PRODUCTS

OpenAI makes 3 big moves to broaden Codex's appeal

Originally designed as a software development tool, Codex is now being used by professionals across many different industries to streamline workflows and automate tasks. OpenAI wants to supercharge that process with new add-ons.

On Tuesday, OpenAI launched six Codex plug-ins that are purpose-built for roles across a wide range of industries, from sales to creative production. The aim is to expand Codex beyond coding to other areas of knowledge work and make the setup as turnkey as possible. 

Like all plug-ins, the new role-specific ones bundle relevant apps, skills, instructions and workflows, and are designed to work straight out of the box. Because the plug-ins have access to the tools you use in your workflow, they provide users with powerful prebuilt assistance without requiring any coding skills. 

For instance, the data analytics plug-in helps users answer questions with data, while the product design plug-in helps turn ideas into prototypes. Additional plug-ins include creative production, sales, public equity investing, and investment banking. 

OpenAI said in its press release that more industry-specific plug-ins are coming soon, including corporate finance, marketing strategy, private equity investing, strategy consulting, and legal. It is also building an open ecosystem where partners can create and deploy their own plug-ins directly in Codex and ChatGPT. 

Beyond the plug-ins, other Codex updates included: 

• Sites: Now available in preview for business and enterprise customers, Codex can create interactive, hosted websites and apps that users can then share. OpenAI shares its work with early partners such as Wix, Base44, Replit, Lovable, Figma, Webflow, and Emergent to build a site partner ecosystem. 

• Annotations: The ability to point to something and ask Codex to refine it now extends to any user-created content, including documents, spreadsheets, and slides.

Tools like OpenAI's Codex and Anthropic's Claude Code can be intimidating for everyday users because they were launched and marketed specifically as coding tools, and that's exactly the audience they were designed for. Expanding beyond developers is a smart move on OpenAI's part, and the numbers suggest it's already working. In the release, OpenAI shared that non-developers now make up about 20% of Codex users and are growing at more than three times the rate of developers. Though Anthropic has attempted to bridge the same gap with Claude Cowork, OpenAI’s market success with a less technical audience may give it an edge.

IN PARTNERSHIP WITH LAMBDA

Cut your AI training costs by 25% or more

Most large-scale AI training runs use less than half the computing power they're paying for. Lambda's team found the root causes and built a reproducible framework that boosted efficiency by over 25%, without changing the model itself.

Lambda’s whitepaper shows you how to address:

• Memory inefficiencies silently inflating your costs

• Training configurations that aren't making full use of your hardware

• Bottlenecks that slow down GPU communication

Get the guide.

ENTERPRISE

Microsoft seizes AI opportunity in claws and safety

With agents becoming the center of the AI narrative, Microsoft wants to be a bigger part of the conversation. 

At Microsoft’s annual Build conference in San Francisco on Tuesday, CEO Satya Nadella announced a deluge of new AI products. Unsurprisingly, agents took center stage, along with the models to support them and the defense products to secure them. Taken together, these offerings seek to fill needs at several different layers of the AI stack while bringing agents to every layer of the enterprise.  

Along with revealing its Copilot "superapp" that combines coding, chat, and the Copilot Cowork agent to bring "coding to all knowledge work," Nadella revealed what he calls "autopilots," or long-running, enterprise-grade "claws" that can have connectors, context, and memory — as well as a personality and name. 

The first of these agents is Microsoft Scout, a new personal agent that’s built on OpenClaw and can proactively handle tasks like meeting prep, scheduling and routine tasks. However, Nadella said, "In the coming months, we will build this out into a complete digital team of autopilots." 

Here’s what else is new: 

• On the agent side, the company made Microsoft IQ, its enterprise knowledge and intelligence system for agents, generally available. It also unveiled WebIQ, a platform that gives agents access to real-time web search. 

• On the model side, Microsoft introduced MAI-Thinking-1, its first reasoning model, with 35 billion parameters and a 128K context window built to lower token costs, something that’s vital as agents drive up enterprise AI bills. Additionally, the company released MAI-Image-2.5, its latest image model; new voice models with MAI-Transcribe-1.5 and MAI-Voice-2; and MAI-Code-1, its "ultra-efficient" coding model tuned for GitHub. 

• On the defense side, the company unveiled ASSERT, an open-source tool to automate AI safety evaluations; Agent Control Specification, an open-source standard to regulate agent controls; and Codename MDASH, an agentic bug-hunting system. Additionally, Microsoft revealed Frontier Tuning, which applies reinforcement learning within your compliance policies to let agents learn your organization. 

"We need all of these pieces of the puzzle and many more, and they have to play well together," Sarah Bird, chief product officer of responsible AI at Microsoft, told The Deep View in an interview. "We need to build an ecosystem where agents can function inside of organizations effectively and be safe and secure."

Microsoft is playing catch-up with its agent and model offerings as it tries to meet the needs of the enterprise AI stack. None of these are particularly groundbreaking, given that frontier labs have dominated the model market and other tech giants have cornered the agent market. Still, there are two reasons Microsoft may have a leg to stand on: existing market penetration and security. Microsoft, being a legacy tech brand with 1.5 billion Office users worldwide, has already established the trust of countless enterprises. Additionally, with agents presenting a major security conundrum, the company has plenty of room to do good work in that arena. And the mix of open-source tools announced at Build offers the kind of granular control and defensive capabilities that enterprises will need to make agents safe and reliable.

TOGETHER WITH PLAID

3 ways to stay ahead of next-gen financial fraud

Generative AI is helping fraud evolve faster than traditional defenses can handle. But these same tools can be put to good use. Read MIT Technology Review and Plaid’s report to see how technology is reshaping financial defenses. 

Read the report to discover:

• AI-powered fraud tactics targeting your customers

• Why layering AI into your defense is key

• How industry collaboration is reshaping fraud prevention

Get the report

GOVERNANCE

Snowflake's new AI tools will make enterprises sing

With enterprises hungry for ways to use AI that won’t compromise their data, Snowflake is delivering the goods. 

On Tuesday at the keynote for its annual Snowflake Summit conference in San Francisco, the company announced a flurry of new AI-powered updates to its enterprise platform. The big takeaway: These updates allow enterprises to leverage AI and agents with more context, without worrying about data vulnerabilities. 

Here are the highlights: 

• Snowflake CoWork: Formerly called Snowflake Intelligence, the company has positioned the platform as a personal AI agent for knowledge workers, integrating data, context and tools into a single platform. Some of the new features include Cortex Sense, which brings together data, business definitions, and operational knowledge; User Memory, which allows users to schedule tasks and take actions with agents using tools like Gmail and Slack; and Skill Catalog, which lets users discover, share and reuse agent skills. 

• Horizon Catalog: This is Snowflake’s central hub for AI governance, context and security. Horizon Context ensures that every person, tool, and AI agent operates from the same business definitions. Meanwhile, Adaptive Compute automatically optimizes compute resources and new security capabilities, such as verified agent identities, continuous security posture management, and automatic defenses against jailbreak attempts and zero-day vulnerabilities. 

• Snowflake CoCo: The renamed coding assistant, formerly Cortex Code, can now run tasks autonomously, without requiring users to stay on their screens. Additionally, the company announced Datastream, which eliminates the need for data brokers, connectors, or data streaming infrastructure, allowing AI apps to continuously have access to fresh data. 

“Our whole mission, our whole innovation train is based on the premise that we are the platform that will help organizations make everyone in the organization be more productive, and reinvent how they're able to work through the benefits of AI — but most importantly, do so being able to sleep well at night, because of security, because of compliance, because of governance overall,” Christian Kleinerman, executive vice president of product at Snowflake, told the media at a private round table that The Deep View attended last week. 

Additionally, on Monday evening Snowflake revealed the fruits of its partnership with Anthropic, including significant momentum with Anthropic’s Claude models on Snowflake Cortex AI, the enterprise tech firm’s suite of AI products. 

“Part of why we've chosen to primarily build for businesses, build for the enterprise, partner with Snowflake, is this concept that trust is an accelerant,” Anthropic President Daniela Amodei told Snowflake CEO Sridhar Ramaswamy in the opening keynote for the Snowflake Summit on Monday evening. “Trust is something that helps you go faster. I've never had a customer meeting where the CEO said to me, ‘I would love if Claude could hallucinate more.’”

Snowflake’s announcements seek to remove two key barriers to AI adoption: Seamlessness and trust. At the end of the day, the easier an AI tool is to adopt, the more likely it is to be used. Additionally, an AI system that doesn’t have access to a company’s context is limited in its ability to do anything beyond providing general answers, with unique, enterprise-specific data as the key differentiator. Snowflake’s partnership with Anthropic certainly doesn’t hurt — and could be symbiotic for both companies, giving Snowflake's AI repertoire more clout, while giving Anthropic a boost in enterprise trustworthiness and security posture. By getting rid of connectors and streaming infrastructure, putting in place security and governance guardrails, and forging partnerships with one of AI’s hottest frontier labs, Snowflake is positioning itself as the interface that allows enterprises to leverage, and more importantly, trust, frontier AI models.

Disclosure: Nat Rubio-Licht's travel to the Snowflake Summit 2026 event was paid for by Snowflake. The Deep View's coverage is editorially independent from the companies we cover.

LINKS

• Anthropic to expand Mythos reach to 15 more countries, 150 more orgs

• Trump signs cut-down executive order addressing AI cyber threats

• Mathematicians publish declaration warning of AI’s use in the field

• OpenAI claims it has not donated to any Super PACs

• Martin Scorsese joins Black Forest Labs as advisor for AI storyboarding

• Meta Chief Security Office Guy Rosen to depart the company

• Microsoft unveils its answer to the Mac mini agent craze: Project Solara

• Hermes Desktop: The agentic platform Hermes now has a dedicated desktop app. 

• TwelveLabs Rodeo: The company’s first API product, an AI-powered video editor. 

• ChatGPT: The chatbot now has editing features for longform writing, letting users edit longer pieces in full-screen and save them to your Library.

• Dreamina Octo: The newest AI video model from Dreamina, now available in Beta.

• Anthropic: Research Engineer/Research Scientist, Audio

• Google DeepMind: Research Scientist, Gemini Vision, DeepMind

• PwC: AI Engineer/Data Scientist, AI Senior Associate

• SonyAI: Research Intern- AI Ethics

GAMES

POLL RESULTS

Do you think AI will be embedded in many small businesses by 2030?

Yes (64%)
Somewhat (29%)
No (4%)
Other (3%)