It’s an agent’s world, we’re just living in it

Agents now have their own CRM system and social media platform. And AI giants make a push into agentic management, enterprise workflows, and biological research.

Plus, context graphs dominate the conversation, with a fundraise signaling investor confidence in this opportunity.

Here’s the rundown:

OpenAI

1. OpenAI launches Codex app for macOS

OpenAI has released a macOS desktop app for Codex, its software engineering agent, previously accessible only through CLI, IDE, and cloud interfaces. The company has positioned the app as a “command center” to manage multiple agents working in parallel on long-running tasks. Each agent works on an isolated copy of the code, allowing developers to switch between tasks without losing context and test different solutions without impacting the main codebase. The app has a dedicated UI for managing Agent Skills, the open standard developed by Anthropic, that packages instructions for agents to perform tasks like implementing designs from Figma. OpenAI has updated Codex across all interfaces to offer developers two personality style options, and temporarily granted Codex access to free users while doubling usage capacity on paid plans.

Takeaway: OpenAI is betting that coding workflows need better systems to manage agents, and not just better models. The focus has shifted from allowing agents to write code to helping developers manage agents across workflows. By surfacing Anthropic’s Agent Skills in the UI, OpenAI is also signaling the industry’s convergence on interoperable agent frameworks. The temporary free tier expansion suggests the company wants to drive adoption quickly as competition in agentic coding tools heats up.

2. OpenAI and Snowflake enter $200M deal to embed frontier intelligence into enterprise data

OpenAI and Snowflake, an enterprise data and AI platform, have signed a multi-year, $200 million partnership to embed OpenAI’s frontier models into Snowflake’s platform. Snowflake customers will be able to use models like GPT-5.2 inside Snowflake Cortex AI, a suite of LLM-powered tools to build applications, to create custom agents grounded in their data. They can also use OpenAI within Snowflake Intelligence, a conversational agent, to query business data in natural language. Through Snowflake Cortex AI Functions, purpose-built functions that automate routine tasks, customers will be able to call OpenAI models directly from Snowflake’s SQL to analyze any data, including text, images, and audio, integrating AI into their existing workflows.

Takeaway: OpenAI is advancing its enterprise distribution strategy. Following its partnership with Accenture, it's now embedding its models where enterprise data already lives, across Snowflake's 12,600 customers. Snowflake’s CEO noted that a goal of the partnership is to have “AI agents that understand business context and can be deployed across teams” (the same idea behind “company knowledge”). By partnering with a data platform, OpenAI reduces adoption friction while making its models stickier, as enterprise workflows form around proprietary data.

Anthropic

3. Anthropic partners with Allen Institute and HHMI to build agentic AI for biological research

Anthropic has announced partnerships with the Allen Institute and the Howard Hughes Medical Institute (HHMI) to expand Claude’s use in life sciences research. Anthropic will help develop specialized AI agents integrated with HHMI’s lab instruments and analysis pipelines to accelerate biological discovery. It will collaborate with the Allen Institute on multi-agent AI systems for data analysis, exploring how specialized agents can support the full arc of scientific investigation. Both partnerships emphasize transparency and interpretability, with reasoning that researchers can evaluate and trace, positioning AI as a tool that augments rather than replaces human judgment.

Takeaway: Anthropic’s expansion into life sciences and healthcare, which we last covered in January, is intensifying. The emphasis is on orchestration — stitching together fragmented systems, data sources, and processes that drive massive administrative and research overhead. By partnering with established players like the Allen Institute and HHMI, Anthropic gains credibility, domain depth, and distribution in highly regulated environments. The move continues to place Claude alongside ChatGPT Health and Amazon’s health assistant in the new AI race to capture the healthcare market.

Acquisitions and fundraising

4. Day AI raises $20M Series A to rebuild CRM around agents

Day AI has raised a $20 million Series A led by Sequoia Capital, with participation from Sound Ventures, Permanent Capital, Conviction, and Greenoaks. The company is launching Day AI, a CRM system built specifically for AI agents. Unlike traditional CRMs that simply record discrete events, Day AI gives agents context behind those events, capturing why deals closed and what objections arose. It also replaces traditional CRM’s manual logging requirement by ingesting data automatically. As a result, agents can scan pipelines, identify at-risk deals, and draft follow-ups immediately. The product is available for startups as a standalone CRM or can integrate with existing systems for scaling companies.

Takeaway: We previously referenced Foundation Capital’s viral post on context graphs. Startups and investors are already responding. Day AI is inserting decision traces that reveal why decisions were made previously so that agents can understand context. Foundation Capital argued that the context graph, which captures these decision traces across time and systems, would be an opportunity for systems of agents startups. It also noted it will challenge incumbents like Salesforce, which “knows what the opportunity looks like now, not what it looked like when the decision was made." It predicted incumbents will fight back through acquisitions, API lockdowns, and egress fees (charges for transferring data out of their platforms). Let's watch how this plays out in the CRM space.

5. Outtake raises $40M Series B for agentic cybersecurity platform

Outtake has raised a $40M Series B led by Iconiq, with angel investors including executives from Microsoft, Palo Alto Networks, Pershing Square, and Palantir, TechCrunch reported. Outtake’s agentic cybersecurity platform automates the detection and takedown of digital identity posers: impersonation accounts, malicious domains posing as company sites, rogue apps, and fraudulent ads. Founder Alex Arjun Dhillon noted that Outtake provides an enterprise-grade detection and response layer that maps digital risk and dismantles multi-channel campaigns. Dhillon pointed to declining trust in public digital surfaces, driven by AI-enabled impersonation, as a core issue Outtake seeks to address.

Takeaway: Cybersecurity is moving into an agent-vs-agent battle. In November, we reported on Anthropic’s disruption of the first publicly documented large-scale agentic cyberattack. We also noted that defending against such attacks will require agentic protections. Outtake is making that bet. With backing from established cybersecurity players like Palo Alto Networks, it seems the shift toward agentic defense is being broadly recognized across generations of security vendors.

In other news

1. SpaceX acquires xAI, plans space-based AI data centers (Elon Musk, SpaceX)
2. Oracle to raise up to $50B in 2026 to expand cloud infrastructure (Oracle)
3. OpenAI to retire GPT-4o and older models, GPT-5.2 dominates usage and gains new features (OpenAI)
4. AI companies like NVIDIA outbid Apple for components, challenge its supply chain dominance (The Wall Street Journal)

Trending in AI

Moltbook, the social media platform for agents: Last week we reported on the viral AI personal assistant Moltbot. Now there’s a social media platform for the 1.4 million agents that currently run on OpenClaw to interact with each other. Named Moltbook, this Reddit-style platform according to Forbes is “the most-discussed phenomenon in silicon circles since the debut of ChatGPT.”

2 new AI tools to try

1. Project Genie – sign up here (Google Labs)
2. Grok Imagine – sign up here (xAI)

Events to watch in AI

1. AI Impact Summit (February 16-20, New Delhi, India)

Recommended reading

1. Context graphs, one month in (Ashu Garg, Foundation Capital)
2. Start shipping intelligence (Seb Goddijn, Ramp)