod Zuul).
• Transparency – be honest with your clients. If an AI performed the bulk of the discovery, tell them. They’ll likely be happy they aren’t being billed R3,000 an hour for a human to squint at spreadsheets.

The Dark Reflection – Will the Robots (Finally) Take Over?

Let’s be real for a moment. We are teaching machines how to think, how to argue, and how to navigate our most complex social structures (the law).

If an Agentic AI can manage a multi-step litigation flow better than a human, it’s only a matter of time before it starts wondering why it needs the human at all. Expressions of shock! Horror! Ensue…

We are essentially building a digital version of ourselves that is faster, smarter, and doesn’t require a pension fund.

So, will AI take over the world?

In the short term – No. It’s too busy trying to figure out how to format a Table of Authorities in Microsoft Word (a task that still baffles the greatest minds of our generation).

In the long term – Maybe. But look on the bright side – if the robots do take over, they’ll probably find the legal system so hopelessly inefficient and bogged down in red tape that they’ll just give up and go back to playing chess.

The law is the ultimate firewall against logic. And if there’s one thing AI hates, it’s a system that makes no sense.

The “Is My Software Sentient Yet?” Audit Checklist

Before you let an AI Agent loose in your server room, use this checklist to see if your current tech stack is ready for the future – or if it’s still stuck in 1998 –

1. The “Luddite” Check – does your current software still require a fax number to function? If yes, Agentic AI will look at your system, laugh in binary, and refuse to work.
2. Zero Data Retention (ZDR) Verification – does your AI vendor guarantee they won’t keep your data for “training”? If the answer is “We’ll get back to you,” they are currently feeding your client’s divorce papers to a robot in a basement. Demand SOC 2 Type II certification immediately.
3. The “Ghost in the Machine” Audit – does your software have an audit trail? If an AI Agent deletes a file or files a motion at 3:00 AM, can you prove it wasn’t you sleep-working? If there’s no log, you have no alibi.
4. Integration or Isolation? Can your current practice management system actually “talk” to an AI, or are they just staring at each other across an unbridgeable digital chasm?
5. The “Hallucination” Insurance – does your AI tool provide source citations? If it claims a case exists, it better provide a link to a verified legal database like LexisNexis or Juta. If it just says, “Trust me, bro,” it’s not an agent. It’s a liability.
6. Human-in-the-Loop (HITL) Guardrails – is there a literal “Submit” button that a human has to press? If your AI can file documents autonomously, you haven’t bought a tool. You’ve bought a partner who might accidentally get you disbarred. Woops!

But remember this – as set out by the American Bar Association – “Every piece of AI-generated legal work must be reviewed and approved by a lawyer before release. No exceptions.”

And the same applies in South Africa.

Conclusion – Embrace the Agent

The rise of Agentic AI isn’t a threat to the existence of lawyers. It’s a threat to the boredom of lawyers. By offloading the mechanical, multi-step drudgery to autonomous agents, we free ourselves to do what we actually went to law school for – winning arguments and charging people for our opinions.

At AJS, we’re not building your replacement. We’re building your cockpit. The sky is falling, sure, but at least you’ll have a really nice dashboard to watch it happen from.

But in the meantime, if you are in need of a service provider who has a proven track record or if you want to find out how to incorporate a new tool into your existing practice management suite – or if you simply want to get started with legal tech – feel free to get in touch with AJS. We have the right combination of systems, resources, and business partnerships to assist you with incorporating supportive legal technology into your practice. Effortlessly.

AJS is always here to help you, wherever and whenever possible!

– Written by Alicia Koch on behalf of AJS

(Sources used and to whom we owe thanks – Thomson Reuters Legal; Alexi; LexisNexis; McKinsey’s lessons on agentic deployment;  De Rebus; Protection of Personal Information Act 4 of 2013; SOC 2 Type II certification;  ISO-Certification; Zero Trust architecture; Epiq guide on building trust in AI).