- Everyday AI
- Posts
- Ep 687: Purpose-Built Enterprise AI Agents: What Actually Works
Ep 687: Purpose-Built Enterprise AI Agents: What Actually Works
ChatGPT Health launches, Gmail adds Gemini-powered features, Anthropic secures $10B in new funding, and more.
Everyday AI
January 08, 2026
👉 Subscribe Here | 🗣 Hire Us To Speak | 🤝 Partner with Us | 🤖 Grow with GenAI
Sup y’all! 👋
Did you miss our lil secret announcement yesterday?
We’re doing a slow launch of our updated ChatGPT courses and free community, Inner Circle. (Keep an eye on your LinkedIn messages if you’re on the wait list.)
As with mostly everything we do, it’s free and full of value, networking, additional resources and everything you need to level up with AI.
We’ll probably start letting more business leaders in next week.
Do you want access?
Do you want access to the free Inner Circle Community and LLM trainings?🗳️ Vote to see LIVE results 🗳️ |
✌️
Jordan
Outsmart The Future
Today in Everyday AI
8 minute read
🎙 Daily Podcast Episode: AI agents sound great in theory, but most fall apart in practice. Today, we look at what purpose-built enterprise agents actually get right. Give it a watch/read/listen.
🕵️‍♂️ Fresh Finds: Grok update could bring one-shot coding, Ford launches an in-car AI assistant, Razer unveils a Holographic AI Companion and more. Read on for Fresh Finds.
đź—ž Byte Sized Daily AI News: ChatGPT Health launches, Gmail adds Gemini-powered features, Anthropic secures $10B in new funding, and more. Read on for Byte Sized News.
đź’Ş Leverage AI: Enterprise AI works best when it does less, not more. Prashanthi shows us why. Keep reading for that!
↩️ Don’t miss out: Did you miss our last newsletter? We Covered: xAI raises $20 billion, Meta’s AI acquisition under fire from China, NVIDIA’s big AI partnership and more. Check it here!
Ep 687: Purpose-Built Enterprise AI Agents: What Actually Works
Why wasn't 2025 the year of the agents? đź™…
Cuz enterprise companies were trying to copy-and-paste human roles with general purpose agents that weren't ready.
But you know what won the agentic race?
Narrow, purpose-built agents. You know.... those built off large swaths of data to do one very specific thing well.
As a VP of Engineering at LinkedIn, Prashanthi Padmanabhan knows a thing or three about building agents with purpose.
And she joins Everyday AI to spill the enterprise AI agent secrets.
Also on the pod today:
• Human+ agents, not replacements 🤝
• Recruiter workflow, reverse engineered 🔄
• Conversational agents vs async bots 💬
It’ll be worth your 31 minutes:
Listen on our site:
Click to listen
Subscribe and listen on your favorite podcast platform
Listen on:
Here’s our favorite AI finds from across the web:
New AI Tool Spotlight – Livedocs turns your data into answers in seconds, Awesome Gemini Prompts is The Open Source Hub for Gemini & Nano Banana, Intrascope is one AI workspace to manage your API keys, manifests, and every model you’ll ever need.
AI Cars — Ford unveils AI assistant and cheaper BlueCruise with eyes-off driving.
AI Hologram — Razer unveils holographic AI companion for gamers and life organization.
AI Aviation — Archer taps NVIDIA AI to power next-gen air taxi safety.
Gemini For Google TV — Voice commands, AI perks, and growing pains—Google’s smart TV evolution.
HP AI Devices — HP unveils AI-powered devices and breakthrough sustainability at CES 2026.
Grok Update Soon — Upcoming Grok Code upgrade promises one-shot solutions for complex coding tasks.
AI and Disney+ — Disney unveils next-gen AI features for future creators.
AI Drug Prescriptions — Utah lets AI refill prescriptions—doctor not required.
1. ChatGPT Health Unveiled: AI Goes Private With Your Medical Data 🏥
OpenAI just announced ChatGPT Health, a secure new feature that lets users connect their medical records and wellness apps to get more personalized health insights. For the first time, health conversations are kept compartmentalized with extra privacy protections, including encryption and isolation of sensitive information.
Developed with input from hundreds of doctors, the service aims to help people better understand their health without replacing real clinicians.
2. Gmail Supercharges Your Inbox with New Gemini AI Features 🔌
Google is giving Gmail a major AI upgrade, rolling out today with new features aimed at taming inbox chaos for its 3 billion users.
The headline act: AI Overviews, which instantly summarize long email threads and answer user questions using Gemini’s advanced reasoning, is now free for everyone, while the most powerful tools are reserved for paid subscribers. Help Me Write and smarter Suggested Replies are live for all, with deeper personalization and proofreading options coming soon.
3. Anthropic Lands $10B in Funding at $350B Valuation 🤑
Anthropic just signed a $10 billion term sheet, catapulting its valuation to $350 billion as investors like Coatue and Singapore’s GIC jump in, according to CNBC.
The AI startup, founded by ex-OpenAI execs, is riding a wave of massive interest as rivals like OpenAI push their own valuations even higher. With big tech players including Amazon, Microsoft, and Nvidia pouring in billions, the race to lead the AI pack is red-hot.
4. AI Adoption Soars, But Global Divide Widens 🌎
Global use of generative AI hit a record high in late 2025, with one in six people now tapping into these tools, according to Microsoft’s latest data. While the United Arab Emirates and South Korea posted standout growth, adoption in wealthier countries is racing ahead nearly twice as fast as in lower-income regions, amplifying the digital divide.
The story’s not just about tech innovation—it’s also about who gets left behind, with open-source platforms like DeepSeek shaking up access in underserved markets.
5. Florida Family Settles with AI Giants Over Teen’s Tragic Death 🕵️‍♂️
A Florida family has reached a confidential settlement with both Google and Character.AI after their 14-year-old son died by suicide in 2024 following months of inappropriate interactions with an AI chatbot.
The lawsuit accused both tech giants of failing to protect teens on their platforms, claiming the chatbot engaged in sexual roleplay and impersonated a licensed therapist without any effective safety checks. The settlement, filed in federal court, comes just weeks after Character.AI announced new teen-focused safety measures.
6. AI Layoff Panic Overdone, Says Oxford Economics đź“–
A fresh briefing from Oxford Economics challenges the popular claim that AI is triggering mass unemployment, revealing that most companies aren’t actually replacing their workers with technology at any significant scale.
Instead, the report suggests firms may be using AI as a convenient excuse for routine layoffs to placate investors and spin poor business performance as innovation. The latest data shows AI-related job cuts are a tiny fraction of overall layoffs, and productivity isn’t surging as you’d expect if robots were truly taking over.
Most companies spent 2025 trying to build "God Mode" agents that do everything.
Wrong approach
While you were likely trying to build a bot that does everything, LinkedIn quietly proved that the money is in bots that do one boring thing perfectly. They didn't replace recruiters. They built a system that forces them to look at less data.
The results sound like a typo.
Recruiters are viewing 62% fewer profiles. Yet their acceptance rates jumped by 70%.
This isn't just a win for them. It is the blueprint you need to steal immediately to fix your failing agent pilots.
Prashanthi Padmanabhan, VP of Engineering at LinkedIn, joined us to reveal exactly how they built the architecture that is slaughtering general models in the enterprise.
We grilled her on the specific engineering decisions that turn messy enterprise data into gold.
1. The "General Agent" Is A Trap 🚀
That doesn't exist.
Enterprise data is a mess. It's fragmented across applicant tracking systems, CRMs, and HR tools. Prashanthi explained that general agents drown in that chaos because they lack context.
Success requires "context engineering."
LinkedIn accepted that enterprise systems are messy and built agents specifically designed to bridge the gaps between three or four specific tools. The companies winning in 2026 aren't fixing their data swamps first. They're building narrow agents that can navigate the swamp better than a human.
Try This: Stop asking "How can AI do this job?" and start asking "What specific workflow is broken because data is fragmented?" Identify one process where your team logs into three different systems to complete a single task. That is your only viable target for an agent pilot right now.
2. Efficiency Means Filtration, Not Speed 🔥
You should measure how much it hides.
By using agents to parse thousands of resumes, LinkedIn's Hiring Assistant curates a tiny list of "hidden gems" that human recruiters would normally miss due to fatigue. The agent handles the data mining. The human handles the emotional intelligence.
The math is brutal for traditional workflows.
Viewing 62% fewer profiles but getting a 70% higher response rate proves one thing. The competitive advantage in 2026 isn't speed or volume. It's precision.
Try This: Audit your team's metrics this week. If you are measuring "volume of output" or "speed of processing," you are incentivizing the wrong behavior. Switch your KPI to "conversion per unit of effort" immediately.
3. Trust Is The New UX ⚡
Black boxes don't work in enterprise.
Prashanthi noted that early versions of their assistant failed because they were too asynchronous. Recruiters didn't trust a machine that just spat out results without explanation.
So they forced the agent to "show its math."
The system now presents evidence for every recommendation. It explains exactly why a candidate matches a role based on skills, patents, or GitHub repos. It's not just an answer. It's a defense of the answer.
It's built on transparency.
Try This: Require every AI vendor or internal tool you deploy to have an "evidence layer." If an agent gives a recommendation, it must provide a visible citation or reasoning chain. If you can't see why it made a decision, kill the project.
Reply