 The $18 Email App That Zapier Says Beats Superhuman
|
|
Lindy AI Promises a $50 AI Employee. The Credits Run Out Before Lunch.
|
|
Google's New AI Agent Never Sleeps. That's the Problem.
|
What else happened today?What AI tools should I be using?Good Morning Thorium Valley, welcome back to The Lab.
Today we're looking at a cheaper email app that claims it actually beats the one everyone's been paying $30 a month for.
We're also testing an AI assistant that promises to be your $50 employee, and finding out how fast it burns through its own budget.
And we're breaking down Google's new AI agent that never stops running, reads your inbox while you sleep, and eventually wants permission to spend your money.
Quickly before we dive in — How much would you actually pay per month for an AI assistant that handles your email and calendar?
PRODUCTIVITY
Remember Google Inbox? The email app Google killed in 2019 that people still post about like a lost loved one? The team that built it went and made their own thing. It's called Shortwave, and its Pro plan costs $18 a month. Zapier just ran a head-to-head comparison against Superhuman and concluded Shortwave is the better choice in most areas.
Superhuman has been the default "serious person's email app" for years. It costs $30 a month, has no free tier, and recently killed its free trial entirely. Shortwave has a free tier, a trial on paid plans, and its Pro plan runs $18. The AI features are where the gap gets embarrassing. Shortwave can organize your entire inbox with a single prompt, write drafts that match your writing style, search your full email history, and connect to your calendar, Slack, Notion, and other tools. Zapier's reviewer put it bluntly: "Honestly, I think Shortwave offers a superior email experience in nearly every aspect. Its AI is smarter, it's easier to learn, it's cheaper, and it keeps your inbox clean."
Shortwave's CEO Andrew Lee, who also founded Firebase before selling it to Google, isn't shy about the rivalry. He's said the company is "going right at Superhuman" and that they're about to hit feature parity on everything that matters.
The catch is Gmail. Shortwave only works with Gmail accounts. If your company runs on Outlook, it's not an option. Superhuman supports both.
On Reddit, the sentiment splits predictably. Some users who've tried both say they prefer Shortwave's UI, while others find Superhuman's auto-sorting faster. Some who just use Gmail for work praise the AI features and the ability to turn emails into tasks.
If you use Gmail and you're paying $30 a month for Superhuman, switch to Shortwave and save the difference. If you need Outlook support, stay put. If you've never paid for an email app before, try Shortwave's free tier first. Most people don't need Pro. For AI help drafting and organizing email outside of a dedicated client, Claude handles both really well at $20 a month and does a thousand other things too.
PRODUCTIVITY
Lindy AI wants to be the assistant you text instead of the app you open. Connect your email, calendar, and Slack, and it handles triage, meeting prep, and follow-ups over iMessage or SMS. Lindy's pitch is that all that time you spend manually copying data between apps could just disappear. At $49.99 a month, it sounds like a steal compared to a human assistant.
The setup is genuinely easy. Templates for meeting notes, sales call prep, and inbox sorting work almost immediately. One blogger had a discovery call agent running in minutes that automatically researched attendees on LinkedIn and emailed her a briefing ten minutes before each meeting. G2 reviewers call it simple and fun. For basic personal automation, people really do like it.
Then the credits run out.
Lindy's free plan gives you 400 credits a month but blocks premium actions, which are required for almost every useful workflow. The paid plan gets you 5,000 credits for $50, and users report burning through them shockingly fast. One reviewer who tested it for two months described the experience as "credit anxiety" that made her avoid experimenting with her own AI agents because every interaction cost something. She eventually went back to Make.com for anything complex.
This isn't just a Lindy problem. Credit-based AI pricing is becoming the norm everywhere, and the costs are genuinely unpredictable. A Ringg AI analysis put it well: "Looking at an AI pricing page should not feel like deciphering a complex mathematical sequence." Even a Microsoft executive has said that for his team, "the cost of compute is far beyond the costs of the employees." The whole industry is selling autonomous AI workers on metered pricing, and the math doesn't add up for most people.
Lindy also struggles with complex workflows. One small business owner spent an hour building a single flow and got a confession from Lindy's own Agent Builder: "I need to stop and be honest with you. We've been going in circles for a while." Zapier's review confirmed the same thing, noting Lindy has hundreds of integrations versus Zapier's 8,000 and calling it "narrow on purpose."
If your work life is mostly email, calendar, and meetings and you want a dead-simple text-based assistant, Lindy's $50 a month might be worth trying for one month. If you need anything beyond basic personal automation or you're budget-conscious, skip it. You can get Claude for $20 a month and pair it with Zapier or Make.com for complex workflows and in many cases spend less while getting more flexibility. Lindy's concept is right. The credit meter just kills it.
ASSISTANT
Google just announced Gemini Spark at I/O 2026, and it's unlike any AI product a major tech company has shipped before. It's not a chatbot you open when you need help. It's an agent that runs on its own virtual machine in Google's cloud, monitoring your inbox, drafting emails, and assembling documents while you sleep. You don't need your laptop open. You don't need your phone unlocked. It just works, all the time, across Gmail, Docs, Calendar, and Chrome.
Google VP Josh Woodward described the experience as "tossing things over your shoulder" and having Spark catch them. Early testers are using it for everything from status update emails that pull data across multiple Google apps to monitoring inboxes for customer inquiries. That part sounds great. The part that sounds less great is what comes next.
Google is building something called the Agent Payments Protocol, which will eventually let Spark make purchases on your behalf within spending limits you set. When asked how Google plans to keep the AI from making unauthorized purchases, Woodward reached for an analogy that was either reassuring or disqualifying depending on your perspective: "On the team, we think a lot of it is like if you're giving a teenager their first debit card." That's a direct quote from Google's own VP describing their own product.
The pricing lands at $100 or $200 a month depending on the Ultra tier. That puts it in line with Claude Max and ChatGPT Pro. The $100 tier gets you Spark, 5x usage limits over the Pro plan, and 20TB of storage. The $200 tier bumps limits to 20x. Both are U.S. only for now, rolling out to beta subscribers next week.
The competitive landscape is crowded. OpenAI has ChatGPT's agent running tasks through a virtual browser. Anthropic's Claude Cowork handles tasks autonomously on your desktop. Google's bet is different because it doesn't need to learn your apps. It already owns them. If you live inside Gmail and Google Docs, the integrations are native. If you don't, Spark is limited to whatever third-party connections roll out over MCP in the coming months.
The privacy concern is not theoretical. This is an AI with a persistent, always-on view of your entire email history, calendar, documents, and chat history. CNET called it giving Google "way too much access." And regulators are already watching this space closely. The FTC just settled a case against companies that marketed device surveillance as a feature, with the Bureau of Consumer Protection warning that misrepresenting data collection has consequences.
If you already live inside Google's ecosystem and you want an AI that can actually do things across your apps without you babysitting it, Spark is the most integrated option available. But you're paying $100 a month minimum to beta test a product whose own creators compare it to an impulsive teenager. If you want capable AI assistance without handing over 24/7 access to your entire digital life, Claude at $20 a month gives you strong writing, research, and document analysis without the always-on surveillance tradeoff. Wait six months and see if Google earns the trust before you hand over the keys.
EVERYTHING ELSE IN AI
+ Trump pulled his AI safety executive order hours before signing after Musk, Zuckerberg, and Sacks called him directly to kill it
+ Meta tracked employee keystrokes to train AI, then laid off 8,000 of those same employees the next day
+ Andrej Karpathy left his own startup to join Anthropic, where he'll use Claude to train the next Claude
+ Pope Leo XIV's first encyclical takes direct aim at Big Tech, calling to "disarm AI" from military and corporate interests
+ DeepSeek reportedly raising $10 billion at a $45B valuation, with the founder personally putting in 40%
+ Tim Cook steps aside as Apple CEO, handing the role to hardware chief John Ternus in September
+ George Hotz says coding agents will be "one of the most costly mistakes" in software development after six months of testing
OTHER TOOLS
+ Wispr Flow (sponsored): The new viral voice-to-text AI for iPhone and Mac that actually understands what you're saying every time
+ Superpower (sponsored): The health app that tests your blood and uses AI to tell you what's actually going on with your body before your doctor does
+ Codex: OpenAI's coding agent can now control your Mac apps, watch your screen, and run tasks on a schedule — even when your laptop is locked
+ Scion: A new open-source tool from Google Cloud that lets you run multiple AI coding agents side by side in separate containers
+ Reallusion AI Studio: A new creative tool that pairs 3D animation software with top AI video generators like Seedance 2.0 and Veo 3
+ Gemini 3.5 Flash Low: Google released a lighter version of Gemini that uses 45% fewer tokens for simple tasks
That's the Lab for this week. If a tool in here saved you time or wasted it, tell us — reply directly.
Written by Jason Chen, Advait Prakash, Andrew Hales, and the Thorium Valley crew.
That's all for today's Lab. See you next time.