Why Glass Had to Become Whisper

Jul 28, 2025

Product

We Got It Wrong: Why Glass Had to Become Whisper

1. Whisper vs. Glass: What’s the Difference?

When we launched Pickle Glass, our goal was to support human cognition in real time by moving beyond traditional prompt-based assistants. Glass was designed to passively observe your screen and understand your personal context. It also captured audio and visual signals from the start, offered in-the-moment suggestions, and learned from every interaction to get smarter over time. We believed that by understanding what you were doing in the moment, AI could stop depending on apps and commands and start offering truly intelligent support.

It quietly observed your digital environment and waited for the right moment to help. But as we used it more, we realized we got it wrong.

We asked ourselves a simple question: Does the future of AI wait for us to ask? The answer was clear. Absolutely not. The future of AI must be proactive. It should understand your intent, anticipate your needs, and take action before you even reach for it. That’s why we moved fast to evolve Glass into something entirely new. We launched Whisper.

With Whisper, we are moving from passive observation to active anticipation. Unlike the "Ask" feature in Glass, Whisper does not wait for you to click or speak first. It recognizes what you need and acts before you have to ask. It understands what you are doing, detects what you might need, and shows up without being asked. It is always one step ahead.

Proactiveness is not just a technical feature. It is the foundation of how we believe AI should work. A truly intelligent assistant should not behave like a passive tool. It should act more like a teammate. It should recognize friction, notice patterns, and offer help before you reach for it.

There is also a major difference in how the assistant presents itself. Glass came with a floating toolbar that lived on your screen. It was minimal, but it was static and predefined.

We came to realize that future AI should not rely on fixed interfaces. It should behave more like a chameleon, adapting its form and behavior based on the situation.

A predefined interface, like what we built with Glass, no longer makes sense. Whisper is different. It has no fixed position, no standard layout, and no always-on presence. Most of the time, it stays completely hidden. When it does appear, it responds to your context. Sometimes it offers a quick suggestion. Other times it surfaces a small card. And often, it stays quiet and lets you work.

The interface is not designed in advance. It emerges when needed, in the format that best fits the moment.

There is also a major difference in how the assistant presents itself. Glass came with a floating toolbar that lived on your screen. It was minimal, but it was static and predefined.

We came to realize that future AI should not rely on fixed interfaces. It should behave more like a chameleon, adapting its form and behavior based on the situation.

The interface is not designed in advance. It emerges when needed, in the format that best fits the moment.

We also realized something deeper. The way humans naturally communicate with machines should begin with voice, not text. Just like Jarvis in Iron Man, we believe spoken interaction is the most seamless, expressive, and intuitive interface. That is why voice is the primary way you interact with Whisper.

At the same time, Whisper understands your context. If you are in a meeting, it will switch to text so you can respond quietly. And if you are busy, there is no pressure to reply right away. Whisper works more like a walkie-talkie. You can respond when the timing feels right.

We believe this is the best way to create the feeling of talking to a real teammate. It supports your workflow without getting in the way, and feels human without demanding your attention.

Whisper is not just a new version of Glass. It is a complete rethinking of what an assistant can be. It watches you work. It understands your context. And it responds in the most natural and helpful way for that situation.

We believe this is the best way to create the feeling of talking to a real teammate. It supports your workflow without getting in the way, and feels human without demanding your attention.

2. Proactiveness Through Context: The Core Idea

The core of Whisper is simple.

AI should act first.

Whisper

Seems like you’ve been looking for

Whisper

Seems like you’ve been lookin

Most AI tools today behave like smart, obedient servants. They wait for your command, respond quickly, and get the job done. But the burden is still on you. You have to know what you need, phrase it clearly, and manage the interaction every time.

That’s not how the human brain works. And that’s not how we believe assistants should work either.

Real intelligence doesn’t wait. It pays attention. It connects dots across apps, moments, and context. It sees patterns, understands goals, and steps in before you ask. The best kind of support feels intuitive. It helps without being told.

That’s the foundation of Whisper. It learns how you work. It watches for friction. It remembers what you’ve done and anticipates what you might do next. And when the time is right, it quietly and proactively steps in with exactly the support you need.

This isn't just about speed or convenience. It’s a shift in how we think about AI altogether.

Whisper moves us from command-based systems to context (or “memory”)-based systems. Instead of waiting for prompts, it builds a mental model of how you work across tasks, screens, conversations, and time. It blends memory, intent, and timing into a continuous understanding.

Imagine finishing a meeting in Zoom where someone asks you to send over a follow-up document. Before you even leave the call, Whisper has already recognized the context, retrieved the file you need, and surfaced a draft message to send. You don’t need to search, copy, or switch tabs. It just appears, ready to go, at the exact moment you need it.

That’s what real assistance should feel like.

Whisper is not just smarter. It’s more human.

It senses, it anticipates, and it moves proactively so you can stay focused and in flow.

That’s not how the human brain works. And that’s not how we believe assistants should work either.

This isn't just about speed or convenience. It’s a shift in how we think about AI altogether.

That’s what real assistance should feel like.

Whisper is not just smarter. It’s more human.

It senses, it anticipates, and it moves proactively so you can stay focused and in flow.

3. Why Proactiveness Matters

The cost of the reactive model is larger than it seems. Traditional assistants struggle to truly complement human cognition for two key reasons:

External Cognitive Offloading
Most AI tools turn users into their own managers. You are responsible for remembering what needs to be done, structuring it into a prompt, and guiding the assistant through each step. This creates a subtle but constant form of external cognitive offloading. You begin to rely on the tool not just to assist, but to think for you.

Over time, this can reduce active engagement, limit creativity, and increase dependence on reactive support. You are outsourcing mental effort without gaining real momentum.

Whisper takes a different approach. It builds a memory of how you work. It understands your habits, your flow, and your intent. Instead of making you explain everything, it recognizes the moment and offers support before you ask. It helps you stay focused without adding more decisions to your day.

The Hidden Cost of Context Switching
Modern work is fragmented. You are constantly moving between tools, tabs, conversations, and tasks. Each switch forces your brain to reset. This process, known as context switching, carries a measurable cognitive cost.

Studies show that frequent switching reduces performance, increases error rates, and creates mental fatigue. Even short interruptions can take 20 to 30 minutes to fully recover from. What seems like a small pause is often a serious disruption.

Whisper is built to reduce these micro-breakdowns. It anticipates what you are trying to do and quietly offers help without requiring you to stop, reframe your goal, or shift your focus. It supports your momentum and keeps you in flow, where meaningful work actually happens.

External Cognitive Offloading
Most AI tools turn users into their own managers. You are responsible for remembering what needs to be done, structuring it into a prompt, and guiding the assistant through each step. This creates a subtle but constant form of external cognitive offloading. You begin to rely on the tool not just to assist, but to think for you.

Over time, this can reduce active engagement, limit creativity, and increase dependence on reactive support. You are outsourcing mental effort without gaining real momentum.

Whisper takes a different approach. It builds a memory of how you work. It understands your habits, your flow, and your intent. Instead of making you explain everything, it recognizes the moment and offers support before you ask. It helps you stay focused without adding more decisions to your day.

The Hidden Cost of Context Switching
Modern work is fragmented. You are constantly moving between tools, tabs, conversations, and tasks. Each switch forces your brain to reset. This process, known as context switching, carries a measurable cognitive cost.

Studies show that frequent switching reduces performance, increases error rates, and creates mental fatigue. Even short interruptions can take 20 to 30 minutes to fully recover from. What seems like a small pause is often a serious disruption.

Whisper is built to reduce these micro-breakdowns. It anticipates what you are trying to do and quietly offers help without requiring you to stop, reframe your goal, or shift your focus. It supports your momentum and keeps you in flow, where meaningful work actually happens.

4. How Whisper Works

Whisper runs quietly in the background. There are no visible toolbars, buttons, or floating UI elements. Most of the time, it stays out of sight until it sees a moment where it can help.

Case: Whisper live-checks context during remote meetings.

Here’s how it works:

It continuously observes your screen and listens for context signals. All processing is secure and end-to-end encrypted.
It builds memory across actions, screens, and moments. Whisper knows if you're writing a doc, designing slides, in a Zoom call, or switching between tabs during research.
When it recognizes a moment of friction or need, it offers help in a way that fits your situation. For example:
- While you’re writing yet another quarterly peer review, it might say: “This sounds just like your last one. Want to try rephrasing it like this?”
- If you always use the same sentence in emails, it might suggest: “Instead of ‘Please inform me about the current progress of your tasks,’ how about ‘Bring me up to speed’ this time?”
- When you're trying to build a spreadsheet formula and getting stuck, it might say: “On the current screen, paste vlookup(...) into cell F9 and press ⌥ (Option) key.”
Whisper decides how to speak to you based on context. If you're alone and hands-free, it might speak out loud. If you're in a meeting or deep in focus, it will show a quiet text suggestion on screen. There is no fixed UI. The interface adapts to your environment instead of staying rigid or permanent.
You can respond by voice or text, either right away or later. Whisper acts more like a smart walkie-talkie. It gives you space to reply when the moment feels right, without demanding your attention.

Here’s how it works:

It continuously observes your screen and listens for context signals. All processing is secure and end-to-end encrypted.
It builds memory across actions, screens, and moments. Whisper knows if you're writing a doc, designing slides, in a Zoom call, or switching between tabs during research.
When it recognizes a moment of friction or need, it offers help in a way that fits your situation. For example:
- While you’re writing yet another quarterly peer review, it might say: “This sounds just like your last one. Want to try rephrasing it like this?”
- If you always use the same sentence in emails, it might suggest: “Instead of ‘Please inform me about the current progress of your tasks,’ how about ‘Bring me up to speed’ this time?”
- When you're trying to build a spreadsheet formula and getting stuck, it might say: “On the current screen, paste vlookup(...) into cell F9 and press ⌥ (Option) key.”
Whisper decides how to speak to you based on context. If you're alone and hands-free, it might speak out loud. If you're in a meeting or deep in focus, it will show a quiet text suggestion on screen. There is no fixed UI. The interface adapts to your environment instead of staying rigid or permanent.
You can respond by voice or text, either right away or later. Whisper acts more like a smart walkie-talkie. It gives you space to reply when the moment feels right, without demanding your attention.

Whisper

is listening..

Whisper

is listening..

Whisper is not driven by prompts. It does not rely on predefined scripts or fixed interactions. It is powered by memory, context, and timing. It blends into your workflow, responds fluidly, and feels more like a helpful teammate than a tool.

5. Whisper’s Future: From Assistant to Agent

Whisper is just the beginning.

We built Whisper to move beyond reactive tools and into something smarter. It understands what you're doing and offers help before you ask. But the next step is even more powerful.

We believe the future of AI isn't just about assistance. It's about agency.

Right now, Whisper supports you in the moment. It reads context, suggests actions, and helps you stay in flow. But a true agent will do more. It will understand your long-term goals, remember how you like to work, and take initiative to move things forward without waiting for instructions.

Imagine an agent that doesn't just remind you to follow up, but actually drafts the message, finds the right contact, and sends it once you approve. Or one that schedules meetings, updates documents, and handles small tasks across different tools, quietly working alongside you.

Whisper is the foundation for this future. It builds memory, learns your preferences, and adapts to your routines. It removes friction, one step at a time.

An agent doesn't just help you when you ask. It helps you accomplish more, even when you're not paying attention.

Whisper is already changing how we work. And this is only the start.

Whisper is just the beginning.

We built Whisper to move beyond reactive tools and into something smarter. It understands what you're doing and offers help before you ask. But the next step is even more powerful.

We believe the future of AI isn't just about assistance. It's about agency.

Whisper is the foundation for this future. It builds memory, learns your preferences, and adapts to your routines. It removes friction, one step at a time.

An agent doesn't just help you when you ask. It helps you accomplish more, even when you're not paying attention.

Whisper is already changing how we work. And this is only the start.