How to Give AI Context About Your Goals

It’s Tuesday. Your AI drafts your emails and researches your competitors. But it has no idea what goals you’re pursuing this quarter, let alone this year.

I’m known as the goals guy. I wrote Your Best Year Ever, and goals sit at the center of the Full Focus Planner and everything we teach at Full Focus.

You’d think I would have handed my goals to AI on day one. I didn’t.

When I first started using AI, I saw goals as something outside it. The most I thought AI could do was help me write a goal so it fit the framework in my book. I even built a custom GPT for that, the Goal Achievement Assistant.

Then I started building a workforce of agents and bots, and something felt off. Each one was busy. Each one was working on its own project. But nothing pointed at what I wanted for my life or my companies. There was no North Star. My agents had started to feel like silos, all moving fast in different directions.

That was my turning point. With agents and bots, my job had become aligning a workforce, and a workforce without a destination just produces activity.

So I made a change. Now every agent on every LLM, and every bot I run in Grok Bot, starts the morning by rereading my goals. Those goals shape everything we do that day.

I’ve never been more goal-oriented than I am right now.

I’ve been writing about AI agents and bots for the last couple of weeks, including the team I’ve built on Grok Bot. Every one of them depends on something I skipped at first: AI agent goals that you define. Notice who does the defining. You do. Skip that job, and your agents will work hard at nothing in particular.

If you want every agent and bot pulling in the same direction, you can give them a destination, whatever platform you use. Here are four MOVES for putting your goals in front of your AI every day.

Move #1: Define the Destination

Good AI agent goals are specific enough that an agent can’t misread them.

Psychologists Edwin Locke and Gary Latham summarized thirty-five years of goal-setting research and found that specific, difficult goals consistently beat telling people to do their best.¹ “Help me grow the business” is a do-your-best goal. An agent has no way to know whether it’s winning.

Gartner predicts that more than 40 percent of agentic AI projects will be canceled by the end of 2027, citing rising costs, unclear business value, or weak risk controls.² Its analysts don’t blame missing goals. Still, unclear business value is what you get when nobody says what an agent is for.

So write the result and the date. Say, “Open a second location by June 30, 2027.” Then add one sentence on why it matters to you. Research they summarize found that a goal assigned tersely, with no explanation, led to significantly lower performance than one set with the person’s input. Explaining the purpose closed the gap.³ An agent can’t lean over and ask why. Put the reason on the page.

Move #2: Divide It by Time Horizon

A ten-year vision can’t tell an agent what to do on Tuesday.

My goals file runs in layers: life vision, multi-year goals, annual goals, quarterly goals, then a Big 3 for the quarter, the week, and the day. Each layer feeds the one below it. A Tuesday task traces all the way up to the vision.

The research backs the layering. Locke and Latham describe a business simulation in which players given near-term goals alongside a distant one earned significantly higher profits than players given only the distant goal.⁴ Near goals tell an agent what to do next. Far goals tell it why.

You don’t need six layers. Start with two: where you’re headed, and the three results that matter most this quarter. Add the rest when you feel the gap.

Move #3: Deliver It at the Start of Every Session

How to give AI context is mostly a delivery problem.

Agents often start cold. Anthropic’s documentation for Claude Code puts it plainly: each session begins with a fresh context window. One of the two fixes it offers is a file the tool reads at the start of every session, “the place you write down what you’d otherwise re-explain.”⁵ Look for your platform’s version of this, and put your goals there.

Think of it as your AI context file. You have three ways to deliver it, from simplest to most powerful:

1. Paste it. Drop the page at the top of any new chat. It’s low-tech, and it works everywhere.
2. Store it in the tool. Put it in whatever your platform calls its standing instructions, so it loads without you.
3. Make it a file every agent reads. Mine is a markdown file in my Obsidian vault, where I keep my notes. Every agent and bot opens it when it starts. Any shared document your agents can open works too.

I’ll walk you through all three in the Implementation Kit. If you already have an Obsidian vault, goals.md sits in your bootstrap sequence, the short list of files your agents read first every morning.

Whichever you pick, keep it short. Anthropic’s guidance for files like this is to stay under 200 lines, because longer files use more context and reduce how well the model follows them.⁶ One page is plenty.

Move #4: Direct Every Task Toward It

A goals file does nothing if it only sits there. Put it to work.

Locke and Latham call this the directive function: goals steer a person’s attention and effort to the work that serves the goal and away from the work that doesn’t.⁷ That’s what you want from every agent. Tell each one to check a task against your goals before it starts, and to name the goal the task serves. If it can’t name one, the task waits.

I built this into the file itself. Agents can propose new goals, but nothing enters the file without my yes. One agent’s job is to ladder my weekly and daily priorities up to the goals. And anything an agent can clear on its own never lands on my desk.

Alignment matters even more with a team. One study Locke and Latham cite found that when a person’s specific, difficult goals matched the group’s goal, group performance improved. Without that match, personal goals hurt it.⁸ Siloed agents are individuals with no shared goal.

If you haven’t handed an agent its first assignment yet, start with my September 29 issue, What Can AI Agents Do for Your Business? Then come back here.

Give Your AI a North Star

Define the destination. Divide it by time horizon. Deliver it at the start of every session. Direct every task toward it. Four moves, one short AI context file.

Imagine opening your laptop on Monday and knowing that every agent you run, on every platform, already knows what you’re after this quarter. No re-explaining. No busywork pointed the wrong way. Just work aimed at the result you care about most.

What’s the one result you’d want every agent you run to be working toward by the end of this quarter?

Comments

If you have a question about giving your AI your goals, click here to send me an email. I read every one. Seriously. Your experiences help me write better content, and sometimes the best insights come from readers like you. 

Transforming AI from noise to know-how,

P.S. Consider the AI Business Lab Mastermind: Running a $1M+ business? You’re past the startup chaos but not quite at autopilot. That’s exactly where AI changes everything. The AI Business Lab Mastermind isn’t another networking group—it’s a brain trust of leaders who are already implementing, not just ideating. We’re talking real numbers, real strategies, real results. If you’re tired of being the smartest person in the room, this is your new room. 👉🏼Learn more and apply here.

---

REFERENCE

1. Edwin A. Locke and Gary P. Latham, “Building a Practically Useful Theory of Goal Setting and Task Motivation: A 35-Year Odyssey,” American Psychologist 57, no. 9 (September 2002): 706. ↩︎
2. “Gartner Predicts Over 40% of Agentic AI Projects Will Be Canceled by End of 2027,” Gartner, press release, June 25, 2025. ↩︎
3. Locke and Latham, “Building a Practically Useful Theory,” 708. ↩︎
4. Locke and Latham, “Building a Practically Useful Theory,” 709. ↩︎
5. “How Claude Remembers Your Project,” Claude Code Docs, Anthropic, accessed September 30, 2026. ↩︎
6. “How Claude Remembers Your Project.” ↩︎
7. Locke and Latham, “Building a Practically Useful Theory,” 706. ↩︎
8. Locke and Latham, “Building a Practically Useful Theory,” 712. ↩︎

How I use AI in my writing process →