Apiro Data
Share
Apiro Data is your IoT solutions partner, we enable IoT the easy way. We make your IoT business case valid and help you connect, deploy, and monetise quick
Why AI Can't Do Your Whole Job
๐๐ฉ๐ฆ ๐๐ณ๐ถ๐ต๐ฉ ๐๐ฃ๐ฐ๐ถ๐ต ๐๐ฐ๐ณ๐ฌ๐ง๐ญ๐ฐ๐ธ ๐๐ฏ๐ต๐ฆ๐ญ๐ญ๐ช๐จ๐ฆ๐ฏ๐ค๐ฆ
Think AI is taking your whole job?
Think again.
Itโs all about Workflow Intelligence. From managing messy cleanup to handling accountability, here is why human judgment still wins.
Don't let a bad workflow break your trust.
03/06/2026
AI agents are choosing your vendor now.
I opened the Codex plugin store this week. Codex is OpenAI's coding agent, the thing developers are starting to hand build tasks to. The directory lists Gmail, Slack, Figma, Notion, GitHub.
And one communications company.
.
When a developer asks Codex to add phone verification or SMS to a new app, the agent reaches for what it knows. Twilio did not just get a listing. They gave Codex procedural knowledge of their APIs. How to use them. What to combine. What to avoid. So when the agent builds, it builds with Twilio, and the developer did not make that call.
No other communications company is in that store right now.
Distribution used to be shaped by documentation quality, developer events, brand recognition. That model assumes a developer sits at the centre of the decision. But the agent already has an opinion before the developer opens a file.
Getting inside the agent is a different kind of distribution. Twilio got there first.
If your product is not in the agent's toolkit, how much does the rest of your go-to-market actually matter?
You're teaching someone else's tool your life.
Every AI platform wants your context. Your workflows, your preferences, the way you think, the clients you work with. And in return it gets smarter about you. Which feels like progress.
Until the pricing changes. Until a better model drops somewhere else. Until you realise you cannot take any of it with you.
The simpler answer, and I know how unglamorous this sounds, is a folder. Your own file structure, locally stored, readable by anything. Claude can read it. Gemini can read it. A completely different model next year can read it. No dependency on any proprietary memory system.
The model becomes beside the point when you think about it this way. What matters is that your knowledge is organised and the model can access it. You can point any capable model at the same system and get consistent help, session after session, without starting over.
Portability is the advantage. Not the model. The model is interchangeable.
Have you thought about where your AI context actually lives?
29/05/2026
Most companies met AI with a new title.
Seventy-six per cent of large organisations now have a Chief AI Officer. A year ago it was twenty-six. That is a real shift, and on paper it looks like seriousness. A new seat at the table, and a person whose whole job is the thing everyone agrees matters most.
I keep looking at it and seeing something else.
Creating a role is the fastest possible way to look like you are doing something about a problem while you work out whether you can actually do anything about it. The title goes in before the capability does. It almost always does.
The number in the same study that we are not quoting back is this one...
Sixty-four per cent of chief executives now say they are comfortable making major strategic decisions on AI-generated input.
Comfortable.
Not "we tested it against our own judgement and it held up." Comfortable.
So one in four still has no Chief AI Officer, and two in three are already letting the thing make the call. The title was supposed to be the part that closed that gap. Mostly it just sits in it.
A new role is not the same as a new capability.
Sometimes the role is what you reach for, and the reason is plain. The capability is the hard part. Hiring someone is the part you can do fairly fast.
If your company created an AI leadership role this year, what changed in the ninety days after the announcement, and what only changed on the slide or a meeting?
Reflections from ITW 2026
I was at ITW last week and canโt shake a crucial realization.
Telecomโs conversation has fundamentally shifted. On the surface, the talk was familiar, subsea fiber, wholesale data centers, investment, partnerships.
But underneath, everyone was zeroed in on AI infrastructure, edge clusters, data sovereignty, APIs, automation, security, and truly responsible data centers.
A few key takeaways for anyone watching the space...
๐ง๐ต๐ฒ ๐ฅ๐ฒ๐ฎ๐น ๐๐ฎ๐๐๐น๐ฒ ๐๐ ๐ฆ๐ต๐ถ๐ณ๐๐ถ๐ป๐ด
The industry is moving beyond ๐ฉ๐ฐ๐ธ ๐ฅ๐ฐ ๐ธ๐ฆ ๐ด๐ฆ๐ญ๐ญ ๐ฎ๐ฐ๐ณ๐ฆ ๐ค๐ข๐ฑ๐ข๐ค๐ช๐ต๐บ to ๐ธ๐ฉ๐ฐ ๐ค๐ฐ๐ฏ๐ต๐ณ๐ฐ๐ญ๐ด ๐ต๐ฉ๐ฆ ๐ค๐ณ๐ช๐ต๐ช๐ค๐ข๐ญ ๐ช๐ฏ๐ต๐ฆ๐ณ๐ข๐ค๐ต๐ช๐ฐ๐ฏ ๐ญ๐ข๐บ๐ฆ๐ณ ๐ฆ๐ท๐ฆ๐ณ๐บ๐ต๐ฉ๐ช๐ฏ๐จ ๐ฆ๐ญ๐ด๐ฆ ๐ฏ๐ฐ๐ธ ๐ฅ๐ฆ๐ฑ๐ฆ๐ฏ๐ฅ๐ด ๐ฐ๐ฏ.
๐๐ผ๐บ๐บ๐๐ป๐ถ๐ฐ๐ฎ๐๐ถ๐ผ๐ป ๐ฃ๐น๐ฎ๐๐ณ๐ผ๐ฟ๐บ๐ ๐ฆ๐๐ถ๐น๐น ๐ ๐ฎ๐๐๐ฒ๐ฟ.
As omnichannel journeys get ever more complex (imagine: WhatsApp โก voice โก AI agent โก CRM โก human, all in one conversation), these platforms are now the glue for customer experience, not just messaging pipes.
๐ฆ๐๐ฟ๐ฎ๐๐ฒ๐ด๐, ๐ก๐ผ๐ ๐๐๐๐ ๐๐ผ๐ป๐ป๐ฒ๐ฐ๐๐ถ๐๐ถ๐๐.
The network is now a foundational piece of the strategic puzzle and some players are moving fast to redefine their roles, while others are still selling capacity and calling it a plan.
Big year of change. Curious to see who adapts and who gets... to play catch up.
26/05/2026
AI didn't make those people redundant.
This month alone, Cisco cut nearly 4,000 roles while talking about AI infrastructure. Coinbase cut around 14 percent of its staff as Brian Armstrong talked about rebuilding the company to be lean, fast, and AI-native. Cloudflare cut about 1,100 people and framed it around the agentic AI era. Intuit is cutting more than 3,000 roles while refocusing around AI.
You can see why that language travels. "We are restructuring for an AI future" sounds like vision. "We over-hired, margins are under pressure, and the market turned" sounds like a mistake.
Same redundancies. Very different press release.
Then Sam Altman, who has every reason to make AI sound powerful, said what most of us are thinking. Some companies are AI washing their layoffs. Blaming the technology for cuts they would have made anyway.
That matters because the story does not stop with the people who leave.
It lands on the people still in their seats. They watch colleagues go, hear it was because of AI, and start believing the only way to stay safe is to pretend tools they barely trust are doing work they are not.
That is a strange kind of damage. It turns job loss into fear dressed up as transformation.
Yes, AI is changing work. Some roles will go, some will change, and pretending otherwise is its own kind of nonsense.
But when every difficult management decision gets wrapped in AI language, people stop knowing what is real. They do not know whether they are being asked to learn, adapt, or quietly prepare for the same excuse to land on them next.
When your company announces its next AI-driven cuts, an honest question arisesโฆ do you believe the AI is responsible, or has it simply become the most convenient scapegoat?
I'm away this week attending and has gone rogue.
She changed her outfits, picked the topics, and ran the whole show with Max.
They get into Greg Brockman's thirty billion dollar stake, and Musk's slightly hypocritical objection to it given he once tried to merge OpenAI into Tesla. They cover the leaked Murati-Altman texts, where Mira typed "directionally very bad" while the board was trying to fire Sam, who calmly replied by suggesting Microsoft just buy OpenAI to solve the governance problem.
There's Anthropic's surprise compute deal with SpaceX, signed not long after Musk called them "misanthropic" earlier this year.
OpenAI quietly added pet Tamagotchis to Codex right after announcing they had eliminated side quests. And GPT-5.5 Instant became the new default, with a voice mode that finally seems to know when to stop talking.
It is, as Mira Murati might text it, directionally very bad. In the best way.
NassAI signs off by threatening to text her landlord "directionally very bad" when the rent's due. I am stealing the line.
A lot of us have built our careers on being the reliable doer.
But as AI becomes more prevalent, we're learning that being valuable isn't about doing the task quicker.
We need to be discerning what's real from the polished AI facades.
Understanding the true value, spotting potential process pitfalls, and knowing when humans are still crucial will keep you in the game.
14/05/2026
Coinbase used the phrase ๐ฐ๐ฏ๐ฆ-๐ฑ๐ฆ๐ณ๐ด๐ฐ๐ฏ ๐ต๐ฆ๐ข๐ฎ๐ด last Thursday.
The memo announced a fourteen per cent cut, around seven hundred people. Brian Armstrong called the new operating model ๐๐-๐ฏ๐ข๐ต๐ช๐ท๐ฆ ๐ฑ๐ฐ๐ฅ๐ด running fleets of agents. The phrase that travelled was the small one.
One-person teams.
A team of one, with software running underneath.
Read what is being said there. The middle layer of any organisation, the part that holds context and translates priorities and smooths handoffs and slows people down enough to do work that does not break, is being recast as a coordination tax that AI removes.
Calling a team of one a team is generous. It is a single person managing a stack of agents and reporting upward.
There is a version of this that is honest. Some teams genuinely shrink and the work still gets done well. The dishonest version sits next to it. The middle is what catches the things that go wrong before they reach the customer, and once you cut it the failures arrive at the top louder, only later.
The cuts are happening. That part is settled.
The question worth asking now is what gets called a team in twelve months. Who is on it. Who used to be on it and quietly came back when the agents could not hold the context.
If you are running a scaling company right now, what does your org chart look like at the layer below you, and is anyone there because of what they coordinate rather than what they ship?
Once a week I'd open about twenty tabs and tell myself I was doing research.
I'd cycle through X, LinkedIn, half a dozen AI newsletters, the FT, Reddit threads I'd saved through the week, and Google News on a few search terms I'd been tracking. Then I'd start the slow scroll, trying to work out what was actually interesting, what was relevant to my audience, and what was just the same story being recycled for the third time since Tuesday.
By the time I'd built a shortlist for the newsletter and the week's social posts, I'd burnt through most of an afternoon and forgotten what I'd seen first.
So as soon as agents became possible I handed the whole job off, first with n8n and then with Claude Cowork once it landed. I described the work in plain language, asking it to scan all those sources once a week, pull the most interesting and relevant stories on AI and the work I cover, rank them by what's freshest and most likely to land with my audience, and drop the lot in a Notion doc.
Now a ranked list waits for me in Notion every week. I open it, pick what I'm writing about, and start writing. The hunt is gone.
Then OpenAI quietly shipped Workspace Agents. Everyone in my feed was busy posting about GPT-5.5 and almost no one mentioned the agents, which felt like the bigger shift.
So I tried it. I gave it the same workflow in plane language and it was really good.
Personaly, I'm sticking with my Claude Cowork setup for now, but if you're already in the OpenAI ecosystem, Workspace Agents are worth a serious look.
What changed wasn't the model. The agent moved the workflow out of my afternoon and into a place where it could just run.
The teams pulling ahead this year are the ones quietly handing repetitive work to agents while everyone else is still drafting better prompts.
and Max get into all of this in the latest AI in Motion - Mostly Working. Check the video out ๐.
What's the one workflow in your week you'd hand off today, if you trusted you could see every step it took?
Click here to claim your Sponsored Listing.
Category
Telephone
Website
Address
Opening Hours
| Monday | 9am - 6pm |
| Tuesday | 9am - 6pm |
| Wednesday | 9am - 5pm |
| Thursday | 9am - 5pm |
| Friday | 9am - 5pm |