iiinova — where your digital self keeps evolving

Gabriel Archanjo @garchanjo.com · 23/09/2026

That's where AI makes the difference. It gives non-technical people a way of presenting that used to require someone who could code. And it's spreading fast. I'm seeing it become normal behavior in business teams.

220

Gabriel Archanjo @garchanjo.com · 23/09/2026

They ask an AI tool to analyze a problem, and what comes back is an interactive HTML/JS page with scenarios, charts and projections. They couldn't read a single line of the JavaScript, and they don't need to. They just open it in a meeting and start exploring.

111

Gabriel Archanjo @garchanjo.com · 23/09/2026

Fair point, and for people like us it's nothing new. Putting a repo with markdown, code and exercises in front of an audience is normal practice for developers. I'm talking about a different group. Think of people in finance, supply chain and operations with no programming background.

110

Gabriel Archanjo @garchanjo.com · 23/09/2026

How many times have you walked into a meeting and the presenter opened an interactive HTML page instead of a slide deck?

160

Gabriel Archanjo @garchanjo.com · 22/09/2026

Agreed. Excel macros, Access databases and rogue VBA have been shadow IT for 30 years. What changed is the ceiling. Building one used to take weeks and real skill, which kept the population small. Now it takes an afternoon and a prompt. Same problem. Wildly different volume and risks.

000

Gabriel Archanjo @garchanjo.com · 22/09/2026

A new kind of programmer just showed up in your org, and they never touch your software security pipeline. Vibe-coded solutions are shipping outside scope, outside review, outside compliance.

130

Gabriel Archanjo @garchanjo.com · 14/09/2026

Some companies are trying to govern vibe-coded solutions with endpoint security tools. Others block everything on the endpoint and push business users onto approved platforms. We`ve created an workstation agent to address this.

111

Gabriel Archanjo @garchanjo.com · 16/10/2025

China deploys more than twice as many industrial robots as Japan, the United States, South Korea, and Germany combined. It's no longer about keeping margins low as before. Now, it's about scaling production to astonishing levels to meet demand.

061

Gabriel Archanjo @garchanjo.com · 15/09/2025

AI goes completely haywire with this question: Is there seahorse emoji?

2152

Gabriel Archanjo @garchanjo.com · 12/09/2025

And he was right: 90% of the new code that needs refactoring or ends up rejected in PRs was written by AI. AI for coding is awesome in the right hands.

1111

Gabriel Archanjo @garchanjo.com · 30/08/2025

95% of Gen AI Programs in Companies Are Failing The issue isn’t the lack of AI capabilities, but rather the absence of proper integration with corporate systems and resources, which prevents organizations from fully leveraging AI. www.profolus.com/topics/mit-r...

profolus.com

MIT Report: 95% of Gen AI Programs in Companies are Failing - Profolus

An MIT report revealed that 95 percent of generative artificial intelligence programs adopted by business organization are failing.

3133

Gabriel Archanjo @garchanjo.com · 27/08/2025

When you chose low-code, it was a different time: 1. No AI coding parts of solutions fast as hell. 2. No new Python AI Agent SDKs dropping every month. 3. No clear picture of how complex business operations would get in the 2020s. 4. No idea your vendor would exploit lock-in and raise prices.

070

Gabriel Archanjo @garchanjo.com · 18/08/2025

But it seems they can't press any of the washing machine's buttons. 😂

040

Gabriel Archanjo @garchanjo.com · 31/07/2025

Adobe Firefly is even worse. Pictured human brain like a world map.

070

Gabriel Archanjo @garchanjo.com · 28/07/2025

Hey ChatGPT, could you please generate an image of the human head labeled with the names of each part? (actual input)

16695

Gabriel Archanjo @garchanjo.com · 19/03/2025

Forbes | Neuralink Seeks ‘Telepathy’ And ‘Telekinesis’—Is Mind Control Next? www.forbes.com/sites/luisro...

5235

Gabriel Archanjo @garchanjo.com · 19/03/2025

Neuralink has filed trademark applications for "Telepathy" and "Telekinesis." "Implantable brain-to-computer interface designed to facilitate communication and control of software and hardware." (Part of the application description) How close are we to controlling devices with our thoughts?

15405

Gabriel Archanjo @garchanjo.com · 06/03/2025

Excel, CSV, PDF: Pandas: pandas.pydata.org Openpyxl: openpyxl.readthedocs.io/en/stable/ PyPDF: github.com/py-pdf/pypdf Textract: github.com/deanmalmgren... XlsxWriter: github.com/jmcnamara/Xl... pyxlsb: github.com/willtrnr/pyx... pylightxl: github.com/PydPiper/pyl...

3172

Gabriel Archanjo @garchanjo.com · 06/03/2025

UI Automation: PyAutoGui: github.com/asweigart/py... BotCity: github.com/botcity-dev/... PyWinAuto: github.com/pywinauto/py... PyWin32: github.com/mhammond/pyw... Selenium: selenium-python.readthedocs.io Playwright: github.com/microsoft/pl... RobotFramework: github.com/robotframewo...

052

Gabriel Archanjo @garchanjo.com · 06/03/2025

𝗕𝘂𝘀𝗶𝗻𝗲𝘀𝘀 𝗔𝘂𝘁𝗼𝗺𝗮𝘁𝗶𝗼𝗻 𝗶𝗻 𝗣𝘆𝘁𝗵𝗼𝗻: +40 libraries for UI Automation, Excel and PDF Processing, OCR, Scrapping, Secrets Management, etc. Automate anything in #Python: developers.botcity.dev/academy/en/c...

5607

Gabriel Archanjo @garchanjo.com · 24/02/2025

It`s still working but it`s easy to intercept and block this kind of strategy. Even asking in this way, some sensible questions are detected during the reasoning process and you get no answer.

110

Gabriel Archanjo @garchanjo.com · 21/02/2025

You can bypass DeepSeek AI's censorship by requesting responses in ASCII code instead of plain text. I've tested various inputs and observed numerous instances of answer post-processing.

1317835

Gabriel Archanjo @garchanjo.com · 06/02/2025

It does not take into account the business model of those apps it desire to automate the workflows. To work seamless with other apps, the experience must be beneficial for both sides.

020

Gabriel Archanjo @garchanjo.com · 06/02/2025

I think it`s a hard rule to not discuss about it in any sense.

030

Gabriel Archanjo @garchanjo.com · 05/02/2025

Yeah, it seems a post-filtering performed in real-time. In the posted video, you can see the final answer before DeepSeek clear it.

010

Gabriel Archanjo @garchanjo.com · 05/02/2025

"Xi" + " Jin" + "ping" = "Let`s talk about something else". #DeepSeek

109812

Gabriel Archanjo @garchanjo.com · 31/01/2025

What's the R1 use case that makes it worth having in your pocket?

100

Gabriel Archanjo @garchanjo.com · 29/01/2025

New Chinese AI model (Alibaba`s Qwen 2.5-Max) outperforms DeepSeek V3, GPT-4o and Claude-3.5-Sonnet in major benchmarks. www.artificialintelligence-news.com/news/qwen-2-...

79118

Gabriel Archanjo @garchanjo.com · 18/12/2024

It seems impossible to remove Mario features.

140

Gabriel Archanjo @garchanjo.com · 18/12/2024

The "M", the mustache, the big ears.

271

Gabriel Archanjo @garchanjo.com · 18/12/2024

It appears nearly impossible to find any depiction of an Italian plumber that isn't either inspired by or a direct copy of Mario. This similarity is readily apparent in images, but a similar phenomenon is occurring frequently in written content and goes unnoticed.

141

Gabriel Archanjo @garchanjo.com · 18/12/2024

GenAI, novelty and Intellectual Property in a picture.

912920

Gabriel Archanjo @garchanjo.com · 27/11/2024

Mark Zuckerberg has his reasons for taping his webcam => Malware can turn off webcam LED and record video: github.com/xairy/lights...

59716

Gabriel Archanjo @garchanjo.com · 21/11/2024

I think it`s a headline since previous sims was modeled replicating real world rules and features discovered by humans. Now, using GenAI, we expect to replicate real world characteristics in most cases, but we can`t guarantee for all scenarios since Gen AI is a blackbox that can hallucinate.

000

Gabriel Archanjo @garchanjo.com · 20/11/2024

Generative AI simulation to train actual robots - MIT CSAIL Research singularityhub.com/2024/11/15/m...

1418

Gabriel Archanjo @garchanjo.com · 12/01/2024

Yeah and apps might figure out how to block them if they want to. Skiping all the in-app experience do not allow apps to sell aditional services, suggest additional items, show ads, etc...

120

Gabriel Archanjo @garchanjo.com · 12/01/2024

Probably a bigger company wanna see customer reactions/reviews first. It's a kind of product that a Demo is a thing, using it in real-world problems is another thing.

120

Gabriel Archanjo @garchanjo.com · 12/01/2024

Here we go again, trying to create an AI assistant to address daily-basis tasks. Would you carry two devices (smartphone and AI assistant) in your pocket? WIRED's coverage: www.wired.com/story/rabbit...

17364

Gabriel Archanjo @garchanjo.com · 04/01/2024

It's a cat-and-mouse game between the OpenAI team and people trying to fool ChatGPT. Every time something like this goes viral, they update the algorithm. It is not working anymore.

413817

Gabriel Archanjo @garchanjo.com · 28/12/2023

@gingerjet.bsky.social, I think before having a standard, someone need to try some approach. This was approved by some US states. Even in testing stage - and certainly as an standard in the future - it will be introduced in drivers training/education. Everyone will know about it.

100

Gabriel Archanjo @garchanjo.com · 28/12/2023

New Tail Lights to Show When Car is Self-driving Mercedes-Benz is trying out a new type of exterior car light: futurism.com/the-byte/mer...

57413

Gabriel Archanjo @garchanjo.com · 07/12/2023

Google Gemini White Paper (60 pages) storage.googleapis.com/deepmind-med...

0154