The Rundown AI / Articles / AI / Perplexity's agent pivot is on the money

Perplexity's agent pivot is on the money

PLUS: Automate your business with custom Notion Agents

Zach Mink

April 10, 2026

Good morning, AI enthusiasts. Perplexity stopped trying to simply out-Google Google and started going after Mint, TurboTax, and every personal finance app on your phone instead.

The company’s Computer launch in February now doubles as a personal finance hub and tax tool, changing Perplexity’s trajectory from search to something much bigger — with a 50% monthly revenue jump to $450M backing up every agentic move.

In today’s AI rundown:

Perplexity plugs its AI agent into bank accounts
Jassy’s $200B Amazon AI spend now has receipts
Automate your business with custom Notion Agents
Oxford AI catches heart failure five years early
4 new AI tools, community workflows, and more

LATEST DEVELOPMENTS

PERPLEXITY

🏦 Perplexity plugs its AI agent into bank accounts

Image source: Perplexity

The Rundown: Perplexity just rolled out a new Plaid integration that lets users connect bank accounts, credit cards, and loans directly to its Computer agent, turning it into a full personal finance hub.

The details:

Plaid's 12K+ bank network feeds into Computer, with users able to pull in checking, credit, loan, and brokerage data for a read-only view of their money.
The agentic system can then build customized tools like budgets, net worth trackers, debt payoff plans, and retirement dashboards via simple text prompts.
The move comes on the heels of Perplexity’s U.S tax integration that autonomously fills out IRS forms and reviews professional-prepared returns.
Perplexity Computer launched in late February, with the agentic pivot helping push Perplexity's ARR past $450M in March, a 50% jump in a single month.

Why it matters: Perplexity built its name trying to out-Google Google, but it’s Computer has completely changed the trajectory. With smart connectors and a powerful AI agent, the company is suddenly competing with Mint, TurboTax, and every other app area it ends up integrating — not just search.

TOGETHER WITH SERPAPI

🔍 Access Google Search, Maps, and more live data via API

The Rundown: Need real-time public web data without scraper headaches? SerpApi turns search results into structured, ready-to-use data with built-in scale and block resilience — powering AI apps, product research, price tracking, SEO insights, and more without the maintenance.

With SerpApi, you get:

Real-time, reliable results from any search engine
ZeroTrace Mode to keep your searches confidential
U.S. Legal Shield for scraper legal protection

Try SerpApi playground for free today — Rundown readers can get 50% off for 3 months via chat or email.

AMAZON

✍️ Jassy’s $200B Amazon AI spend now has receipts

Image source: Amazon

The Rundown: Amazon CEO Andy Jassy shared his annual shareholder letter with the company's first-ever AI revenue figures and a defense of the $200B planned capex, dismissing bubble talk and floating the idea of selling Trainium chips to outside buyers.

The details:

Amazon's $200B AI spending rattled investors this year, with Jassy's letter firing back with first-ever revenue figures and locked-in customer demand.
AWS's AI arm crossed $15B in annualized revenue, a number Amazon had never disclosed — and 260x where AWS itself stood at the same point.
The custom Trainium, Graviton, and Nitro chips crossed $20B in yearly revenue, and Amazon may sell “racks of them to third parties in the future.”
Two unnamed AWS customers asked to buy the company's entire Graviton chip supply for 2026, with Amazon declining to protect other clients' access.

Why it matters: If you only tracked models as a barometer for the AI race, Amazon might look like it's behind — but the $20B chip numbers tell a different story. Nvidia has dominated AI compute, but the supply side of the boom is finally getting real competition at exactly the moment demand has never been higher.

AI TRAINING

⚙️ Automate your business with custom Notion Agents

The Rundown: In this guide, you will learn how to build Notion Custom Agents that run your company's recurring work on a schedule, automating inbound leads, campaigns, accounting, or any recurring job in your business without adding complexity.

Step-by-step:

In Notion, create two databases — Tasks (Name, Source, Priority, Status, Assigned To) for every to-do, and Reports, so that agents can log their work
Open Notion AI, click +Create custom agent, and prompt: “Create a Weekly Planner agent that reads my last 7 days of emails every Monday morning, adds action items into the Tasks database, and writes a one-paragraph summary of the run into the Reports database”
Notion drafts the agent with a name, trigger, instructions, and data sources. Review, connect Gmail, confirm the schedule, and save
Run it on demand to test. Tighten the instructions if the output is off, then leave it running. Notion shows the next scheduled run, and the work is off your plate

Pro tip: Clone the pattern for every recurring job. Same two databases, different agent, with every run traceable in one place.

PRESENTED BY UNFRAME

📊 Enterprise AI ROI: 2026 benchmarks

The Rundown: While enterprise AI is widely deployed and already delivering measurable gains, many organizations struggle with translating those gains into business outcomes. Unframe surveyed 255 enterprise leaders to reveal current trends in ROI, adoption, and scaling of AI.

In this report, you’ll find:

4 in 5 enterprises report productivity gains
ROI drops 25% in environments with 6+ tools
Half of potential value is lost between insight and action

Get the report.

AI MEDICAL RESEARCH

🫀 Oxford AI catches heart failure five years early

Image source: Lovart / The Rundown

The Rundown: Researchers at the University of Oxford introduced an AI system that picks up invisible changes in heart fat from routine CT scans, flagging patients at high risk of heart failure up to five years out — with 86% accuracy across 72K patients.

The details:

Fat around the heart shifts texture when the muscle beneath is inflamed, with the AI reading the patterns invisible to doctors on any current scan.
In the highest-risk bucket, 1 in 4 patients ended up with heart failure within five years — a 20x gap versus those the AI flagged as safe.
Oxford is already working with regulators to bring the tool to National Health Service hospitals, and plans to extend it to all chest CT scans within months.

Why it matters: Heart failure's biggest problem isn't treatment, it's timing. Doctors usually can't act until damage has set in, so an 86%-accurate early warning system built into scans patients are already getting could shift the equation of a serious condition from reaction to prevention for better diagnosis and outcomes.

QUICK HITS

🛠️ Trending AI Tools

💼 Claude Cowork - Anthropic's desktop agent, now generally available
💻 Perplexity Computer - New agentic connections for finances, taxes
🧠 Muse Spark - Meta’s multimodal reasoning AI with multi-agent mode
🐈 Meow - Infra that lets agents open bank accounts, issue cards, and more

📰 Everything else in AI today

Spacelift Intelligence just launched, an AI infrastructure suite that helps platform teams ship infra as fast as developers code. Start for free.*

OpenAI has built a model with advanced cybersecurity skills similar to Anthropic’s Mythos, with Axios reporting the company plans to release it to a “small set of partners”.

xAI is undergoing a reorg of its engineering division, with CFO Anthony Armstrong leaving the company as SpaceX execs are installed ahead of the company’s IPO.

OpenAI launched a $100/month Pro tier with 5x more Codex usage than Plus, designed for heavy agentic coding, coming amid anger over Claude usage limits.

Florida's attorney general opened a probe into OAI with subpoenas incoming, citing allegations that ChatGPT helped plan a campus shooting at Florida State University.

*Sponsored Listing

COMMUNITY

🤝 Community AI workflows

Every newsletter, we showcase how a reader is using AI to work smarter, save time, or make life easier.

Today’s workflow comes from reader A.M in New Zealand:

"I had an interview for a role that was a leap up from where I was currently at. I loaded the job ad into ChatGPT and asked it to run me through interview questions that I voice-answered. Chat reviewed my answers, and I repeated this process until I scored highly.

I then made it to the 2nd round of interviews with the board. I researched all 5 board members and then had Chat run me through another list of questions I could expect, and then repeated the process of reviewing my answers until I scored highly. Bonus points for Chat helping me answer questions in a way that would appeal to the specific board members’ roles and interests. Fingers crossed for a response any day now!"

How do you use AI? Tell us here.