Gemini Decision 1 is out: using it with your Focalyst plan

What Google released

Google announced Gemini Decision 1 as a model that answers typed questions with probabilities. It operates through a new Decisions API and does not write text. This model is multimodal, meaning it can process various types of input.

Google announced Gemini Decision 1 on November 11, 2026¹. It returns probabilities and confidence scores for up to 1,000 independent questions per request, with end-to-end latency between 150 and 250 milliseconds⁴. The model is multimodal, processing text, images, audio, video, and PDFs, and has a large context window of 262,144 tokens³.

Google uses Gemini Decision 1 internally for tasks such as routing queries in Gemini Notebook, acting as a safety layer for MCP tools in Search, and even picking emoji reactions². The independent Artificial Analysis Decision Index v2.1 ranks Gemini Decision 1 ninth out of 18 decision models at launch, scoring 66, which places it just above the field's median of 64⁵.

What a stronger model can do with your own data

A model like Gemini Decision 1 can help you make more self-directed choices about your Focalyst plan. By analyzing your Focalyst data, it can reveal patterns and probabilities, such as the likelihood of completing a task or sticking to a habit, helping you understand your own behavior better. This understanding is key to building agency and deliberate effort, which are core to owning your time. This is about using AI Leverage: multiplying what you can accomplish rather than using AI merely as a shortcut.

AI Leverage means getting probability-driven insights for tasks and planning. These insights do not replace your thinking; they amplify it. For example, the model can help you optimize your study schedule by looking at past focus sessions, like those tracked with the Pomodoro technique for short attention span: build focus, and habit completion rates [/blog/pomodoro-technique-for-short-attention-span]. If the model shows a higher probability of completing focused work in the mornings, you can deliberately schedule your most demanding cognitive tasks then. It can also assist in identifying potential bottlenecks in your workflow by analyzing your task history. You could ask it to suggest the likelihood of completing a complex project given your current habits and focus time, helping you adjust your approach before you even start.

Use it to determine the best time for a habit based on your past activity and calendar events. This helps you make choices based on your actual behavior, for example, identifying the best time for a habit based on past completion rates, not just assumptions. This moves you beyond guesswork, allowing you to apply deliberate practice to your routine. For example, you could ask the model to identify a two-hour block in your calendar events and past focus sessions with the highest probability of uninterrupted work. This kind of insight helps you see your current ability as a starting point, not a final limit, and then work to expand it.

How to connect Focalyst to Gemini

You can connect Gemini Decision 1 to your Focalyst data through our AI connections feature. This allows the AI app to read your tasks, habits, focus time, countdowns, and notes. The connection is made through the AI app itself, not built directly into the model.

To use AI connections, you need a Focalyst Pro plan. The specific Gemini models you can use depend on your plan with Google's AI app. To connect, go to the AI connections guide in Focalyst and follow the steps for Gemini [/guides/connect-ai-apps]. You can disconnect the AI app at any time, ensuring your data remains under your control. This feature provides a way for external AI apps to access your Focalyst data, including tasks, habits, focus sessions, countdowns, and voice note transcripts and summaries.

Prompts to try with your Focalyst data

Once connected, you can ask Gemini Decision 1 specific questions about your Focalyst data. This helps you get targeted insights, such as the probability of meeting a goal, and make better decisions about your time and priorities. Remember, AI Leverage means AI should amplify your thinking, not replace it.

Start by crafting prompts that ask for probabilities related to your specific goals and challenges, like the chance of completing a task or sticking to a new habit. Here are some examples to try this week:

• "Given my current tasks and priorities, what is the probability I will complete [Task X] by [Date Y]?"
• "Based on my habit streaks and completion rates, what is the likelihood I will stick to a new habit if I start it on [Day of Week]?"
• "Considering my past focus sessions and distractions, what is the optimal time block for deep work on [Project Z]?" This could help you plan your day in ten minutes for more control [/blog/how-to-plan-your-day-in-ten-minutes].
• "What is the probability that I will meet my [Countdown Event] goal if I maintain my current study habits?"
• "Summarize my voice notes from the last week and identify any recurring themes or urgent actions with their likelihood of being missed."

After running a prompt, analyze the probability scores and adjust your plan. For example, if the model suggests a low probability of completing a task, you might break it into smaller steps or reschedule it.

Limits to know

Gemini Decision 1 has limitations. Understand them before relying solely on its output. The model provides probabilities, not guarantees. Your human judgment remains essential; always question the outputs and use them as a guide.

We believe AI should amplify your thinking, not replace your ability to think independently. Over-reliance on any AI tool can weaken the very skills you are trying to develop, like problem-solving, reasoning, and decision-making. Use the probabilities to inform your choices, but always maintain your own direction and verification.

Gemini Decision 1 costs $0.10 per million input tokens on the paid tier, though a free tier is also available². There is no output charge because it produces no output tokens, only probabilities. Focalyst Pro is required for AI connections, and your access to Gemini models depends on your plan with Google.

The independent Artificial Analysis Decision Index v2.1 ranks Gemini Decision 1 ninth out of 18 decision models, scoring 66⁵. This means other models might offer different strengths. For example, if you are planning a complex project, you might cross-reference Gemini's output with your own intuition. This approach ensures you are using AI as an amplifier of your intelligence, not a substitute for it.

Another specific action you can take this week is to charge your phone outside the bedroom. This creates a physical distance from potential distractions, helping you manage your attention better regardless of AI outputs. We know that willpower alone is often not enough to compete with sophisticated attention-optimization systems. Designing your environment, like creating friction between you and your phone, is itself a form of discipline. These small environmental changes, like charging your phone outside the bedroom, support your decisions even when probabilities are uncertain. They help you regain control over your attention and build behaviors that eventually become more automatic.