Ricardo Shiroma

Resetting the order. Writer of The Contextual Edge

[Ricardo Shiroma @r_sh8]

@Apple did a much better job on find my for air pods than for the Watch Ultra. I can’t find the Ultra because it doesn’t locate with my phone.

[Ricardo Shiroma @r_sh8]

Hey @Google -- NotebookLM was a much better name than @GeminiApp Gemini Notebook.

[Ricardo Shiroma @r_sh8]

@SmoothieKing needs a new app provider. The app is so bad.

[Ricardo Shiroma @r_sh8]

@elonmusk YOLO mode!

[Ricardo Shiroma @r_sh8]

@venturetwins @SensorTower @venturetwins , I am married, so AI it is.

[Contextual Intel @contextualint]

Serious career moves are relationship-context problems, not just application-tracking problems.

The hard part is remembering who you spoke with, what they said, who can introduce you, and what follow-up matters next.

That is what Contextual Intel is built around.

[Ricardo Shiroma @r_sh8]

Se alguém está interessado em bolao para copa, eu criei um aqui esta https://t.co/PI0QzeMalp

[Ricardo Shiroma @r_sh8]

@Lowes I was stuck at https://t.co/efUSbDbct3 . I tried my credit card, Apple Pay, and PayPal, but none completed the transaction. All the options just keeping coming back to the checkout page. I moved to @HomeDepot as I need to buy the appliance today.

[Ricardo Shiroma @r_sh8]

I am impressed how bad service is at big home improvement retail stores over 15min from waiting for help @Lowes - I will pay extra going to @AceHardware where they have staff that is knowledgeable and helpful.

[Ricardo Shiroma @r_sh8]

I interviewed Felipe Vieira about The Leadership Challenge of AI https://t.co/uebL2Aj0bm

[Ricardo Shiroma @r_sh8]

Deskilling by Default https://t.co/ZOdVmCi4sM

[Ricardo Shiroma @r_sh8]

Maya Shiroma has begun a wonderful reading adventure. Please click the link to support Maya Shiroma's reading efforts https://t.co/1T1oX0W34o

[Ricardo Shiroma @r_sh8]

AI agents aren’t just features—they’re dependency graphs. The question isn’t “can it do the task?” it’s “what happens when it’s wrong?” Ship least privilege, receipts, rollback, circuit breakers.

[Ricardo Shiroma @r_sh8]

Please help South Montgomery Softball Girls League by buying raffle tickets https://t.co/t0z2d4xI06

[Ricardo Shiroma @r_sh8]

@AnthropicAI didn’t just leak Claude Code—they leaked the coding-agent blueprint.

Not the embarrassment: the harness.
Permissions, isolation, tool execution, and MCP.

Models converge. Harnesses don’t.

Moat = trust architecture.

[Ricardo Shiroma @r_sh8]

AI can scale research. It can’t lead to discovery. @simile_ai and @cvspharmacy “agentic twins” look great for validation. But discovery is upstream: learning agenda, contradictions, weak signals, judgment. What would you never delegate to a model? https://t.co/JayURlioeI

[Ricardo Shiroma @r_sh8]

@profzeke, today in the classroom, I mentioned the 5-person team. I agreed that it will not work in every setting, but that is how the new AI-firtst start-ups are organizing themselves -- https://t.co/E4Lu38X0x3**

[Ricardo Shiroma @r_sh8]

@AzorFrank, I am taking a module this week at @Wharton. One person in my cohort looks like you. I almost called him Frank today. :-)

[Ricardo Shiroma @r_sh8]

@Topps will there be a team Brazil deck for the @WBCBaseball 2026?

[Ricardo Shiroma @r_sh8]

@Rakuten → 50% faster MTTR with @openAI Codex.

AI agents: experiment → infrastructure.

Map the bottleneck first. Slot the agent second.