Full autonomy sounds like the finish line for AI. @surfas0x says it should not be, and not for ethical reasons. Underneath the reasoning, a model is still matching patterns in context, not thinking, and manipulated context can push it somewhere else entirely.
Watch the second episode of The Intelligence Series through the link in our bio.
#intelligenceseries #ai #subvisual
Tech La∩ds #3 done. AI discussion, two lightning talks, good people. Thank you to everyone who came. See you at the next one.
#techlands #portugal #ai
At Subvisual, people build the tools they need. This time, it was our Brand & Marketing Specialist.
Mariana Oliveira isn't a developer. But she built SubSocial, our social media dashboard, with Claude, Claude Code, and help from one of our developers.
You can find the story of how it came together, and what broke along the way, in the link in our bio.
#ai #claudecode #socialmedia
AI provider pricing does not stay flat, and Manuel Pina builds Tally on top of that uncertainty. His fix: put a buffer in your P&L before the swing becomes a crisis.
Watch the second episode of The Intelligence Series through the link in our bio.
#intelligenceseries #artificialintelligence #aiadoption #startups
We joined Tesser at day zero. No product, no screens, just an idea. The design challenge: make dense information feel clear and navigable. New Dribbble post at the link in our bio
#productdesign #ux #fintech #web3
Tech Lands #3 is tomorrow!
Join us for an evening of lightning talks, open discussions, and networking.
Still time to RSVP 👉 luma.com/1brk459u
Tech Lands #3 is dropping by our office on October 14th
What to expect:
🎤 Lightning talks & show-and-tell: Planned or impromptu. Show us something you've been exploring in AI, tech, or anything adjacent.
🗣️ Open discussion: Tech Lands is still evolving. What events, formats, and topics would make you show up every time?
🍻 Networking: Plenty of time to meet people and let conversations go wherever they go.
RSVP 👉 luma.com/1brk459u
1/ The calendar labels in me.subvisual finally reached the pipeline. @rccsousa uses that card to show what the review stage is for and what it costs.
2/ What the reviewer found, what a similar change gets today, and the one rule that sits under all of it.
Link in bio.
#agenticworkflows #ai #softwaredev
The question that caught @jpoliveira_pt, developer at Subvisual, sideways was an exceedingly easy one: how do you add a new recorder? He couldn't walk them through it.
Link in bio for the full article.
Episode 2 of The Intelligence Series is live!
We're talking directly with the people building, deploying, and regulating AI. This time: Manuel Pina (@ManelMPina), Filipe Pereira (@surfas0x), and João Moreira (@Johnpfmoreira) on why liability cannot be delegated to software.
Watch the full conversation through the link in our bio.
#intelligenceseries #artificialintelligence #aiagents #fintech
If you use two levers to push the same obstacle, it won't move.
@alvarorrbezerra, on the catch behind every AI leverage pitch: knowing where to point it matters more than having it.
Watch the full conversation through the link in bio.
#intelligenceseries #artificialintelligence #aiadoption #subvisual #futureofwork
More important than reading is being critical about what you're reading. That's the real go/no-go factor with AI, according to Álvaro Bezerra, PM at Subvisual. It's how he sums up "A PM's Guide to AI."
Link in bio for the full article.
#productmanagement #aiproduct #criticalthinking
Thank you to everyone who joined us for DSPT's 10-Year Meetup last Wednesday!
Special thanks to Margarida Lobato Costa from Cegid and Pedro Costa from Condukt for their insightful talks.
See you at the next one? 👀
me.subvisual's calendar labels were barely visible. The fix was two CSS classes. It sat untouched for months anyway.
@rccsousa built something with three ways of saying no and a merge button no agent ever touches.
Link in bio.
#agenticworkflows #softwaredevelopment #ai
September recap is up.
Two events. Four pieces on the Content Hub. The Intelligence Series launched.
October: Tech Lands #3 on the 14th and Ep.02 of The Intelligence Series on agentic systems at scale.
Full breakdown in the carousel 👇
#subvisual #AI #web3
Just one week to go until the third edition of Tech Lands arrives at our office.
Whether you're AI-curious or an AI expert, an expat or international living in Portugal, or simply curious and care about technology and meeting interesting people, this is the place for you!
Take a look at the agenda, RSVP, and even submit a talk 👉 luma.com/1brk459u
Tech Lands #3 is dropping by our office on October 14th
What to expect:
🎤 Lightning talks & show-and-tell: Planned or impromptu. Show us something you've been exploring in AI, tech, or anything adjacent.
🗣️ Open discussion: Tech Lands is still evolving. What events, formats, and topics would make you show up every time?
🍻 Networking: Plenty of time to meet people and let conversations go wherever they go.
RSVP 👉 luma.com/1brk459u
As a human, I cannot say the agent made this decision, so it's not my fault. That's not exactly how it works.
@Johnpfmoreira talks about why @ConduktHQ keeps a human in the loop by design.
Episode one of The Intelligence Series is live (link in bio). Another conversation is coming soon.
#intelligenceseries #ai #fintech #subvisual
DSPT Braga is tomorrow!
Still time to grab your spot 👉 tinyurl.com/296nzwua
We're hosting DSPT's 10-Year Meetup in Braga next Wednesday! Join Margarida Lobato Costa from Cegid and Pedro Costa from Condukt for an evening all about Braga's AI and data community.
See you there?
📅 September 30
🕕 18:00–20:30
📍 Subvisual — Pólo de Negócios de Braga
🎫 RSVP through the link in bio.
The job's essence hasn't changed. Everything sitting underneath it has.
@alvarorrbezerra, PM at Subvisual, breaks down four things that shifted: documentation is free now, AI has to be trained like an apprentice, speed isn't the same as direction, and the human still has to know exactly where to stand.
Link in bio for the full article.
1/ A service we built with AI assistance shipped fast and tested clean until a routine onboarding question broke our confidence in it. Six months later, we went back.
2/ The second pass looked nothing like the first, on purpose. Full piece at the link in bio.
#software #engineering #ai #documentation





































