Make a deal with Romano Roth
🔒 Paid proposals held safely in escrow — released only when the work's approved.
Known for
3 views
"Product teams with zero software engineers." The new Team Topologies model for agentic work.
The catch: the engineers didn't vanish. They moved into the platform, the guardrails, the eval frameworks, the sovereign inference.
The real question isn't whether teams need engineers. It's whether your platform is mature enough to carry the ones you moved.
https://blog.owulveryck.info/2026/06/24/who-doe
1 views
Three words kill more AI projects than any failed model: "We do AI now."
The moment it's said, the thinking goes big. A lighthouse project. Months of scoping before anyone names the problem.
The bottleneck was never the tech. It's the organization. Start small: one real problem, live in weeks, and if it fails it hurts no one.
Episode 1 of Prompt & Proper, my new German AI podcast with Steffen Ochs
1 views
Multi-agent AI systems fail for the same reasons badly run companies do.
A new survey of ~60 agent benchmarks cites the study on why: 14 failure modes in three buckets. Unclear roles. Agents misaligned. Nobody owning verification.
Read that without the word "agent." It's every dysfunctional org, rebuilt in software.
You don't fix coordination with a smarter model. You fix it with roles, feedback l
1 views
Friday Fun 🤣
AI that does what?
#fun #humor #meme #fridayfun #fridayhumor
📊 Post engagement
🔥 Top post: "Product teams with zero software engineers." The new Team Topol · 3 likes + reposts
📊 Activity & format
Recent posts
View on Mastodon ↗
🔥 Top post
"Product teams with zero software engineers." The new Team Topologies model for agentic work.
The catch: the engineers didn't vanish. They moved into the platform, the guardrails, the eval frameworks, the sovereign inference.
The real question isn't whether teams need engineers.…
Ask your AI the same question with a typo, and it may route you to a different model.
A Google DeepMind paper: routing is judged on accuracy and cost, but that misses whether it's meaningful.
- 112 models scored less diverse than 15 hand-p…
Three words kill more AI projects than any failed model: "We do AI now."
The moment it's said, the thinking goes big. A lighthouse project. Months of scoping before anyone names the problem.
The bottleneck was never the tech. It's the orga…
Multi-agent AI systems fail for the same reasons badly run companies do.
A new survey of ~60 agent benchmarks cites the study on why: 14 failure modes in three buckets. Unclear roles. Agents misaligned. Nobody owning verification.
Read tha…
The bottleneck was never the model.
A study of 5,000 codebases: healthy code cuts AI refactoring failures 15-30%. Same model, same prompt. Only the code changed. Discipline beats intelligence, now with numbers.
Edition 7 of The CAIO: the c…
A Google DeepMind researcher just published a formal argument that AI cannot be conscious, ever. Not biology, structure. Computation is a description, not a process. A conscious mapmaker has to exist before symbols can mean anything.
AI is…
Friday Fun 🤣
AI that does what?
#fun #humor #meme #fridayfun #fridayhumor
Your test suite is green. Your test data is fiction. Guess which one production believes.
A suite built on stale or invented data reports green while real edge cases wait in production. Test data is a feedback loop, and a broken one lies t…
Two-thirds of executives say they use AI. Actual usage: 1.5 hours a week, and 90% of firms report zero impact.
Same year, same tools, Trail of Bits went from 5% buy-in to auditors finding 200 bugs a week.
The difference was never the model…
For three years, "the great flattening" meant middle managers cut while the top stayed put.
The 2026 LeadDev report (600 leaders): where roles were cut, upper management went from 13% to 23% in two years. The wave is climbing.
My read: thi…
By 2029, the AI bill per engineer could hit $596k, 230% of their salary. That's the bull case from Tomasz Tunguz.
Anthropic already spends 2.3x payroll on compute. The median software firm spends $137 per engineer a year. A 680x gap.
The s…
▶
Everyone is arguing about whether AI is a bubble. Wrong question.
Both camps fight about demand, which is loud and can be faked. The 200-year signal is supply: real capital committed upfront. Railways, fiber, shale, all real tech, all bubb…
🐘 Community & instance
💡 Facts
🕵️ Fake follower check
Estimated- Est. 89% real, active audience · Low fake-follower risk.
- Engagement (~1.2% of followers engage each post) is around typical for Mastodon.
- Established account (3+ years old).
Heuristic estimate from engagement, follower ratios, account age & growth — a screening signal, not a guarantee.
About
📸 Gallery
🔀 Audience overlap
EstimatedEstimated shared audience with similar creators — useful for avoiding overlap (or doubling down) when planning a campaign.
More like this
Find more →✉ Message Romano Roth
Reaching out to influencers is a Pro feature. Upgrade to message any influencer directly — perfect for brands and agencies booking sponsorships.
- ✓ Message any influencer from their listing
- ✓ The influencer gets notified by email
- ✓ Manage every conversation in one inbox
Already Pro? Log in.
🎤 Event / appearance with Romano Roth
Booking an event / appearance is a Pro feature. Upgrade to book Romano Roth for an in-person or virtual appearance — payment held safely in escrow until the event is done.
- ✓ Book them for events, livestreams, panels & more
- ✓ Romano Roth gets notified by email
- ✓ Fee held in escrow, released after the appearance
Already Pro? Log in.
You're out of free requests this month
Free accounts get 5 per month. Go Pro for unlimited sponsor pitches, collab requests & sponsorship deals — plus featured placement, the Verified badge, free withdrawals and more.
Upgrade to Pro — $9.95/mo →Your free limit resets on the 1st of next month.
Auto-generated from Mastodon's public data — no affiliation with or endorsement by SocialDB. Is this you? Claim it · Remove