Scarlett: The AI Co-Worker That Makes You Ask Why

Scarlett is an AI co-worker that lives in Slack and iMessage. It drafts messages, summarizes threads, sets reminders, and answers questions about your company’s documents. It launched on Product Hunt this week, and the pitch is straightforward: treat the AI like a colleague you message instead of a dashboard you log into.

The product is listed on Product Hunt with the tagline “Your AI Co-Worker in Slack & iMessage.” It is not a new model. It is not a breakthrough in reasoning. It is a thin wrapper around existing large language models, delivered inside the two messaging apps where most knowledge workers already spend their days.

That is exactly why it matters.

Scarlett represents a specific bet about how AI enters the workplace. Not through a new IDE plugin, not through a custom chatbot portal, not through an API that developers integrate. Through the chat window that is already open. The bet is that the path of least resistance wins, and the path of least resistance for most office workers is the message bar.

The product does not appear to offer anything technically novel. The Scarlett website describes features that dozens of other AI assistants already offer: meeting notes, document search, task management. What is different is the distribution strategy. Slack and iMessage are not just channels. They are the operating systems of modern office work. By embedding itself there, Scarlett bypasses the adoption problem that plagues enterprise software. There is no onboarding flow. No new tab to open. The AI is just another contact.

This is the same logic that drove Microsoft to put Copilot into Teams and Google to put Duet AI into Gmail. The difference is that Scarlett is a startup, not a platform giant. It does not own the chat infrastructure. It rents access. That makes it vulnerable to platform risk. Slack could change its API terms tomorrow. Apple could restrict iMessage bot access. Scarlett’s entire business model rests on permission that could be revoked.

The more interesting question is what Scarlett reveals about the market’s appetite for AI co-workers. The Product Hunt launch suggests genuine demand. The comments section shows users who want exactly this: an AI that does not require them to change their habits. They do not want to learn a new tool. They want the tools they already use to get smarter.

That is a reasonable ask. It is also a trap.

An AI that lives in your chat window is an AI that optimizes for chat-shaped tasks. It summarizes, it drafts, it reminds. It does not build, analyze, or create in any deep sense. It makes you faster at the shallow work that already fills your day. It does not help you escape that work. Scarlett, like most AI co-workers, is a productivity tool for the status quo. It helps you do more of what you already do, not do something different.

That is the critique that matters. The AI industry has spent two years selling the idea that agents will transform knowledge work. Scarlett is a reminder that the market, left to its own devices, prefers incremental improvement. Users do not want an AI that reimagines their workflow. They want an AI that handles the parts of their workflow they find annoying.

The economics support this. Scarlett’s pricing is not listed on its Product Hunt page, but comparable products charge between $20 and $50 per user per month. That is cheap enough to be an expense-account purchase. It is not cheap enough to be trivial for a company of 500 people. The unit economics of AI co-workers are still being worked out. Each query costs the provider money in inference compute. If Scarlett gains traction, its margins will depend on how efficiently it can route queries to the cheapest capable model.

The real test for Scarlett is not whether it works. It is whether it sticks. Many AI productivity tools see a spike of interest followed by abandonment. The chat interface is familiar, but it is also noisy. A message from Scarlett competes with messages from actual colleagues. Over time, users may find that the AI’s summaries are not quite right, its reminders are slightly off, its answers are plausible but wrong. The cost of checking the AI’s work can exceed the time saved.

Scarlett’s creators likely know this. The product’s emphasis on being a “co-worker” rather than a “tool” is a deliberate framing. A co-worker you can correct. A tool you discard. The language matters.

What Scarlett signals to the broader AI industry is that the next battleground is not capability. It is context. The models are good enough. The question is where you put them. Scarlett puts them in the message bar. Other startups will put them in the calendar, the email composer, the code editor, the spreadsheet. The winner will not be the company with the best model. It will be the company that puts a good enough model in the place where the user already works.

That is a harder problem than it sounds. It requires integration with systems that were not built for it. It requires trust from users who have been burned by bad AI. It requires pricing that makes sense for both buyer and seller. Scarlett has solved none of these at scale. It has only launched.

The product is worth watching for what it represents, not for what it is. It is a test of whether the chat-first AI co-worker model can sustain a business. If Scarlett succeeds, expect a wave of imitators. If it fades, the industry will learn that embedding AI in existing interfaces is not enough. Users need a reason to talk to the AI, not just a way to.

Scarlett’s launch day is not the story. The story is what happens in the six months after.