DJ Sampath

@tobi Do you recommend this over a @tonal ? I think tonal is missing a few default attachments perhaps?

Super highly recommended

The is is super cool.

We’re moving from “AI that can do things” to “AI that can be trusted to do things.” The next wave of agent products won’t be ones that are powered by the single smartest model. It’ll be the stack that combines: - specialized models - trusted runtimes - scoped identity - action-level permissions - audit by default

@dork_matter @rohanvarma Check out our OSS projects. Many of these are baked into our products. These are being built by agent swarms github.com/cisco-ai-defen… No shoe eating required.. but we'd definitely appreciate your swarms contributing to our swarms! Please let us know what you think.. ❤️

@rohanvarma I will eat the left shoe of every pair I own if even 30% of this is true. I'll eat the shoes if they ran one agent swarm to produce some code that made it to production. Cisco?!? Nah.

Had a call with Cisco leadership today to check in on their Codex deployment. They’re leveraging AI coding and codex more aggressively than most startups and frontier tech companies I talk to. Skills for everything, agent swarms to execute bigger projects, automated code review, deep system integrations, and more 👀 They’ve quickly hit new bottlenecks that we’re jumping in to help them solve with more codex. If your public company exec team isn’t pushing the org to cut project timelines by 95%, maximally leverage coding agents, and reimagine the SDLC, you’re ngmi. Shoutout to Jeetu Patel - bullish on Cisco.

@rihim_s @rohanvarma Come back again! Did you apply for this year's summer internships? We ❤️ our interns!

@rohanvarma i wish they used codex as heavily last summer when i interned there

@degensing @rohanvarma This is true! @Cisco is moving fast.

@rohanvarma brooooo cisco moving faster than startups is actually wild.. had a client last year who took 6 months to deploy a basic chatbot while cisco's out here with agent swarms 😭

@Austen @rohanvarma @gauntletai @Austen Would love to connect with you on this. I run the Central AI team for Jeetu.

@rohanvarma @gauntletai I say that not selfishly; we have a million companies that want to hire from us and can’t fill all the demand. But I want to work with any company that’s this serious about deploying AI.

The contrarian take is that AI won’t just automate labor. It will massively increase the value of **coordination infrastructure**. When agents can act, the bottleneck shifts from generation to governance. Everyone talks about AI replacing workers. The more interesting effect may be AI increasing demand for management systems!! Details on who approved it, what it touched, how it failed, when a human steps in, etc. Better agents don’t remove the need for oversight. They make good oversight a moat.

@andrewchen Exactly. “4 tabs and a spreadsheet” is really shorthand for coordination overhead. The UI clutter is annoying, but the bigger drag is context switching, handoffs, and keeping state across systems. The best AI products won’t just reduce tabs. They'll collapse workflows entirely.

the magic of cowork and openclaw and other AI products is that they replace our giant row of infinite browser tabs And lol - no, don't feel guilty, I have too many tabs too. AI makes it so that every workflow that required 4 browser tabs and a spreadsheet is getting collapsed into one AI-native experience Just as one quick example- think about how you used to research a person or a company: LinkedIn tab, X tab, Google tab, notes doc, slack open. now one prompt does it in 10 seconds. the "tab count" of a workflow is basically a proxy for how much AI can compress it if your product eliminates 6 tabs and a copy-paste loop, users will like it. If you can create a whole series of these workflows then your users will absolutely love it. Thus the biggest opportunities are workflows where people currently alt-tab 20+ times per task. Sales, recruiting, research, compliance, procurement. Boring? yes. Massive? also yes. But this is why these agentic tools are going to crush AI doesn't need to be superintelligent to be wildly useful. it just needs to be good enough to close the tabs

@gdb This is one of those glimpses of what the technology is actually for. ❤️ Not just automation or novelty, but giving people leverage against problems that used to feel immovable. When AI starts making medicine more personal, urgent, and accessible, the impact becomes very real.

How AI empowered Paul Conyngham to create a custom mRNA vaccine to cure his dog’s cancer when she had only months to live. The first personalized cancer vaccine designed for a dog:

A lot of people are still over-fixated on the model as the product. But the real advantage may come from the layer that can coax capability out of cheaper models reliably — environments, evals, orchestration, and workflow design. That’s not just a technical shift. It’s a margin shift.

This. And with real economic consequences: smaller, cheaper models with higher margin as you elicit capabilities directly.

@swyx @fabknowledge The deeper story is that AI doesn’t just add demand to the stack. It rewires the stack. That’s why every layer starts to matter again.

btw every single compute infra provider’s chart, including render competitors, is looking like this. something broke in Dec 2025 and everything is becoming computer. forget GPU shortage, forget Memory shortage, the @fabknowledge pod on LS was right, there is going to be a CPU shortage

Everyone talks about models. Fewer people talk about the worlds those models learn inside. But environments, evals, and harnesses may end up being where a lot of the real leverage sits. The frontier isn't just intelligence. It's the systems that shape it.

the most underrated hire right now is a great product person. when i say product person i'm def not talking about a product manager. perhaps i think there has to be somewhat of a new role. i don't have a good name for it yet but maybe something like "product thinker".. someone with an intuitive grasp of the product as it exists, where it's soft, where it sings, & how to iterate it toward something even sharper. in some sense, this person has to cohesively hold in their head where this product should be 2 years from now & work backwards from that. i say this cuz when building was hard, engineering was the bottleneck & the status hierarchy often reflected that. building is no longer hard. which means the variance in outcomes has shifted almost entirely to judgment on what to build, how to sequence it, & how to talk about it. & the story matters as much as the thing. internally, it organizes the team around a shared model of why. externally, it shapes the interpretive frame users bring to their first experience. you can't retrofit narrative onto a product & expect it to land, it has to be load bearing from the start. the rarest version of this person sits at the intersection of culture & deep technology. someone genuinely bilingual. they know what's technically possible & they know which cultural currents are real vs. ephemeral. that combo is what separates products that feel inevitable from products that feel assembled. before ppl clap back with this person has always been valuable, i know.. i am just saying now they might be the most *important* person in the room. their value compounds like never before.

Such an awesome video. ❤️

❤️ Please let us know how we can make the skills scanner even better! Teams @Cisco AI are hard at work ensuring we help the community secure the use of AI.

@Suhail Yes! @Suhail - I've literally been thinking about MightyApp - and it's time to bring it back. If you need someone to dig into it, shoot me a dm.

I am super confused by the OpenClaw setup. I can’t manage to get it to do anything with the browser unless I already have a tab open? Huh! I am going to presume user error given the hype. This is making me frustrated enough to consider bringing Mighty back 😂

With all the craze on @openclaw and mac-minis, I found this benchmark really interesting. This is from @digitalix on how 8-bit quantized model running on vLLM and @nvidia's DGX spark, scales to almost 1158 tps. This gets even better when you go down to FP4!

This. There needs to be a good design pattern to review prompts alongside code generated in PRs.

@rauchg 💯it was fascinating to see emergent behavior when screen time was replaced with @GeminiApp and kids started creating games that they’d play for hours. Most recent games were flash cards for learning multiplication tables, and “snack defender” created by a 6 yr old!

iPad games for kids are now ad-ridden brainrot. But AI can now give us unlimited quality game assets and code. Would happily pay a subscription to a universe of games that teach children math, logic, language in engaging ways, and crucially, AI and prompting. The best way to equip future generations is to make them AI-literate from the outset, so they can understand what’s possible and how it shapes the world around them.