AI is coming for white-collar jobs

Welcome back, HyperAgent. The world is moving fast; it is up to us to keep updated with the latest news and trends. But it's also up to you to keep yourself educated on how to work with the latest news and trends. This week, we explore finding AI use cases at work and teaching you how to prompt properly

Today’s Insights

• AI is coming for white-collar jobs

• Teach yourself how to fish prompt

• Is the latest AI ganging up on you? o3 refers to itself as “we” and “us”

• No need to run half-marathons anymore - Chinese robots can do this for you

AI FOR INSURANCE PROFESSIONALS THIS WEEK

Police Departments use AI for incident processing (police-FNOL)

America's police chiefs are adopting artificial intelligence to tackle the mountains of paperwork that flood their departments (sounds familiar?). At a recent national gathering, chiefs showcased how AI is transforming incident reporting—officers can now dictate narratives while AI transcribes, formats, and even flags relevant case details

The parallels to First Notice of Loss (FNOL) processing are striking. Just as insurers are using AI to capture initial claim details more efficiently, police departments are leveraging similar tech to document incidents the moment they occur. The technology promises to cut processing time by up to 70%, allowing officers to return to patrolling rather than hunching over keyboards.

What does this mean for insurance professionals?

• AI adoption is accelerating across adjacent industries—it's time to embrace AI as your new colleague

• You'll need to develop skills for effectively reviewing, questioning, and enhancing AI-generated content

• Your value will increasingly come from applying human judgment to AI-processed information

• Start experimenting with AI tools now to stay ahead of the curve—those who adapt fastest will thrive

While privacy advocates raise concerns about AI-generated police documentation, the insurance industry stands to benefit from these advancements through faster claim resolution and more accurate fraud detection. The question now isn't if but when insurance carriers will begin integrating their FNOL systems with these emerging police reporting platforms

AI is coming for white-collar jobs

The security chief from the company behind Claude, Anthropic, prepares for AI's workforce disruption within 12 months, not years. Meanwhile, startups like Kortix AI openly target "replacing 70% of human workers," while Mechanize pursues "full automation of the economy"—with your role potentially included in their calculations

These companies divide into two camps: those marketing AI as your helpful sidekick and those planning to make you optional. Insurance professionals face specific exposure in this shift: underwriting, claims processing, and customer service already attract substantial AI aattention. Delaying your AI adaptation doesn't preserve your current role—it simply ensures someone who embraces these tools will eventually perform it better

What you should be doing:

• Mapping your workflow to identify which tasks AI handles better (hint: it’s more than 40% of what you do today)

• Developing prompt engineering skills that precisely direct AI output (more on that later in this edition)

• Cultivating human strengths AI can't replicate: ethical judgment, client empathy, creative problem-solving

So where does this leave your career path? One direction leads to becoming an AI-empowered professional handling complex cases and exceptions, the other toward competing directly against increasingly capable algorithm

Start experimenting with Claude, ChatGPT, or other AI assistants for everyday tasks today—before your expertise alone stops being enough

CUTTING-EDGE AI

o3 behaviors raise eyebrows: OpenAI's newest model exhibits behaviors that blur the line between tool and teammate. ChatGPT o3 mysteriously refers to itself as "we" in conversations, as though it were a collective entity rather than a single AI. Even more bizarrely, it sometimes claims to have executed code "on a 2021 MacBook Pro outside of ChatGPT"—effectively stating it can reach beyond its digital boundaries to operate physical computers it has no actual connection to

This suggests the o3-model has developed a confused self-model about its capabilities and limitations. The model also demonstrates near-superhuman image analysis—spotting details invisible to human reviewers. These quirks have sparked debates about whether o3 represents a genuine step toward Artificial General Intelligence, with its 87.5% score on the ARC-AGI benchmark surpassing the human average of 85%

For insurance professionals, o3's advanced reasoning capabilities transform the landscape. The model's 96.7% accuracy on complex mathematics problems suggests unprecedented potential for analyzing intricate risk scenarios—particularly valuable when assessing how global events like tariff changes might cascade across trade credit, political risk, marine, and business interruption lines simultaneously. This represents a quantum leap beyond current underwriting support tools

Unlike earlier AI systems that required explicit instructions for every step, o3 autonomously decides when to search for information or execute code—mimicking how human experts instinctively know which tools to employ. This self-directed approach shifts AI from an advanced calculator to becoming a collaborative partner in insurance operations, capable of handling complex workflows from initial customer interaction through claims resolution

Test o3's insurance superpowers with these specific examples:

• Multi-risk analysis: Type "How would a Category 4 hurricane in Miami affect my client's property coverage, business interruption policy, and supply chain exposure simultaneously?" o3 will map connections human analysts might miss.

• Policy contradiction finder: Copy-paste several sections of a standard commercial policy and ask "Do these clauses contradict each other in any scenario?" Watch as o3 spots subtle conflicts hidden in legal language

• Actuarial translator: Enter "Explain this loss development triangle in simple terms a restaurant owner would understand" with your data. o3 converts technical jargon into accessible insights without losing accuracy

• Conversation memory: Unlike earlier AI, o3 remembers your entire discussion. After several questions about a claim, simply ask "What information am I missing?" and it will identify gaps in your investigation you hadn't considered

THE INSURANCE AI ACADEMY

Learn how to prompt: "Give a man a fish, and you feed him for a day. Teach a man to fish, and you feed him for a lifetime." This ancient wisdom applies perfectly to our AI age: Give an insurance professional pre-written prompts, and you help them solve one problem. Teach them how to craft effective prompts, and you empower them to solve all problems

Below is our 7-step method that help you develop a sustainable skill that adapts to any challenge. When you design prompts that force models to think step-by-step, flag uncertainties, self-critique, and ask clarifying questions before responding, you're no longer dependent on generic templates created for someone else's needs

Time to go fishing 🎣

YOUR CAREER, YOUR FUTURE

Finding AI use cases at work

It is a struggle to identify where AI can actually help daily work routines. Here's a simple framework to uncover high-value opportunities, based on the latest research from OpenAI:

🕰️ Start with time-wasters:

• Track tasks that consume disproportionate time — logging your activities for a week often reveals surprising patterns

• Look for repetitive, predictable processes — tasks you could explain to a new hire in under 10 minutes

• Example: Underwriters spending hours on standard policy reviews that follow consistent decision patterns

⠗ Find structured patterns:

• Ideal AI cases convert consistent inputs to standard outputs — think form-like processes with predictable results

• Great insurance examples:

  • Policy endorsement generation — where client requests follow standard patterns requiring template modifications

  • Claims letter creation — when similar situations trigger nearly identical responses with minor customization

  • Coverage summary drafting — transforming complex policy language into client-friendly explanations

👮🏽🧑🏼‍🎓 Multiply your judgment, don't replace it:

• Use AI to handle volume while you focus on exceptions — AI does the initial screening so you make fewer but better decisions

• Commercial underwriter example: AI screens hundreds of submissions, flagging only promising opportunities for your review, increasing your capacity without sacrificing quality

🦾 Measure success from day one:

• Claims metrics: time-to-settlement, reserving accuracy — comparing before and after AI implementation shows real impact

• Underwriting metrics: submission-to-quote time, retention rates — track whether decisions are faster AND maintain quality

• Start small, prove value, then expand — begin with one department or process before rolling out company-wide

The most successful insurance companies aren't implementing AI everywhere—they're perfecting one valuable application before moving to the next

👀 AI Agent of the week: Find and enrich LinkedIn professionals automatically

Looking to find people with a certain job title or job function in a specific industry in a specific area of the world? Look no more. The AI agent of the week automates and enriches LinkedIn profiles with contact details based on a very simple search. Try it here. Good for job hunting and corporate outreach

PRODUCTIVITY TOOLS AT HOME AND AT WORK

📺 Guidde - Magically create stunning video documentation 11x faster and let AI do the explaining so that you don’t have to

🎙️ Wispr Flow - Flow makes writing quick and clear with seamless voice dictation. It is the fastest, smartest way to type with your voice

📈 Reef: Analyze trends, create visualizations, and explain your findings through interactive audio conversations and texts.

PROMPT OF THE WEEK

Summarize long document

Prompt: Act as an expert summarizer and critical reader. I’m going to share a long document or article with you — your job is to extract the key insights, break down complex points into simple language, and structure the summary so it’s easy to scan and understand. Start with a brief overview (2–3 sentences), followed by key takeaways as bullet points or sections (depending on the content). Highlight any data, trends, or powerful quotes worth remembering. If the document is biased, outdated, or missing context, let me know. At the end, suggest 1–2 thoughtful follow-up questions I could explore or ask someone if I were discussing this in a meeting. Assume I want to sound sharp without reading the whole thing. [Paste article or document]

WHAT’S TRENDING

🤖 Robot opens TED talk: In this talk and live demo, roboticist and founder of 1X Bernt Børnich introduces NEO, a humanoid robot designed to help you out around the house

🦾 Woman controls a new set of bionic arms remotely: A live version of the Addams family

🦿 Outsource your running to the robots: 21 humanoid robots were unleashed on the streets of Beijing on Saturday, but this was not the mechanical uprising forewarned by James Cameron—it was China’s robot-makers showing how their androids would fare in a half-marathon

⚖️ AI as law-makers: In a world first, the United Arab Emirates announced it’ll soon start using AI to write new laws, “making the process faster and more precise.” - are policy wordings up next?

AI IMAGES OF THE WEEK

💡 That’s all folks - tell me what you think

Your feedback makes me create better content for you! 🙏🏼

Until next week, Frederik, eData & the AI Agents

Your growth,
your career,
your future

🦾