Appenue AI

06/05/2026

AI is not killing people.

It is killing assumed optionality.

Basic admin, basic research, basic formatting, basic reporting, basic coordination - all of that is getting cheaper, faster, and easier to automate.

The real question is what survives.

Proof of work.
Visible capability.
Strong bridges.
Clear ownership.
Trusted reputation.

For leaders, the use case is not “replace juniors with AI”.

That is lazy strategy wearing a blazer.

The better use case is to use AI to map repeatable workflows, give people scoped ownership, let them build internal tools, and turn invisible effort into visible capability.

AI adoption done properly should build people, not just reduce headcount.

28/04/2026

The Manus case is not an isolated AI story.

It sits inside a much bigger pattern.

In 2022, Nvidia warned that new US licensing rules could affect around $400M of China sales for A100/H100-class chips.

In 2025, the H20 - the “China-compliant” chip - was pulled into licensing restrictions as well, with Nvidia expecting a $5.5Bn charge.

AMD was hit, too. Its MI308 products faced a new China export licence requirement, with around $800M in related charges.

ASML saw the same logic on the machinery side: advanced lithography exports to China moved from commercial sale to licence-by-licence political discretion.

So the question for businesses is not “which AI tool is better?”

That is the easy question.

The real question is binary:
Are we in control?

Read the full article in the comments

26/04/2026

With hiring freezes almost everywhere, how do you handle it when asked by a recruiter "what do you bring to the table?"

Do you own the entire process?

Do you build your own pipeline?

Can you use tools to write the copy, tweak the campaign, can you do some technical stuff as well?

26/04/2026

𝐓𝐇𝐄 𝐍𝐄𝐖 𝐁𝐔𝐈𝐋𝐃 𝐄𝐐𝐔𝐀𝐓𝐈𝐎𝐍

The economics of building software are shifting.

GitHub previously reported that Copilot was behind an average of 46% of developers’ code, rising to 61% in Java. GPT-5.5 is now generally available in GitHub Copilot for Pro+, Business, and Enterprise users. Cursor 3 has moved further into agent-based development, with multiple agents running across repos and environments. ChatGPT has also expanded app integrations with tools like Notion, Linear, Box, and Dropbox.

For SMEs and solopreneurs, that changes the practical build equation.
Internal tools, dashboards, lead-scoring systems, client portals, MVPs, and workflow automations no longer require the same level of upfront technical capacity.

But this is not a free pass.

AI does not fix unclear thinking.
It scales it.

A broken workflow becomes a faster broken workflow.
A weak data structure becomes a nicer-looking liability.
Poor product logic remains poor product logic, just shipped sooner.

The advantage now belongs to operators who understand the business problem clearly enough to direct the tools properly.

The constraint is no longer just coding ability.

It is clarity, judgement, governance, and ex*****on discipline.

Send a message to learn more

16/04/2026

Do you trust AI with your company’s secrets?

The real issue is not only whether your data is used to train models.

It is what happens to your prompts, outputs, client details, numbers, strategy notes, and internal documents once they enter someone else’s system.

Privacy settings matter.

But they are not a governance SOP.

Before using an LLM, businesses should know:

What data is being used.
Which environment is safe.
What must be anonymised.
What should be replaced with dummy data.
Who reviews the output.
Who logs the usage.

Rule of thumb:

If the data would hurt you in a leak, do not paste it raw into a public LLM.

Governance first. Magic later.