The Superhuman Suite: the software starts doing the work.
I have been getting hands on with the full suite, Mail, Calendar, Grammarly, Docs, Databases, and Go. Here is what I learned in each one, and the shift I keep coming back to: the app stops being the product, and the finished work becomes the product.
AI made me faster. It did not change how my team works. Superhuman's own launch data says the same thing: 65% of workers say AI helped their personal productivity, but only 12% say it changed how their team works.
What I noticed using the suite is where that gap starts to close. The software stops handing you a tool to operate and starts handing you the finished work. Docs builds the thing you would have bought. Go acts across your other apps. Grammarly writes in your voice without being asked. The agent does the task; you make the call.
That is the shift I keep coming back to. For thirty years you bought software and ran it yourself. This is the first stack I have used where the software runs the work and hands you the result.
Six things that stood out once I actually used it.
Most AI answers a question and waits. Go actually does the thing, across the apps you already have open, and often before you ask. That is the part that felt new.
I kept reaching for a new tool out of habit, then realizing I could just describe the workflow and have it built. The purchase decision quietly disappears.
It is already on 40M machines. The suite ships with distribution most products spend years trying to buy, and it keeps one voice across everything you send.
Scheduling from inside the inbox, availability shared inline, no app switch. It is the small, believable version of the AI acting across your tools.
A nice doc is a demo. A doc backed by a real data layer, with permissions and a million rows, is something a team can actually run on.
When Claude can read and write your docs in real time, the doc stops being a place you visit and becomes something your AI operates for you.
What each product actually does, once you use it.
A quick, honest review of all six. Mail and Calendar handle the individual work. Grammarly is the voice layer already on every machine. Docs and Databases are where the work gets built and kept. Go is the agent that carries the context across all of it, and into the 100+ other tools you already run.
Superhuman Mail
The email client rebuilt for speed and relationships, not inbox archaeology. Keyboard-first and AI-assisted, built to get you to done and keep the relationship warm.
- 1Connect your Gmail or Outlook account and run the guided onboarding, which learns your workflow.
- 2Set up Split Inbox so VIPs, team, and noise land in separate lanes.
- 3Turn on follow-up reminders and Send Later so nothing you are waiting on slips.
- Triage fast with keyboard shortcuts and AI summaries of long threads.
- Use snippets for the messages you send constantly, personalized on the fly.
- Let AI draft and follow up, so the inbox becomes a relationship engine instead of a time sink.
Grammarly
The writing layer that learns your voice and rides along everywhere you type. In the suite, it is how one company voice shows up in every message, already installed across the enterprise.
- 1Install Grammarly across the surfaces your team writes in: browser, desktop, and mobile.
- 2Set up your style and brand-voice profile so suggestions match how you actually sound.
- 3For teams, roll out shared style guides so everyone writes on-brand by default.
- Get real-time clarity, tone, and correctness as you write, in your voice.
- Enforce one consistent company voice across every rep and every channel.
- Lean on its 40M-user distribution: it is already on enterprise machines, so adoption is not a new procurement fight.
Superhuman Docs
THE ENGINECoda, reborn AI-native. A plain-language prompt becomes drafted content, structured tables, and interactive views, backed by Databases that scale to a million rows and one-click MCP into Claude, ChatGPT, and Cursor. This is not a doc. It is where a team builds the tool it would otherwise buy.
- 1Start one doc for a real team ritual: a tracker, a launch plan, an OKR board.
- 2Ask Docs AI in plain language to build the tables, views, and automations you need.
- 3Add a Database for the data that has to scale, and connect your sources with Packs (Jira, Salesforce, Asana).
- 4Turn on MCP so Claude or ChatGPT can read and write to the doc and tables in real time.
- Replace a point tool with a build: describe the workflow and let Docs assemble it.
- Keep the team working in one surface. Review, assign, and act, instead of exporting to a spreadsheet.
- Let AI run the recurring work in the doc, or via MCP from your assistant: draft, track, update.
Superhuman Databases
The data layer under Docs. One connected store, up to a million rows on Enterprise, that stays current in every doc and view. This is what turns a Doc from a page into the team's system of record, with the permissions and durability an enterprise actually evaluates.
- 1Promote the tables that matter into a Database so many docs read one source.
- 2Set granular permissions: who can view, comment, edit rows, or manage the base.
- 3Connect the sources that feed it with Packs, so the records stay live.
- Let every team read the same current records right where decisions get made.
- Scale past spreadsheet limits without losing the doc-native feel.
- Ask Docs AI to chart or answer questions off the live data in plain language.
Superhuman Go
The proactive assistant that works across 100+ apps and tabs. It knows what you know and acts, without you switching tools. In the suite, Go is the agent layer that carries context from Mail, Docs, and Databases into wherever you are working, and the Agent Store plus Agents SDK make it a platform, not a feature.
- 1Turn on Go and connect the apps your team lives in: Gmail, Calendar, Jira, and more.
- 2Add agents from the Agent Store, or build your own with the Agents SDK.
- 3Let it observe your workflow so its help gets sharper over time.
- Get proactive help in context: polish a reply, prep for a meeting, file a ticket.
- Pull account and doc context into the moment without leaving the app you are in.
- Extend it with partner and custom agents so it acts across your whole stack.
Superhuman Calendar
Scheduling built into the inbox. Turn an email into an event, share availability inline, and find mutual time without leaving the thread. It is the small, believable proof of the whole suite promise: the AI acts across your tools so you never switch apps.
- 1Connect Google and Outlook calendars so you see them in one view.
- 2Set your availability and booking preferences once.
- 3Use natural language: ask AI to find time and draft the follow-up.
- Create an event from an email in one step, with the right attendees and time.
- Share availability and a booking link inline, with no double-booking.
- Type a teammate's name and see mutual free time instantly.
Build me a [workflow] operating surface for my [team]. Include: a table for [the records], the views we need ([by owner], [by status]), an automation that [does the recurring thing], and a summary section the team reads first. Ask me for anything you need before you build.
The app stops being the product. The outcome becomes the product.
A thesis I am still testing, but the pattern is hard to unsee.
SaaS sold you a tool and left the work to you. You bought the seat, learned the app, and did the job yourself. What I keep seeing in this suite is the software closing that last step: the agent runs the task and hands you the result to approve.
Go acts across your apps. Docs builds the tool you would have bought. The recurring work happens on its own and waits for your call. You are not operating software anymore, you are reviewing outcomes.
I am not certain how far it runs, and I would want to pressure test it with a real team. But if it holds, the thing you pay for stops being access to an app and starts being the work itself.
Nine pieces of work the suite just does for you.
This is where the thesis stops being a thesis. Each one is a real outcome the software produces, with the agent doing the task and you making the call. Read the AI-does / you-do split on every card, that is the shift in one line. Every one is buildable today on a native Pack plus Docs AI. No code.
Deal meeting, CRM updated for you
Every deal call ends with the CRM already updated: next steps, risks, and stage, no manual logging.
- 1Add the Fireflies and Salesforce Packs to one deal-desk Doc.
- 2Fireflies syncs each call transcript into a table automatically.
- 3A Docs AI block pulls next steps, risks, and deal fields into columns.
- 4A Coda button writes the approved fields to the linked Salesforce opportunity.
- 5The Slack Pack posts the recap to the deal channel.
The internal meeting OS
Standups, 1:1s, and staff meetings run themselves: notes, action items, owners, and the next agenda.
- 1Create one Doc per recurring meeting.
- 2The Fireflies Pack pulls the transcript in.
- 3A Docs AI block generates the summary and an action-item table.
- 4An automation assigns owners and due dates and Slack-notifies them.
- 5Docs AI drafts the next agenda from the open items.
Project meeting to live status
Project syncs update the plan itself: task status, decisions, and risks land in the tracker as you talk.
- 1Build a project Doc with a Database (tasks, milestones) and the Jira Pack.
- 2Fireflies pulls the project meeting in.
- 3Docs AI updates task status and logs decisions and risks to the table.
- 4An automation flags blockers and sends a Slack digest to the team.
The self-enriching prospecting table
A target list that enriches itself and drafts the first touch, with Coda as the operating table.
- 1Pull target accounts into a Coda Database (Salesforce Pack or paste).
- 2Point Clay at the Coda table, or add the Apollo Enrichment Pack, to fill emails, phones, and firmographics.
- 3A Docs AI block drafts the first-touch message per contact from the enriched context.
- 4Approve, then the Salesforce Pack syncs the enriched, messaged records back.
The PLG to GTM auto-handoff
Product usage becomes qualified pipeline, with a handoff checklist that keeps both teams in sync.
- 1Product connects product-usage or signup data into a Coda Database.
- 2Clay enriches, and an ICP score is computed right in the table.
- 3When an account crosses the threshold, an automation creates the CRM opportunity and an owned handoff checklist.
- 4The Slack Pack notifies GTM with the context; the checklist tracks the SLA.
The RevOps command surface
Pipeline hygiene, routing, and forecast prep run on one doc instead of ten spreadsheets.
- 1Two-way sync the pipeline in with the Salesforce Pack.
- 2Docs AI blocks flag stale deals and missing fields.
- 3An automation nudges the owning rep in Slack.
- 4Docs AI drafts the forecast-call narrative for the leader to review.
The customized onboarding hub
A personalized onboarding Doc per hire: role-specific 30/60/90, people to meet, tasks that assign themselves.
- 1Build one onboarding template Doc with a Database for hires and tasks.
- 2Docs AI generates the role-specific 30/60/90 and people-to-meet from the role.
- 3Automations assign tasks with due dates and send Slack or Gmail intros.
- 4The Database tracks time-to-value and completion across every hire.
The customer onboarding doc
A success plan per account that starts from the kickoff call and tracks time-to-value.
- 1Fireflies pulls the kickoff call into the account Doc.
- 2Docs AI drafts the success plan and milestones.
- 3A Database tracks milestones and time-to-value.
- 4Automations create CSM tasks and Slack-nudge on risk.
The board and OKR operating doc
Board prep and company OKRs build themselves from the numbers, so the exec edits instead of assembles.
- 1Pipe metrics into a Database via the Salesforce, Sheets, and other Packs.
- 2Docs AI rolls up OKR progress from the team docs.
- 3Docs AI drafts the board narrative from the numbers.
- 4The exec reviews and edits in one surface.
Every one runs on Docs and Databases plus the tools a team already owns, Fireflies, Clay, Apollo, Salesforce, Jira, Slack, Gmail, with Go carrying the context between them. String a few together and you stop opening apps to do work. The work arrives done, and you spend your time deciding, not operating. That is the whole shift, in practice.
Is all of this actually doable, without code?
Yes. Every play uses a native Coda Pack (Fireflies, Salesforce, Apollo, Clay, Slack, Jira) plus Docs AI and automations. You describe it in plain language; nothing here needs an engineer.
Is Superhuman Docs just for GTM teams?
No. It grew up in operations and product, and it is used across Chief of Staff, People, Finance, IT, and engineering. Any team that keeps exporting its work to a spreadsheet is a fit.
Why does Docs matter most in the suite?
Because Docs and Databases are where the work actually gets built and kept. Mail, Calendar, and Grammarly speed up what you already do; Docs is where a task you used to buy a tool for becomes something the software just builds and runs. If you only try one thing, start there.
What makes it a build-not-buy decision?
With Docs AI, Databases, and one-click MCP, the tool a team would otherwise buy for one workflow becomes a Docs build in plain language, connected to the data and the AI it already uses.
How does the AI actually help?
Docs AI can read, plan, write, track, and build inside the doc from plain-language prompts, and MCP lets Claude or ChatGPT read and write to your docs and tables in real time.
Where do Go, Calendar, and Databases fit?
Databases is the data layer that makes Docs a real system of record at enterprise scale. Calendar is the scheduling layer built into Mail. Go is the agent that carries the context from all of them into the 100+ other apps your team works in, and, via the Agent Store and SDK, becomes a platform others build on. Docs still lands the team; the rest of the suite compounds on top.
Is this hype, or does it hold up today?
Most of it holds up today. The product review and every playbook here are things you can build now on native Packs plus Docs AI. The bolder claim, that the software starts delivering the work and not just the tool, is a direction I am still pressure testing, not a finished fact. That is the honest line.
That is where I have gotten to. Tell me where I am wrong.
This is one deep dive in the How to AI series, and a thesis I am still testing in the open. If you are building on the suite, or you think the outcome read is off, I want to hear it. Or see every tool worth knowing and how to connect it.