I'm building an email client. Here's what every AI email tool gets wrong

I'm building an email client. Here's what every AI email tool gets wrong

I'm building an email client. Here's what every AI email tool gets wrong

Last updated: July 28, 2025

Last updated: July 28, 2025

Hey there 👋! Louis here, founder and CEO@Upstream.

Look, we all know the email struggle is real. But here's what I learned testing these tools: they're solving email as if we work in isolation. Having built collaboration tools at Asana, I can tell you- that's not how teams actually work.

TL;DR: I tested 4 leading AI email tools so you don’t have to, and ranked them depending on your use case. Turns out, email tools often optimize for individual speed but miss one key aspect of work: collaboration.

Want to support us? 👉 Join the waitlist!

Executive summary

💡 What users think they want vs. what they actually need (a product perspective)

Surface problem: "Help me write emails faster"

Every tool I tested led with the classic feature-driven approach. Superhuman boasts users write emails "twice as fast on average." Shortwave promises "less stress and newfound sense of peace." Classic individual productivity metrics.

But having built products at Asana where we obsessed over team workflows, I recognized this pattern immediately: they're optimizing for the wrong jobs-to-be-done.

🎯 Real job-to-be-done: "Help my team solve the problems that matter (at 10x the speed)"

From my time at Algolia (B2B SaaS, distributed teams), I saw this daily. The real bottleneck isn't individual email speed- it's the meta-work of keeping everyone aligned.

Picture this: You fire off a quick response to an investor inquiry while grabbing coffee. Meanwhile, your co-founder is crafting a detailed follow-up to the same person because they didn't know you already handled it. We've all been there, right?

At Asana, we learned that individual productivity tools often create team coordination debt. Same thing here.

🔎 Detailed tool analysis: what I actually found

1. Superhuman: the speed demon

Sal the Salmon swimming in notifications
Sal the Salmon swimming in notifications
Sal the Salmon swimming in notifications

Performance metrics:

  • Speed improvement: Users write emails twice as fast on average

  • Technical foundation: OpenAI GPT for instant draft generation

  • Key feature: Instant Reply pre-drafts responses before opening emails

  • User feedback: Some drafts "come off formal or generic"

My hot take ❤️‍🔥:
This is textbook individual productivity optimization. At Algolia, we called this "local maximum thinking"- making one person incredibly efficient without considering system-wide effects.

What's genuinely impressive:

  • The UX is honestly beautiful (that keyboard-first approach feels native)

  • The AI integration doesn't feel bolted-on

  • Their "never drop the ball" follow-up automation actually works

Where the product strategy falls short:

  • Limited consideration for team workflows (classic B2C thinking applied to B2B use cases)

  • Creates what I call "productivity islands"—individual efficiency that increases coordination costs

  • The AI has no awareness of team context or shared state

Perfect for: Solo founders or teams with dedicated executive assistants handling coordination

2. Shortwave: the thoughtful communicator (better user empathy)

Having worked on search and discovery at Algolia, I appreciate Shortwave's approach to AI model selection.

Performance metrics:

  • User satisfaction: Users report "less stress and newfound sense of peace with their inbox"

  • Technical foundation: Anthropic's Claude model (smart technical choice for tone)

  • Key feature: Natural language queries like "summarize last week's customer emails"

  • Integration: Works within existing Gmail workflow

Sal the Salmon swimming in notifications
Sal the Salmon swimming in notifications
Sal the Salmon swimming in notifications

My hot take ❤️‍🔥:
This feels like a team that actually talked to users about communication quality vs. just speed. The Claude model choice shows they understand that tone and context matter more than raw generation speed.

What I loved:

  • AI drafts "often require minimal editing because they mirror the sender's voice well"

  • No workflow disruption (stays in Gmail)

  • The summarization feature is genuinely useful for catching up

The limitation: Still optimized for "your voice," not "your team's voice." It's individual empathy at scale, but not team empathy.

Perfect for: Founders who'd rather sound human than fast, especially if you're already living in Google's ecosystem

3. SaneBox: the email bouncer (solving the right micro-problem)

Think of SaneBox as that friend who's really good at saying "no" to things you don't actually want to do.

Performance metrics:

  • User outcome: "After having an unmanageable email mess for years, I'm finally at inbox zero" (actual customer quote)

  • Technical approach: AI learns your email habits and acts accordingly

  • Cross-platform: Works with any email provider

Sal the Salmon swimming in notifications
Sal the Salmon swimming in notifications

My hot take ❤️‍🔥:
This is actually smart product strategy: they identified one specific job (email triage) and nailed it. No feature creep, just filters that learn.

The magic:

  • Works with whatever email setup you already have

  • That "executive summary" of filtered emails is clutch

  • Doesn't try to change how you work, just removes noise

The limitation: It's still just about you and your preferences. Can't distinguish between emails that matter to you vs. emails that matter to your team's current priorities.

Perfect for: Anyone whose inbox makes them want to hide under a desk

4. Missive: the team player (sort of)

Missive is the only one that seems to remember teams exist. Revolutionary, I know.

Performance metrics:

  • Collaboration features: Shared inboxes, team assignments, internal comments

  • AI integration: OpenAI GPT with commands like "make this friendlier" or "translate to French"

  • Team coordination: Unified email management across team members

Sal the Salmon swimming in notifications
Sal the Salmon swimming in notifications

My hot take ❤️‍🔥:
Finally! Someone thinking about the multi-user problem. But they're still thinking about it like "individual productivity + sharing" rather than "team productivity from the ground up."

What they got right:

  • Shared workspaces that actually work

  • AI that can adjust tone on command (useful for global teams)

  • Internal comments without cluttering the actual email thread

Where it falls short:

  • The AI still thinks email-by-email, not team-by-team

  • Higher learning curve than the individual-focused tools

  • More expensive for small teams

Perfect for: Teams that need to share email duties, international teams that need translation help⚡ The coordination gap: what everyone's missing

🤕 The industry's collective blind spot: productive work is a team feat

The real problem (from a product development perspective)

Here's the thing that keeps me up at night as I'm building Upstream: every AI tool is optimized for this mythical solo founder who never needs to coordinate with anyone.

But that's not how we actually work, is it?

From my Asana days, I learned that the biggest productivity killer isn't individual inefficiency—it's coordination overhead. The same applies to email.

Current AI email tools excel at:

  • Recognizing your writing patterns

  • Prioritizing emails based on your behavior

  • Making you sound consistent

  • Tracking your individual productivity

What they completely miss:

  • Which conversations your co-founder is already handling

  • What your team decided in yesterday's standup that affects how you should respond

  • Whether this "urgent" email conflicts with your shared priorities

  • The context your teammates need to stay aligned

Conclusion: Where this is all headed (my totally biased take)

This whole deep-dive directly shaped what we're building at Upstream. While everyone else is racing to make individual email faster, we're tackling the team coordination mess. If you're interested in where team-context email AI is heading, drop me a line 🙂.

Wave 1 (now): Individual AI optimization

  • Superhuman-style speed

  • Shortwave-style voice matching

  • SaneBox-style filtering

Wave 2 (soon, hopefully): Team-context AI

  • Shared priority awareness

  • "Who's handling this?" intelligence

  • Collaborative decision integration

Wave 3 (the dream): Full organizational context

  • Company-wide communication understanding

  • Strategic priority integration

  • Predictive team coordination

(And yes, this is exactly what we're building at Upstream, but more on that later...)

📢 Want to support us?

👉 Sign up for our waitlist and be the first to try Upstream

Related Resources