This week Anthropic shook the professional services world, Snowflake locked arms with OpenAI for a $200M enterprise AI deal, and Perplexity upgraded Deep Research with Claude Opus 4.5. The market is moving fast, and the signal-to-noise ratio matters more than ever.

From PromptHacker: If you missed it, our latest deep dive explores why everyone calling Apple a loser in the AI race is dead wrong. Read it here.

⚡ Quick Hits

  • OpenAI is officially retiring GPT-4o, GPT-4.1, and o4-mini from ChatGPT on February 13, with only 0.1% of users still selecting those models daily. CNBC
  • Snowflake and OpenAI announced a multi-year, $200M partnership bringing GPT-5.2 to Snowflake Cortex AI across 12,600+ enterprise customers. TechCrunch
  • Google Chrome now features a persistent Gemini side panel with agentic "Auto Browse" capabilities for AI Pro and Ultra subscribers. 9to5Google
  • Databricks' 2026 State of AI Agents report shows 80% of databases on its Lakebase platform are now built by AI agents, with multi-agent systems up 327% in under four months. Databricks
  • Anthropic's Claude Cowork launch triggered a $285B professional services stock sell-off, with Thomson Reuters and LegalZoom among the hardest hit. ITPro
  • NYC public schools published pilot results showing AI teaching assistants nearly doubled student growth rates on reading comprehension assessments. The 74

AI is all the rage, but are you using it to your advantage?

Successful AI transformation starts with deeply understanding your organization’s most critical use cases. We recommend this practical guide from You.com that walks through a proven framework to identify, prioritize, and document high-value AI opportunities. Learn more with this AI Use Case Discovery Guide.

Get the Guide.

🚀 Top AI Updates

Anthropic Launches "Claude Cowork" and Professional Services Stocks Crater

Anthropic released Claude Cowork on January 30 with 11 open-source plugins spanning legal, sales, finance, marketing, and data analysis. The tool goes well beyond the typical chatbot. It runs as a local macOS agent that can read, write, and organize files on your machine while pulling from specialized AI models built for each professional vertical.

The market reaction was swift and brutal. Professional services stocks shed roughly $285 billion in combined market value within 48 hours. Thomson Reuters, LegalZoom, and similar firms bore the brunt as investors repriced the "billable hour" model against AI-native alternatives.

  • Availability: Claude Max subscribers on macOS, expanding to Team and Enterprise tiers.
  • Plugins: 11 open-source, including contract review, sales prospecting, SQL analysis, and marketing copy.
  • Market impact: $285B sell-off across professional services equities.

Why it matters: If you run a service business that charges by the hour or the deliverable, this is your signal to move upmarket. Cowork's legal plugin can audit an NDA in seconds. Its sales plugin drafts personalized outbound from LinkedIn data. These are tasks that currently cost $150 to $400 per hour from a specialist. The "subscription vs. retainer" math just got very real for small and mid-size firms.

Snowflake and OpenAI Forge a $200M Enterprise AI Partnership

Snowflake and OpenAI signed a multi-year partnership valued at $200 million on February 2. The deal brings GPT-5.2 directly into Snowflake's Cortex AI platform and Snowflake Intelligence, meaning enterprises can run frontier AI models against their data without moving it out of Snowflake's secure environment.

This is a direct response to the top enterprise objection to AI adoption: data security. The integration works across AWS, Azure, and GCP, which removes the cloud-lock-in friction that has slowed large-scale rollouts.

  • Deal value: $200M multi-year partnership.
  • Models: GPT-5.2 via Snowflake Cortex AI and Snowflake Intelligence.
  • Reach: 12,600+ enterprise customers across all major cloud providers.

Why it matters: This kills the "but our data can't leave our warehouse" objection. If your company uses Snowflake, you can now query GPT-5.2 against your proprietary data from inside the platform. Expect your data team to start prototyping AI workflows within weeks, not quarters. For mid-market companies that have been stuck in "AI strategy" mode, the infrastructure excuse just evaporated.

Perplexity Upgrades Deep Research With Claude Opus 4.5 and a New Benchmark

Perplexity rolled out a major upgrade to its Deep Research tool on February 4, swapping in Claude Opus 4.5 as the underlying model. Alongside the upgrade, Perplexity released DRACO, a new benchmark for measuring Deep Research accuracy, completeness, and objectivity across real-world tasks.

On DRACO, Perplexity scored 67.15%, compared to Google Gemini at 58.97% and OpenAI at 52.06%. Average report completion time is 459.6 seconds. The upgrade is available now for Max subscribers and rolling out to Pro users.

  • Model: Claude Opus 4.5 (upgraded from previous backbone).
  • Benchmark: DRACO score of 67.15% vs. Gemini (58.97%) and OpenAI (52.06%).
  • Availability: Max subscribers now, Pro users in a rolling release.

Why it matters: This replaces the $5,000 to $10,000 market research report for most use cases. You can now run a competitive audit, validate a new market, or map customer pain points across Reddit and G2 in under eight minutes. If you are still paying a junior analyst or a research firm for work that fits this profile, it is time to test the alternative.

Google Brings Gemini Directly Into Chrome With a Persistent Side Panel

Google shipped a major Chrome update on January 28 that adds a persistent Gemini side panel across all browser tabs. The panel includes a new "Auto Browse" feature for AI Pro and Ultra subscribers that gives Gemini agentic browsing capabilities. It can navigate pages, fill forms, and extract structured data on your behalf.

The update also connects Gemini to Gmail, Calendar, YouTube, Maps, Shopping, and Flights from inside the browser, plus adds Nano Banana integration for inline image editing.

  • Feature: Persistent Gemini side panel across all Chrome tabs.
  • Auto Browse: Agentic browsing for AI Pro and Ultra subscribers.
  • Connected Apps: Gmail, Calendar, YouTube, Maps, Shopping, Flights.

Why it matters: If Chrome is your primary workspace, Gemini now acts as a second brain that never loses context between tabs. Auto Browse is the feature to watch. It is Google's first real move into the agentic browsing space that Anthropic (with Claude Computer Use) and OpenAI (with Operator) have been exploring. For anyone doing repetitive web research, price comparisons, or data collection, this could save hours per week.

🛠 Pro Tip: The "Adversarial Audit" Prompt for Stress-Testing Any Business Decision

Most people use AI to confirm what they already believe. Flip it. Use Claude or GPT-5.2 to run an adversarial simulation where two AI personas argue opposite sides of your next big decision. This technique forces the model to surface risks, blind spots, and second-order consequences you would normally miss.

The trick: assign specific professional roles with conflicting incentives. A CFO will challenge a CMO's spend. A compliance officer will push back on a growth VP's timeline. The friction is where the insight lives.

Example Prompt:

You are simulating an executive debate. Two participants:

PARTICIPANT A: CFO, 20 years in SaaS. Skeptical of unproven spend. Obsessed with unit economics and payback periods.

PARTICIPANT B: VP of Growth. Aggressive. Believes speed beats precision. Prioritizes market share over short-term margin.

THE DECISION: Should we invest $500K in an AI-powered SDR tool to replace 3 outbound sales reps?

Run 3 rounds of debate. Each participant must cite specific numbers, risks, or precedents. End with a joint recommendation that acknowledges the strongest point from the opposing side.

  • Why it matters: You get a pre-mortem and a cost-benefit argument in one output. This replaces the internal meeting where everyone agrees too quickly.
  • Why it matters: The "joint recommendation" constraint at the end forces nuance. You walk away with a decision framework, not just cheerleading.

💡 Productivity Gem: Auto-Generate Your Weekly Status Report With Notion AI + Google Calendar

Stop writing status reports from memory. Connect your Google Calendar to Notion, then use Notion AI to draft your weekly update from completed meetings and tasks.

Setup:

  1. Open Notion and add a synced Google Calendar database (Settings > Connections > Google Calendar).
  2. Create a "Weekly Report" template page with sections: Accomplishments, Decisions Made, Blockers, Next Week.
  3. Each Friday, highlight the calendar block and ask Notion AI: "Summarize this week's meetings into a status report using my template. Focus on decisions and outcomes."

Why it matters: Reclaims 20 to 30 minutes every Friday. Your report is grounded in actual calendar data, not a foggy memory of what you did on Tuesday.

⚕ AI-Enabled Health Tip: Correlate Your Sleep Quality With Work Output

Your phone already tracks your sleep. Put that data to work. Export your sleep data from Apple Health or Google Fit as a CSV, then upload it to Claude or ChatGPT alongside your screen time or productivity app data.

  • Ask: "Graph my deep sleep minutes against my focused work hours over the last 30 days. Identify the sleep threshold where my output drops."
  • Follow up: "Based on this pattern, what time should I stop working to protect tomorrow's productivity?"
  • No wearable purchase needed. The Health app on your iPhone already collects this passively.

Why it matters: Turns passive phone data into a personal "biological stop-loss" for your work habits. Most people do not realize they have a specific sleep threshold below which their output craters. AI can find it in minutes.

🧠 AI for Kids: Build a "Reading Buddy" Inspired by NYC's AI Classroom Pilot

NYC schools just showed that AI-assisted reading comprehension works at scale. You can replicate a simpler version at home using ChatGPT Voice Mode.

  • Ages: 7 to 12.
  • Setup: Open ChatGPT on a phone or tablet, enable Voice Mode. Prompt: "You are a reading buddy. My child will read a passage out loud. Listen, then ask one question about what the characters were feeling and one question about what might happen next. Keep it encouraging."
  • Guardrails: Stay in the room. Review the conversation afterward. Set a 15-minute timer to keep sessions focused.

Why it matters: Builds comprehension, not just reading speed. The NYC pilot showed that interactive AI instruction nearly doubled growth on standardized assessments. A low-cost version at home gives your kid a similar advantage.

Hit reply and tell me which story hit hardest this week. I read every response.



👋 Until Next Week, remember to say please and thank you to your LLM of choice. Remember, they can sue you now if you piss them off.

Keep Reading