01What Is Google Jitro?

Google Jitro is a new AI coding agent that works without prompts. Instead of you telling it step by step what to do, you set a goal and it figures out what changes are needed. Then it works on its own.

It is reportedly the next version of Google's Jules, its existing AI coding agent. Jitro is not officially announced yet, but internal signals and reporting point to a launch soon, possibly at Google I/O on May 19, 2026.

02How Jitro Is Different from Current AI Tools

Most AI coding tools like Cursor, GitHub Copilot, and OpenAI's Codex agent work on a task-first model. You define the specific task, like 'fix this bug' or 'write this function.' You are still doing the thinking.

Jitro flips that. You say something like 'I want our error rate down' and it goes and figures out what to change. This is called KPI-driven development.

You set a metric, and the agent identifies obstacles and works on them.

Key differences include:

  • Runs asynchronously – it works in the background while you do other things.
  • Uses a persistent workspace – it remembers context and builds on what it knows.
  • Goal-based, not prompt-based – you set outcomes, not tasks.

03Setting Up Google Jitro: What to Expect

Since Jitro is not officially launched yet, there is no direct setup guide. However, based on the episode, here is what to expect:

  1. Join the waitlist when it becomes available – the episode mentions a wait list is expected.
  2. Look for Jitro inside Google AI Pro and Ultra subscriptions, as Jules is already there.
  3. Once available, set up a dedicated persistent workspace for your goals.
  4. List your goals and configure which tools the agent can use.
  5. Review the agent's approach and approve the direction before it works on its own.

04How to Use Jitro: Walking Through an Example

In the episode, Julian uses an agency example. You run an agency and want your client's website to generate more leads. With Jitro, you set that goal – 'more qualified leads from this page.'

The agent then identifies what is blocking that goal. It might find that the form is buried, the headline is weak, or the page loads slow on mobile. It flags these issues and proposes changes.

You review the plan and give sign-off.

Key steps in using Jitro:

  • Set a clear, measurable goal.
  • Let the agent analyze the codebase and propose changes.
  • Review the agent's reasoning and approve the plan.
  • Let it run autonomously in the background.
  • Monitor the results against your KPI.

This moves you from managing tasks to managing outcomes.

05Results and Comparisons from the Episode

The episode focuses on Jitro's predecessor, Jules, which already shows real results. During its beta, Jules helped developers make over 140,000 code improvements. It is now out of beta and available in Google AI Pro and Ultra subscriptions.

Julian compares Jitro to competitors:

  • GitHub Copilot – strong for task completion but still task-first.
  • Cursor – excellent but synchronous; you must wait for each output.
  • OpenAI Codex agent – moving toward agentic but still task-based.

Jitro is described as a different product category – goal-driven, persistent, and asynchronous.

06Tips, Limits, and Mistakes to Avoid

Jitro is not a fire-and-forget system. It keeps humans in the loop for goal approval and direction review. This is a design choice, not a limitation, to build trust.

Avoid these mistakes:

  • Treating Jitro like a prompt-based tool – start thinking in outcomes, not tasks.
  • Ignoring reasoning transparency – choose tools that show why they made decisions.
  • Not reviewing the agent's plan – always review before approving.

Julian also warns that the person who just figures out steps will be automated. The future belongs to those who frame outcomes and own the results.

07How to Prepare for Goal-Driven AI Agents

Even before Jitro launches, you can start preparing. Julian gives four practical steps:

  1. Understand the difference between prompting and goal setting – start saying 'I want X to happen, here's what good looks like.'
  2. Pay attention to reasoning transparency in tools – reward those that show their logic.
  3. If using Jules, get comfortable reviewing a plan before approving changes.
  4. Map out which parts of your business have clear measurable outcomes – conversion, turnaround, error rate, leads.

This mental shift works with every AI tool you use today and will matter more as tools get more capable.

08Who Should Care About Jitro and Why It Matters

Jitro matters even if you never write code. If you run a business, agency, or online store, you work with developers or technical tools. The shift from prompt to goal is coming to every workflow.

Julian emphasizes that the people who stay ahead are not the best at prompts but those who understand how to set good goals, frame outcomes, and review AI reasoning. These are human skills that become more valuable.

The episode also mentions the AI Profit Board, a community of 2,800 business owners using AI agents for lead gen and client delivery. If you want to get ahead, joining such a community can help.