Just because AI can, doesn’t mean it should
I met an operations manager from a construction company this week. Let’s call him Darren (because that’s his name). Darren had been a prospect for a few weeks and we’d sussed a time to meet at his office. Unlike a normal prospect, however, Darren had tried automation before, and tried it on a bunch of different tasks across different software. He had it parsing CAD pdfs, drafting quotes, scraping pricelists and drafting emails, among a bunch of other tasks. He hated it. He turned it off pretty soon after turning it on.
Here’s where Darren went wrong: he had jumped straight to using AI for the automations. He had MS Copilot doing all sorts of stuff purely from chat prompts and Claude acting as an integration between the different pieces of software. The software wasn’t actually integrated, it was just pretending to be (using AI). It was incredibly expensive, more than a new employee he reckoned. It also didn’t work that much - most of the time, AI would complete the tasks correctly a couple of times and then find something to cock up on the next run. Darren had identified some really good tasks to automate, put in the automations, and somehow ended up spending more time fixing the automations than working on business growth like he intended.
AI is incredibly powerful, and it can do anything online, but it’s not usually the best solution. If a task happens the same every time with different data, you should automate it using programming and/or middleware instead. This is a fraction of the cost of AI tokens and you can trust it to do the same thing every time it runs.
So, if we shouldn’t use AI for recurring work, where is it best used?
Firstly, what you’re probably doing already - use it as an advisor. Bounce ideas off your chatbot of choice, make it suggest things to automate, and make it come up with the best solution for the automation ideas. It’s also excellent for grunt work: writing code, debugging code, spreadsheet formulas, transcribing, data manipulation and more. Stuff that used to take days of messing around can now be done in an hour.
Furthermore, it can build apps. You might hate your CRM (most people do). Try asking Claude Code or Lovable or any other AI app builder to make you a new CRM, with all the good bits of your current CRM and all the bits your current CRM is missing. It’s surprising and exciting how much it can build, entirely referencing plain text prompts. The new app won’t have any existing integrations, but it will have an API endpoint to enable them.
AI also does a great job of web scraping, but keep your prompts as specific as possible and make sure to double-check the links it references. I find it too much of a yes-man; it’ll explain how to do something, I’ll explain that isn’t possible, then it’ll go round in circles on a handful of ideas without just telling me that the thing isn’t possible. Ask it where it’s getting its information from, and have a look at the source to cut down wasted time on misunderstanding a problem.
Darren was pretty switched off the idea of automation, because his idea of automation was completely reliant on AI prompts. AI is capable of anything, but that doesn’t mean you should use it for everything.
-Fred
If you don’t enjoy work, automate it
Do you enjoy every online task you do at work? Great, you can close this tab and go back to watching wahs highlights. If you don’t, you should get it automated.
There aren’t any limits or constraints on automation. You can automate anything if you can explain how to do it. If it’s difficult to explain, or heaps of different stuff can happen during the workflow, or it’s very intuitive, use AI to make the automation. Wherever you are now, think of your least favourite online task at work. Is it quality control? Internal follow-ups? Missing or incorrect product info? What constantly or regularly pisses you off?
And even if you can’t think of a way of automating it, let us know. I’m 99% sure we can automate it, and we’ll let you know ASAP if we can’t.
-Fred
Use basic software and build it out
80% of business owners and decision makers I talk to about automation hate at least one piece of software in their business. However, they never change or improve it because it’s doing the bare minimum and there are bigger, easier priorities. Often, this software has outsourced management (an IT company is looking after it for you), isn’t integrated with anything, has zero customisation/wiggle room and every task is super manual. Does that sound familiar to your IT stack?
Whenever we scope projects to integrate software of this nature, it almost always emerges that the best solution is to change to a different platform that offers more customisation. The simpler an app, the more you can customise it for your needs. That’s why I love Excel and GSheets; they’re incredibly powerful and they start off completely blank. You can integrate anything, build anything, and AI actually knows about them and can write formulas that work. Oh, and they barely touch your wallet.
You can build any generic business processes on spreadsheets. I know I sound like an IT neanderthal saying that, but ERPs are just a whole bunch of spreadsheets (tables) anyway. I’m not saying nothing else is required - if you run production and manufacturing, an industry production tool is likely going to be necessary. But the point is that the best software stack for your business is starting from blank interfaces and building them out. It applies to CRMs for example: anyone that hates their CRM finds it way too complex and rigid, because it’s not designed for their business. You’re going to save a lot of long-term time, hassle and money if you start with a blank canvas on something like monday.com or ClickUp, as opposed to a function-heavy, expensive CRM like Hubspot or Salesforce.
Build stuff for your business. Avoid getting stuff that’s built for other businesses. Future you will be happy.
Size doesn’t matter
The word ‘automation’ is often perceived as a revolutionary project that will turn a business on its head for a month, like putting in a new ERP or doing a cloud migration. Conversely, I encourage business owners to start as small as possible. The bigger the project, the more stuff can go wrong. That’s why lots of IT projects stall and everyone wishes they had never started.
I’m not saying avoid new software - if your existing software doesn’t work, get new software. I’m saying that automation is a fantastic way of avoiding constant fights with your software and getting your productivity back. Look at every task/workflow happening in your business; what one is going to have the biggest impact on your staff productivity if it were automated and thereby removed?
Let us know. We can tell you if it’s a good thing to automate, or if you’re fighting a losing battle with your software. Starting small with automation is cheapest, quickest, avoids narcissistic consultants, brings immediate tangible results, and more often than not, the best option.
(hopefully you noticed how I purposely made this blog post small as well).
-Fred
You should have done automation yesterday
“But that’s what an automation firm would say.” Correct. It’s going to help our business. However, it’s also going to help yours.
Think about it like two car options for your commute: if you don’t have automation yet, you’re driving a car that you can only use 10 hours a day, it needs a new tank of fuel every day and new tires every week. It’s not broken, but it needs heaps of maintenance, governance and puts a huge amount of pressure on your finances. With automation, you have to stump up the cash to buy a new car. But fortunately, it’s available 24 hours a day, 7 days a week, uses a tank of fuel once a month, and new tires once a year.
Let’s play a quick maths game:
-Car 1 needs a new tank of fuel every working day. At $200 a tank, that’s $40k a year. It needs new tires every week, at $1000. That’s $50k a year. So it works 10 hours a day, 5 days a week and costs $90k per year. Sound familiar?
-Car 2 needs a new tank of fuel every month, and new tires every year. That’s $3400 annually. You would need to pay more than $80k for the car for it NOT to be valuable, and that’s only based on one year. Oh, and it works 24/7.
The fun part is that it really is that simple. Good automation replaces mundane work that admin staff are doing. It means you can replace them or push them into work that will help grow your business.
There’s no ‘best time of year’ or ‘investment quarter’ for automation. Yes, it can be a large outlay to get it done, but it pays itself off so quick that you’re doing your company a huge financial disservice by NOT getting it done ASAP.
-Fred
How to add value in a world of AI
I’ve been talking to a client recently who keeps identifying more and more opportunities for automation in his business. After all, automation is a drug. He’s also super hooked on Anthropic’s Claude, his AI bot of choice. As we find more stuff to automate in his business, he keeps defaulting to Claude as the solution and building automated workflows via Claude, all the time with zero clue what it’s doing. This introduces an awkward dynamic for us; he wants to keep costs down by avoiding our services and doing automation in-house, which is fair enough, but 90% of the time Claude doesn’t have the best solution. He’s constantly beating a dead horse to spit out inflated, unintelligible workflows that don’t holistically achieve his goals.
AI is incredibly powerful for writing code snippets, locating bugs and brainstorming ideas. It’s terrible at identifying the best solution for automation. It’ll soak up heaps of time and tokens to set up a workflow, then require babying through every time it runs. Back in reality, you can ask it to write an entire Python script to run instead, which is completely free, instant and doesn’t break all the time.
I can write code in heaps of languages, but why would I? An AI bot can do that for me. We add value by knowing the best solution for your automation opportunities, and default to engaging as little AI as possible to keep logic high and costs low. We add value by putting intelligible code in intelligible places. Our expertise is valuable to people who want the best end-to-end solution, not a tacky trellis that kills scalability.
People with good oversight of AI are 100x more valuable than people who use AI for everything, and 1000x more valuable than people who don’t use it at all.
It’s actually pretty hard to quantify our value in that area without being speculative. I realised that while writing this blog. Perhaps try AI for yourself first, and let us know when it shits the bed.
-Fred
Automation is a drug
The first thing I ever automated was SMS follow-up messages to contacts in a lead bank. Instead of the admin team going through their call list and sending a text whenever a lead didn’t answer their call (very frequently), they could type a code number beside the contact. This triggered a middleware-powered workflow to send the contact a personalised text based on the code number that the staff member had entered. It saved a heap of time, and errors; staff were able to 3x their outbound call volume, and lead contacts no longer received grammatically retarded texts.
I found it very enlightening. When you automate one process, your eyes are opened to what’s possible with automation, and suddenly you start identifying stuff wholesale. Even if you’re not the one that designs and configures the automation, you now know what’s possible, and develop a thirst to push other mundane tasks into the background.
Recently I met with a prospect (let’s call him Drake) who, at best, was very sceptical that we could provide any value to his business. Drake agreed to meet only on the condition that it would be a sub-1 minute discussion (Drake was a super busy beaver). I visited his office, said hello and handed him a business card. 20 minutes later, Drake had identified 10 different tasks to automate, and I had to cut the discussion short so my calendar didn’t get too harassed. It all started when we talked about their process of checking supplier invoices against their POs. Drake didn’t know that that could be automated, and for that reason, didn’t know that a whole bunch of other tasks could be automated too. We discussed a couple of other tasks, then he began suggesting automation opportunities under his own steam. He pushed for a follow-up consult to decide on their first project and rope in his ops manager (let’s call him Kendrick).
‘Drake’ didn’t want to engage automation services until he unlocked what’s possible. It’s possible that you’re in the same boat.
I tried OpenClaw
I hated it heaps. I think it’s because I like being in control of everything happening on my laptop and at my company. If you don’t like doing work on a computer, and don’t get rage baited by prose that is clearly AI-generated, I think you’ll like OpenClaw.
I got it doing a bunch of tasks: drafting emails to clients, importing bank feeds to Xero, unsubscribing from a few campaigns I hadn’t signed up for and drafting messages to group chats. I told it to use my communication style when drafting messages and let me know if it found anything slightly untoward during other tasks. It did a decent job of the latter, and got horribly lost attempting the former (thanking clients for their email - please don’t do this, it’s unnecessary filler text and you’re perceived as desperate to provide value. Emails should be as short and concise as possible).
OpenClaw also only works when you’re not working. It’s installed as an application package on your PC, and uses your PC to work. Obviously, you can install it on a different PC and provide access to all your stuff so it can work uninhibited, but that means getting another PC and providing it access to all your stuff. It functions as an actual employee, a personal assistant, instead of an ‘automation’. Don’t think of OpenClaw as automation. It simply does tasks that you tell it to do, step-by-step, generally much faster than a human. Automation is writing a program or connecting background middleware to make those tasks happen instantly and automatically. Good automation is inherently more robust and impactful than poking a stick at a bot and telling it to do things and stuff.
If you don’t like computer work and don’t care about being in granular control of everything happening in your work life, OpenClaw is probably really helpful. Install it and give it a go. But it’s not for those of us that like to send good messages and have a handle on everything.
As a side note, I haven’t tried it on Windows. I use a Chromebook which uses the Linux environment for non-play store installed apps. In saying that, it worked exactly as expected and I had no issues with access or limitations.
-Fred
Corporate buzzwords
We’ve reached a point in society where saying the right combination of words and letters can earn you more respect than fixing actual problems. By no means does this apply to everyone - I’m primarily talking big corporate companies. Big corporations have untold levels of hierarchy and employ probably over a billion people worldwide. I often wonder how many of these people have a shred of purpose or drive to ‘outperform Q4 forecasting’ for their big daddy corporation. This phenomenon is gradually getting more attention on social media as individuals realise that being a mindless, guardrailed drone actually sucks; but it’s a job that pays their rent and it’s too risky to do anything else.
I don’t blame them. Financial security is paramount to living the 5-9 life we want to live outside of the 9-5. However, I find it fascinating, and quite funny, that corporations have developed a new dialect of buzzwords, a new LLM if you will. Being able to say things like ‘let’s circle back’ and ‘take a 30,000 foot view’ and ‘low-hanging fruit’ on those entirely pointless weekly Teams meetings makes you more important than someone with drive to solve actual problems.
It’s pure ragebait for me, to the point I deliberately avoid it. I hate LinkedIn because every post is just another chunder of corporate chat (‘here’s what the latest Avatar taught me about supply chain’). It doesn’t do any harm, but it’s silly how a mastery of the corporate language can be more valuable than ambition.
If you’re in a business that values your ability to hide behind a mask of buzzwords instead of actual productivity, it might be time to ‘pursue alternative career pathways’.
Anyway. That’s got nothing to do with automation, it was just fun to write. I guess you can at least be assured that you won’t get any of that garbage when you deal with us. Time to get back to my desk and work on results-oriented ecosystems that move the needle.
-Fred
AI shouldn’t replace your staff
It should enable them. It should 2x, 5x or 50x their productivity. Staff, and especially good staff, are the most valuable resource in your business, and likely always will be. Stop thinking of AI as a staff member; think of it as an enabler, an augmentation of skillsets.
You want to grow your business, right? Don’t replace your staff with AI. Instead, leverage AI in your tasks and workflows to free up staff time and prevent them having to do mundane work that they hate. This way, they have time for work that directly compounds business growth. Having good, loyal staff combined with good, impactful AI automation is going to spark business growth quicker than you can send five stunningly lowball offers on Marketplace. While you’re sending those offers, your business is scaling because it has a super robust core.
I talk to a lot of business owners who are unexpectedly cutthroat, who want AI in their internal workflows by hook or by crook and they actively want it to replace staff. I’m talking businesses with less than 50 employees. It always takes me by surprise a little how business owners can be that dismissive of their staff, and fail to see any opportunities to funnel them into BD work. We don’t avoid projects that replace staff - we suggest alternatives, but at the end of the day our clients can do what they want to their businesses. But it does feel like a bad idea every time it happens. It feels small-minded and defeatist, not to mention the myriad problems that can come with moving employees on.
In saying that, I don’t think it’s a blanket rule; if you’ve got bad staff who are doing next to nothing, and there are no opportunities for growth, you can and probably should replace them with AI. If they’re doing nothing at all, you don’t even need AI to replace them. However, this generally only happens when a company has a huge headcount and the bad staff have layers of corporate to weasel behind. It’s also a much larger problem that automating a few workflows isn’t going to fix.
Stop thinking of AI as a guillotine to replace staff; think of it as a power tool to unlock staff productivity.
-Fred
Over-automation
How much automation is too much? You’d expect us to say that there’s no limit - it’ll help us get more sales, right? However, I believe that it’s possible to over-automate a business to the point that it’s hog-tied by software.
Hear me out; good automation absolutely helps a business scale, and scale dramatically. Automating mundane and repetitive tasks and workflows is going to free up your staff to focus on more business growth-adjacent activities. It means that you can make more money without having to recruit and employ. Good automation and software allows a business to rocket forward without distraction. Though, there are still some tasks and workflows that a human should be doing.
Customer service is the obvious big one. If my [looks around room for inspiration] suitcase shat the bed, and I called the suitcase company, and an AI agent answers the phone, I would honestly just give up and buy a new suitcase from another suitcase shop. People hate talking to AI agents, especially over the phone. AI just isn’t intuitive enough yet to outperform a human in customer service. Sure, replacing your reception and customer service with AI is going to save your business a huge expense. But it’ll cost your business exponentially more in client dissatisfaction, mistrust and eventually, lost sales.
(I know suitcases are a terrible example. Good thing it doesn’t matter.)
Another example is industry- or company-specific software, the stuff that electronics, machines and robots use. Take a laser cutter machine; its manufacturer will have programmed it to do X when you hit the X button. It likely comes with software that’s installed on a PC to adjust shapes and specs, run reporting, and keep it up to date with industry standards. It’s important to have somebody that knows as much as possible about this software, to maximise the machine’s output and champion any issues with it. The automating of industry or company-specific software removes the need for this expertise, so the laser cutter machine is going to be pretty stuffed if something goes wrong.
Automation is great, but it’s going to deskill your staff as they forget the workflow over time. However, if it’s a mundane and repetitive workflow, that doesn’t matter. If it’s a customer-facing workflow or includes skills that are indispensable to a role, you’re best to think twice.
-Fred
How we automate stuff
Good automation feels like magic. It’s identifying a thing that you’re doing manually, and designing a workflow to make it happen in the background so you never have to think about it again. There are heaps of different ways to do this; it depends on the complexity of the task, your software stack and the level of intuition.
If you can explain how you do a task step-by-step, it can be automated. For example, keying a sales order. The email from the client arrives, then this happens:
Identify the client
Identify the order details (number, date, lines etc)
Open your sales order software
Click to add a new sales order
Enter the client and order details from the email
Save the sales order
Attach the order pdf
Email order confirmation to the client
Notify the warehouse for picking
See how every step in the process is literally just clicking your mouse a bunch of times? If you can get this granular when automating a task, you’re going to have a super robust and reliable automation. You can set this up using middleware like Zapier or Make, or an existing integration if your software supports it. If your software doesn’t natively integrate with middleware, you can set it up using webhooks triggers and API calls. There are very few bits of software these days that don’t have API - if you’re using something that doesn’t have API, you’ve likely got bigger problems (you are probably but a humble serf harvesting grain for your lord in the year 500).
If you can explain how you do the task step-by-step, but it requires human intuition to make a calculated decision at some point in the process, it probably requires AI to be automated. An example of this is drafting email replies. If you’ve got your money up and don’t need to do business anymore, you can email however you want. The rest of us have to observe email etiquette and make sure we’re coming across correctly. This requires human intuition to write in a way that expresses the correct tone and sentiment for the situation. You can say whatever you want if you’re emailing your work mate, but not if you’re trying to convince a CEO to commit to your tender application. That’s where AI comes in - it’s not human, but it’s generally intelligent enough to detect the sentiment and intent in an email chain, look at any prior correspondence with the victim, and draft a reply to match.
Almost any task on a computer can be automated. We’re experts at deciding the best way to do it.
-Fred
Automation isn’t nuclear
It can be. It can be tiny. Automation is whatever size you want it to be. It’s like the Lego cars we built years ago; some were tiny and included only the bare essentials to operate as a car, while some were gigantic and used every brick in the box and operated not only as a car but also a rocket ship and a luxury motor home.
I talk to heaps of business owners because they are generally the decision makers. A lot of them don’t want to engage with us because automation in their heads is too large, too confronting and too much like the Lego rocket cruise liner chinook submarine. Don’t get me wrong, it can be that. You can stop business for a month and get every internal system automated. However, this isn’t realistic for most businesses. Unless you sell stuff that no one wants. Then you can realistically stop business for any amount of time. With that in mind, we don’t recommend it to anyone, even if they want to. Good automation involves starting as small as possible and building on it. Good automation means identifying a tiny, two-step task that happens constantly, and setting it up to happen in the background. Good automation means pinpointing one process/workflow that your staff HATE, and asking an automation specialist (who, us?) to engineer it.
Automation is a drug. When you automate one process, your eyes open to see other processes that you never knew could be automated. That’s why we encourage our prospects to think up one eminently basic workflow, then we set it up to happen in the background, then they identify a second workflow to automate. Eventually, our partnership balloons into that Lego bullet train lawnmower rocket ship.
Ironically, your business becomes a bullet train because your staff are freed up to work on stuff that actually makes a difference.
-Fred
‘we need to get AI in our business’
AI is everywhere today. Every piece of software you use has jumped aboard the AI train and now offers an AI service, whether that’s a chatbot assistant (most common), AI-powered navigation, writing suggestions, meeting recordings and more. AI being forced down our throats by the software we use. it’s got to the point that the majority of businesses are looking to integrate it in their business, literally for the only reason that they want to keep up with the crowd.
So is it a good idea to get AI in your business, and what’s the best way of going about it?
I’ll start with a disclaimer - I hate two things:
AI-powered services being forced on me. I’d like to pick and choose which ones I want to use. Generally, it’s too underdeveloped to hold any use for me. It’s like that one drunk dude at the party who laughs too loud and keeps telling the same story (not taking high ground, I’ve been that). They think they are adding value but really they are just shredding any credibility/aura.
People bandwaggoning to get AI in their business. It’s new and shiny and the future and everybody is getting it and what if I miss out? They don’t know what they want it to do, they just want to get it.
In saying that, it’s incredibly powerful when used correctly. If you’re using chatGPT or Gemini, DeepSeek, Grok or Claude (the list goes on) as an SME (subject matter expert), you’re using it correctly. If you’re using your AI chat to ask what to eat for lunch or to write a bunch of code, you’re using it correctly. If you’re asking it for the meaning of life or to proof-read an email, you’re using it correctly. These are all examples of how AI can make your life a hell of a lot easier. Personally, it’s most impactful for me when it’s writing code or telling me which earbuds to buy (Shokz). It regularly saves me literal days on identifying bugs in lines of code or telling me where to find a setting toggle for a function that has been rage baiting me.
In my experience with providing automation services for businesses, people want AI but they don’t know what they want it for. When they identify a workflow to automate, they’ll ask me how can they use AI to automate it. Here’s some news: AI isn’t the answer 90% of the time. AI is unpredictable, unhinged and unfiltered. It’s slow, time-consuming and extremely expensive compared to hard-coded programming or middleware.
I’ll make it make sense in a badly formatted pivot table I made (it takes too long to format):
| Thing | Human | AI | Programming |
|---|---|---|---|
| Speed | Decent | Decent | Lightning quick |
| Reliability | Good | Really good | Perfect |
| Availability | Okay | Really good | Perfect |
| Versatility | Great | Perfect | Terrible |
| Consistency | Bad | Almost perfect | Perfect |
| Takes sick days | On occasion | Never | Never |
| Can you mess with it | Of course | No | No |
| Cost | Enormous | Big | Tiny |
This post is dragging on a little, but tl/dr: don’t assume you need AI in your business. Sure, it can automate your workflows. But 90% of the time, there are better options.
-Fred