AI News: Claude, Walmart Agents, OpenAI Ads
Key Points
- Anthropic unveiled Claude Sonnet 4.5, a model that excels at building/editing Excel sheets, creating PowerPoint decks, and coding, but its performance hinges on clear, well‑crafted prompts.
- Walmart has rolled out a “WB” super‑agent across more than 200 AI tools, achieving a 95% autofix rate on bugs and proving that large‑scale AI agent orchestration is already viable in enterprise environments.
- OpenAI introduced ChatGPT Pulse and Sora, two new advertising‑focused products that expand where AI‑driven content can appear, a move marketers and product teams need to factor into budgeting and platform strategy.
Sections
- Anthropic Releases Powerful Claude Sonnet - The segment spotlights Anthropic's Claude Sonnet 4.5 as a versatile AI excelling in spreadsheet, presentation, and coding tasks—requiring clear prompts for optimal autonomous development—while also noting Walmart's new AI deployment.
- OpenAI's Sora Launch & AWS Agent Core - The speaker notes OpenAI’s upcoming consumer release of Sora 2 as a sign of more 2026 products, and highlights AWS’s new open‑source Agent Core MCP server that aims to make AWS the default platform for building production‑grade AI agents, pressuring rivals.
- Salesforce's Secure AI Coding Push - Salesforce is launching an AI‑powered, security‑focused low‑code co‑pilot called Agent Force Vibes to leverage its enterprise SaaS relationships and challenge rivals like Lovable and Bolt.
Full Transcript
# AI News: Claude, Walmart Agents, OpenAI Ads **Source:** [https://www.youtube.com/watch?v=DWb4SqILvvM](https://www.youtube.com/watch?v=DWb4SqILvvM) **Duration:** 00:10:36 ## Summary - Anthropic unveiled Claude Sonnet 4.5, a model that excels at building/editing Excel sheets, creating PowerPoint decks, and coding, but its performance hinges on clear, well‑crafted prompts. - Walmart has rolled out a “WB” super‑agent across more than 200 AI tools, achieving a 95% autofix rate on bugs and proving that large‑scale AI agent orchestration is already viable in enterprise environments. - OpenAI introduced ChatGPT Pulse and Sora, two new advertising‑focused products that expand where AI‑driven content can appear, a move marketers and product teams need to factor into budgeting and platform strategy. ## Sections - [00:00:00](https://www.youtube.com/watch?v=DWb4SqILvvM&t=0s) **Anthropic Releases Powerful Claude Sonnet** - The segment spotlights Anthropic's Claude Sonnet 4.5 as a versatile AI excelling in spreadsheet, presentation, and coding tasks—requiring clear prompts for optimal autonomous development—while also noting Walmart's new AI deployment. - [00:04:15](https://www.youtube.com/watch?v=DWb4SqILvvM&t=255s) **OpenAI's Sora Launch & AWS Agent Core** - The speaker notes OpenAI’s upcoming consumer release of Sora 2 as a sign of more 2026 products, and highlights AWS’s new open‑source Agent Core MCP server that aims to make AWS the default platform for building production‑grade AI agents, pressuring rivals. - [00:08:22](https://www.youtube.com/watch?v=DWb4SqILvvM&t=502s) **Salesforce's Secure AI Coding Push** - Salesforce is launching an AI‑powered, security‑focused low‑code co‑pilot called Agent Force Vibes to leverage its enterprise SaaS relationships and challenge rivals like Lovable and Bolt. ## Full Transcript
I spend dozens of hours covering the
news so you don't have to. And I'm going
to give you the news that matters to you
for AI in just 10 minutes. Okay. Number
one, Enthropic launches Claude Sonnet
4.5. Why should you care? You should
care because this is the first model
that is extremely good at both building
and editing Excel files and also at
building PowerPoint. It also has a very
strong coding DNA. This model famously
built Slack over again in 30 hours
straight coding. One task just went and
did it. Now I have seen in my testing so
far, not 30-hour run times but longer
run times than I have seen before. This
model likes to take its time, think
through problems and solve them
correctly. And it has a strong check
your work ethos. So why you should care
is it is driving us toward that
autonomous development future. But it
will only work if you set it against the
right problem space. If you architect it
within a framework that works if you
think about the task you're giving it.
In some of my testing outside the coding
environment, I found similar constraints
and opportunities. If I give it a good
prompt for PowerPoint, it's going to
produce a great tech. If I give it a
lousy prompt for PowerPoint, I'm going
to get something difficult to build and
unreadable. So, this matters for you
because it prioritizes your intent and
it prioritizes your ability to define
work. If you can define the work, this
thing is superpowered. Story number two,
Walmart has deployed the WB, which is a
terrible name, super agent across 200
plus AI tools. And you care because it
shows that AI agents are operative in
the biggest companies in the world right
now and they deliver value. They're
getting a 95% autofix rate on bugs and
they're doing it across a complex
ecosystem inside their development
environment now today deployed. And so
this is showing us and reminding us that
AI orchestration is here at scale
already. And the federated agent model
that they're using across multiple
workflows is underlining that you can
get really complex multi-workflow
footprint set up for agents that do
work. So why should you care? Because if
you haven't believed that agents are
coming, they are. And if you have
believed that agents are coming, you can
actually build them and deploy them at
scale. It's just a function of having
the right workflow, defining your intent
properly, and then orchestrating it out.
There's more detail on that coming. I'm
going to do a deep dive on that. Story
number three, OpenAI has launched
ChatGpt Pulse and Sora. And I'm putting
those stories together because they are
intentionally together for OpenAI. Open
AAI has launched two new advertising
surfaces in the last week. Why? They are
going after ads people. Why should you
care? If you're in marketing, you care
because of budgeting. If you are
building for product, you care because
you have to think about where your
product is going to appear. Not just in
chats, but now potentially as a product
offering. Yes, I am not just assuming
this will be a consumer ad platform. It
is likely given the way we use AI, this
will also become a B2B ad platform. This
is going to be one of the biggest
developments in marketing in the last 20
years and we are going to start to see
it and we saw the first product surfaces
for it launch this week. So pay
attention to Sora, pay attention to
Pulse and where they're going. They
point the future and they also, by the
way, underline an important component of
Chat GPT's strategy. They look at a
model and they decide when it's ready
and then they find a surface for it and
kick it out. I would argue, they haven't
said this, but I would argue that pulse
is its own chat GPT model. The results I
get from pulse overnight are quite high
quality and they are different
qualitatively than what I get from
standard chat GPT5.
Sora, of course, is its own model. Very
famously Sora 2, right? They had it in
Sora 1. and it wasn't quite ready. They
didn't launch a consumer product. Now
they feel they have it and they launch a
consumer product. This suggests how
OpenAI is going to approach future
launches of their models in 2026. Look
for more net new products out of that
company. Next, AWS is launching the
agent core MCP server. Why should you
care? It's really about AWS playing
catch-up. They are providing open-source
infrastructure for building production
ready AI agents with built-in runtime,
gateway integration, identity
management, memory, and all the rest of
it. The server integrates with 40 plus
different MCPaware clients including
anthropic cloud code, cursor, etc. And
it gives developers the option to build
agents that can securely call external
tools and also maintain context across
sessions. You know why you should care?
because Amazon is leaning into
open-source to preserve their cloud
revenue. And if you are on AWS, it's
probably a great deal for you. If you
are not on AWS, I don't know that it's a
reason to switch, but it's something to
keep an eye on because Microsoft and
Google may be under pressure now to lean
this hard into open-source agent
infrastructure in their own way. They've
done some, right? They have the A2A kit
from uh from Google and I think there's
the agent development kit from Google.
So, there's some work being done here.
AWS is pushing hard on becoming the
place for developers to build because
they need to preserve the cloud revenue
and that is something that implies that
Azure and Google are really gaining
ground with AI native organizations and
that has come through in recent
quarterly earnings and that's something
that is going to shift the balance of
power and ultimately where developers
and builders prefer to build because if
you're not building with AI these days
where are you all right next story
Microsoft copilot Pilot is opening the
door to other models. For the first
time, we have co-pilot working with
clawed models and we have the idea of a
multi- aent enterprise strategy inside
the co-pilot ecosystem. This is not
super surprising since Microsoft has
already done this with Azure and Azure
has been a place where you can get
multiple models in a secure sandbox. But
they are going a step further with
co-pilot. What Sachin Nadella is
realizing is that he needs people to
care about co-pilot and the AI Microsoft
productivity experience. And if that
means he can't have his own proprietary
AI models and he needs to bring in other
people's models to make that work, he
will do it because he needs to hold on
to his attention and distribution
footprint for co-pilot products. That is
the next generation of office. And if I
were Satcha, I would also be worried and
I would also be releasing now because of
the power of Sonnet 4.5, which was the
first story I shared in this little
update. Sonnet is attacking work
primitives that Microsoft has dominated
for decades. They're going after Excel.
They're going after PowerPoint. They're
going after Docs and they're doing it
well. Microsoft would rather join them,
right? like Microsoft would rather bring
them into the fold, work with them on
co-pilot and make sure they get the
distribution advantage secured for
Microsoft because since the days when we
shipped CDs on software, Microsoft has
built off this distribution advantage in
the office. Everybody runs Windows. This
is why it doesn't really matter what
Slack does. It matters what Teams does
because Teams has the distribution. So
you should care because this is the
first case where a major enterprise
platform is offering competing AI models
inside their same interface they use for
selling their own software. Right? Like
this is this is much more integrated
than we have seen when you have back-end
developer availability of multiple
models. This is no you can choose as the
end customer what model you want. We are
starting to see enough competition where
vendor lockin for AI may just not be
tenable any and that is going to matter.
Competitive pressure is going to
intensify. We are going to start to see
IT departments able to negotiate better
terms because of moves like this. Story
number six. Salesforce is shipping agent
force vibes which I don't know that I
would have called it that but here we
are. They are looking to bring natural
language coding to enterprise
environments and they're leaning on the
salesforciness of it all. They're saying
we'll have built-in security governance
and compliance controls. In many ways,
this is exactly like the Teams and Slack
conversation, except ironically, in this
case, Salesforce is the one with the
distribution advantage versus tools like
Lovable or Bolt, which don't have the
long-term SAS enterprise deals.
Salesforce does. And so, they can go to
a CTO and they can say, "Oh, you've
heard about Lovable. I love Vibe Coding,
too. Why don't you use Agent Force
Vibes?" And the platform includes the
special autonomous AI agent. It connects
to Salesforce orgs. It can talk to your
data securely. It can actually enable
you to build securely.
This is Salesforce's play to build a
vibe coding co-pilot for lack of a
better term. What they want is to give
enterprise developers a way to expand
their footprint and to work with product
managers and marketers and others in
their orgs in ways that are integrated
into their systems instead of having all
of this shadow IT. I get why they're
doing it. I get why security first is
attractive, but what you should watch
for is the actual usage and conversation
around this tool. It is easy for this to
get sold to CTO's. It is not easy to
persuade product managers, marketers, CS
leaders who are already vibe coding
using shadow it like lovable to move to
a tool like this if it's not actually
good. And that's really the trick. You
can sell the software, but you can't
make people use it. And so what I'm
going to be watching for is whether we
get an enterprisebuilt vibe coding tool
that is actually good. I think the jury
is still out. And that's the news that
mattered this week.