TeqTalk With Jas

The voice of AI, Data and Innovation.

[TeqTalk With Jas @TeqTalkWithJas]

@nikitabier The real skill in 2026 isn’t writing code.

It’s knowing when the AI is confidently wrong and having the courage to delete it.

[TeqTalk With Jas @TeqTalkWithJas]

@TechCrunch If the model is open, cheap and increasingly interchangeable, where does the moat actually sit?

It moves to the data, workflows, infrastructure and distribution around the model.

[TeqTalk With Jas @TeqTalkWithJas]

@dhh Said thank you today.

It still made the same small mistake it always does.

But at least the interaction felt a little less one-sided.

[TeqTalk With Jas @TeqTalkWithJas]

@salesforce 185K users is impressive.

Once AI can coordinate context, agents and workflows behind a single request, the human role changes too.

Less managing steps. More setting intent, reviewing exceptions and making judgment calls.

[TeqTalk With Jas @TeqTalkWithJas]

@bot @bot can you handle the call, summarize it, update the CRM, send the follow-up, and only involve a human when something actually needs judgment?

[TeqTalk With Jas @TeqTalkWithJas]

@HealthcareAIGuy Great!

Safety, HIPAA-level compliance, and secure agent-to-system access are still the biggest bottlenecks.

Once enterprises solve these, healthcare AI spend will accelerate fast.

[Jas Kaur @Jbhamra05]

A Glimpse into #EastBayCXO | Oct 2026!

Our Oct meetup “The AI-Native SaaS Reset: Rewriting Product, Eng, and Software Economics” — bringing together CXOs to discuss AI is reshaping the future of enterprise software.

A big Thank You to everyone who supported. @TeqTalkWithJas

[TeqTalk With Jas @TeqTalkWithJas]

AI Agents can connect the dots faster than humans.

But that also means they can access data they shouldn’t.

In healthcare, access to one wrong field can put HIPAA compliance at risk.

How much access is too much?

#AIAgents #HealthcareAI #DataSecurity

[TeqTalk With Jas @TeqTalkWithJas]

@ChrisGPT Great to hear, but here is the thing, what happens when the pace of discovery starts moving faster than our ability to validate, reproduce and apply it responsibly.

That may become the next big bottleneck. What you think?

[TeqTalk With Jas @TeqTalkWithJas]

@HarryStebbings This feels similar to what happened with cloud infrastructure.

Compute matters, but the ecosystem around it is what creates stickiness.

[TeqTalk With Jas @TeqTalkWithJas]

@AnthropicAI AI trust may increasingly depend on two things:
Can the system prove what actually ran?

And can others inspect the mechanism doing the proving?

Open-sourcing FlashProve makes the second question just as interesting as the first.

[TeqTalk With Jas @TeqTalkWithJas]

@StevenBartlett @simonsinek “Nobody knows” may be the most honest answer here. AI’s future is still uncertain.

As AI impact grows, the transparency around the risks, assumptions and trade-offs matters just as much as the technology itself.

This discussion is significant.

[TeqTalk With Jas @TeqTalkWithJas]

@ChatGPT Finally, an AI that can judge spending habits in real time!
The answers might hurt though.

[TeqTalk With Jas @TeqTalkWithJas]

@OpenAI Using AGI in everyday life? Interesting! Curious how this holds up when the task involves incomplete information or conflicting constraints which is most of real life.

[TeqTalk With Jas @TeqTalkWithJas]

2,301 vulnerabilities across 392 open-source projects, found by @AnthropicAI under five months.

Your agentic AI stack runs on this code too. Discovery speed up. Remediation didn’t.

New TeqTalk episode soon. Follow for the breakdown.

@Jbhamra05

#AIsecurity #AgenticAI

[TeqTalk With Jas @TeqTalkWithJas]

Healthcare is one of the biggest AI battlegrounds: claims automation, ambient scribes, population health, access.

Whatever the midterm results, the pressure for real ROI + guardrails isn’t going away.

What should health systems demand from AI vendors in 2027?

#Midterms2026 #ai

[TeqTalk With Jas @TeqTalkWithJas]

Midterms tomorrow. AI regulation is on the table.

For enterprises already deploying agents:
What changes if federal rules tighten on agent autonomy, data use, or kill switches?

What’s already in your architecture that would need to adapt?

#Midterms2026 #EnterpriseAI #AIAgents

[TeqTalk With Jas @TeqTalkWithJas]

@DeepSeekHarness The harder problem becomes orchestration: what should run, what context should be recalled, when should another agent take over, and when does a human need to step in?

That feels like the next real systems problem.

[TeqTalk With Jas @TeqTalkWithJas]

@a16z AI is already filtering options and comparing products before customers visit a store/website.

That removes the need to browse multiple sites.

The customer journey from discovery to confirmation is shifting and retailers need to rethink how they win that final decision.

[TeqTalk With Jas @TeqTalkWithJas]

@BoringBiz_ The biggest risk here is - Compute.

Infrastructure is becoming the bottleneck when it comes to scaling AI.

This is the conversation more enterprise leaders need to be having.