Futures
Access hundreds of perpetual contracts
CFD
Gold
One platform for global traditional assets
Options
Hot
Trade European-style vanilla options
Unified Account
Maximize your capital efficiency
Demo Trading
Introduction to Futures Trading
Learn the basics of futures trading
Futures Events
Join events to earn rewards
Demo Trading
Use virtual funds to practice risk-free trading
CFD
U.S. stock CFD derivatives
US Stocks
Access real US stocks and ETFs
HK Stocks
Trade quality Hong Kong-listed stocks
Korean Stocks
SK Hynix
Real Korean stocks and top assets
Stock Futures
High leverage, 24/7 trading
Tokenized Stocks
Backed by real stock assets
IPO Access
Unlock full access to global stock IPOs
GUSD
Mint GUSD for Treasury RWA yields
Stocks Activities
Trade Popular Stocks and Unlock Generous Airdrops
Launch
CandyDrop
Collect candies to earn airdrops
Launchpool
Quick staking, earn potential new tokens
HODLer Airdrop
Hold GT and get massive airdrops for free
IPO Access
Unlock full access to global stock IPOs
Alpha Points
Trade on-chain assets and earn airdrops
Futures Points
Earn futures points and claim airdrop rewards
Promotions
AI
Gate AI
Your all-in-one conversational AI partner
Gate AI Bot
Use Gate AI directly in your social App
GateClaw
Gate Blue Lobster, ready to go
Gate for AI Agent
AI infrastructure, Gate MCP, Skills, and CLI
Gate Skills Hub
10K+ Skills
From office tasks to trading, the all-in-one skill hub makes AI even more useful.
Google Workspace CLI author Justin Poehnelt has been dismissed, and the official version has been launched.
In April 2026, Google announced the official development of the Workspace command-line tool at the Cloud Next conference. Two days later, Google engineer Justin Poehnelt, who built this tool himself, was dismissed. About a month and a half later, the official version launched, with a design almost identical to the tool he was fired over.
(Background: Google’s promotion system shifted to “AI performance,” and employees fragmented the product ecosystem)
(Additional context: Meta CTO admits fault: the new AI division “manages terribly,” and 6,500 engineers are now eligible to transfer to other roles)
Table of Contents
Toggle
Summary Highlights
On March 5, 2026, Addy Osmani posted on social media: “Introducing the Google Workspace CLI: built for humans and agents.” This tool is called gws, capable of unified access to Google Drive, Gmail, Calendar APIs via a single command-line interface, and it doesn’t hardcode commands. Instead, it dynamically generates the entire interface at runtime from Google’s Discovery Service. When Google APIs update, the tool updates automatically. It includes over 40 AI agent skill files, designed from day one for AI agents.
The post quickly spread within developer communities, garnering thousands of stars and real users.
The creator of this tool is Justin Poehnelt, an engineer on Google Workspace Developer Relations, who had been at Google for nearly 7 years. His work mainly involved building open-source tools on top of Google APIs. He later said that Addy Osmani’s post was “the post that got me fired.”
Google fires him
About a month later, at Google Cloud Next 2026, Google officially announced that the Workspace command-line tool was “in development.”
Two days after that announcement, Justin Poehnelt was dismissed.
A term in tech history called “Sherlocking” originates from 2002, when someone created a Mac tool called Watson that made searching everywhere very easy. Apple later integrated Watson’s core features directly into its own Sherlock, rendering third-party tools obsolete. (The story references Dr. Watson and Sherlock Holmes)
Being “Sherlocked” means your product has been copied or absorbed by the platform owner. Justin Poehnelt’s situation is even worse: he was an employee of the platform company himself and was dismissed simultaneously.
Official version launched like a draft written by hand
About a month and a half after his dismissal, the official Google Workspace CLI was officially launched on GitHub (googleworkspace/cli).
Written in Rust, licensed under Apache 2.0, it dynamically generates command interfaces from Discovery Service at runtime, includes over 100 agent skills, outputs structured JSON, and integrates with MCP server to connect Claude Code and Gemini CLI.
gws has over 40 agent skills; the official version has over 100. Both read from Discovery Service at runtime. gws was designed from day one for AI agents, and the official version’s positioning statement is identical.
Addy Osmani, then Director of Google Cloud AI, who had been at Google for 14 years and led developer experience for Chrome and Google Cloud AI, also left Google later.
Workspace division fears being overtaken by its own product
Justin Poehnelt interprets his firing as stemming from the Workspace team and some managers’ fear of being disrupted—an overall anxiety that AI agents will replace existing products, not just about his tool.
This logic isn’t new. In large organizations, innovation often dies not from technical issues but from internal territorial anxieties. A well-made open-source tool that’s so good it makes product managers uncomfortable can be explained without conspiracy theories. Moreover, employees releasing open-source tools that incorporate Google product features without full company authorization makes firing a “necessary evil.”
The timeline makes the close proximity of the official launch announcement and Justin Poehnelt’s dismissal seem cold.
Frequently Asked Questions
What is Sherlocking, and how does it relate to this case?
Sherlocking originated in 2002 when Apple integrated the core features of a third-party tool called Watson into its own Sherlock, making the original product obsolete. Justin Poehnelt’s case is a more advanced form of Sherlocking: he is an employee of the platform itself, and his tool was absorbed and turned into the official version. He was also dismissed two days after the official announcement.
How similar are the Google official Workspace CLI and Justin Poehnelt’s gws?
The core design is almost identical: both read from Google Discovery Service at runtime to dynamically generate command interfaces, both output structured JSON, and both include numerous AI agent skill files (gws has over 40, the official version over 100). The official version also adds MCP server integration, connecting Claude Code and Gemini CLI. Justin Poehnelt himself has publicly pointed out the similarities in design philosophy.