Build your personal AI copilot

Tal Raviv argues that LLMs become genuinely useful for complex knowledge work only when given persistent, rich context about a person's goals, role, team, and organization. He calls this setup a personal "AI copilot" that works with the user rather than replacing their thinking. The piece walks through building one using Projects-style features: hiring it via instructions, onboarding it with project knowledge documents, running each initiative in its own chat thread, and then applying it to prototyping, automations, and ongoing "gossip" updates. It matters because most people underuse AI by withholding the context that makes it effective.

01Key takeaways

• Give AI tools the background a human colleague would need before expecting strong strategic output.
• Set standing instructions that define your copilot's role, values, and coaching style, and adjust them over time.
• Upload a few core documents such as strategy decks, research, and retrospectives to create lasting shared context.
• Keep each initiative in its own chat thread so context accumulates without repeated explanation.
• Use speech-to-text to share unstructured thoughts, since rambling often surfaces details you would otherwise omit.
• Treat AI output as a prompt for your own judgment: when wrong, ask what context was missing and refine.

02Key sections

Why context unlocks AI

The author admits he initially rejected AI until a user-story task showed how much better output becomes with background and examples. Context, not clever prompting, is what makes the tools useful.

What copilots are good for

A copilot helps with strategy, brainstorming, data analysis, rehearsing hard conversations, and drafting documents. The goal is to amplify the user's own capabilities rather than outsource judgment.

Building the copilot in four steps

Set instructions to define the copilot's role and behaviors, upload company and org documents as shared knowledge, open one chat thread per initiative, and then ask focused questions. Interview prompts fill gaps in the knowledge base.

Prototyping, automations, and gossip

The copilot can help define and generate prototypes, identify event-driven automations, and absorb informal updates by speech-to-text. Each use keeps the shared context current.

Habits and future directions

Success depends on an AI mindset that takes useful outputs and discards bad ones, plus curating knowledge over time. The author also wishes for easier knowledge syncing, shared team knowledge layers, and proactive copilots.

03From the post

“One of my favorite collaborators, Tal Raviv, is back with another incredible piece that will change how many of you operate at work. If you’re looking for the promised AI productivity gains everyone’s talking about, start here. For more from Tal, check out his 63 free video tutorials on using AI agents at work, his other guest post, “PM is an unfair role. So work unfairly”, and his episode on Lenny’s Podcast. You can also book Tal for an AI build sprint with your team. Here’s my embarrassing story: 11 months ago I wasn’t using AI at work—at all. My team and I were building AI products used by tens of thousands of people. But when it came to using AI in my own job, I was a proud Luddite. I didn’t want to sound like “the average of the internet.” I was worried that if I let AI tools do things for me, I’d lose my edge. When I tried using ChatGPT, I found it disappointing for strategic and innovative work—like consulting a chatty Wikipedia. Deep down, I felt…”

04Frameworks mentioned

Summary and takeaways written by PM Atlas; quotes are short excerpts. © the original author.