Claude Code as Product Manager: How We Built a 30-60-90 Day Backlog in One Session

Meta Description: We had no backlog and a vague direction. 90 minutes later: 5 projects, 19 tickets, and a product launch plan — here’s the CLEAR prompt that started it.

We had five product areas, a goal of $10k/month, 10-15 hours per week to work with, and no clear 30-day plan.

We sat down with Claude Code, used a structured CLEAR prompt to brief it like a senior product advisor, and walked out with a complete 30-60-90 day backlog built into Linear — projects set up, labels created, custom views configured, and 19 prioritized issues ready to work.

This is what that session looked like, and how we structured the prompt that made it work.

The Problem With Vague Planning Sessions

Most planning sessions fail in the same way: you go in with too many ideas and come out with a list of notes you never look at again. No prioritization, no dates, no structure.

The instinct is to do a brain dump and call it a backlog. But a brain dump is not a backlog. A backlog is prioritized, assigned to a project, tied to a milestone, and has a due date someone is accountable to.

Getting from brain dump to structured backlog usually requires a product manager — someone who asks the right questions, pushes back on scope, and translates fuzzy direction into discrete, actionable tickets.

We used Claude Code for that role.

The CLEAR Prompt That Started Everything

The CLEAR method is ABT’s framework for writing prompts that produce specific, expert-level output. The five elements — Context, Length, Examples, Audience, Role — prevent AI from defaulting to generic.

Here is the exact structure we used to open the session:

C — Context:

I’m Jared Little, founder of Alien Brain Trust Labs. My goal is $10k/month in revenue, working 10-15 hours/week. I use Linear as my primary project tracker and GitHub as source of truth. My 5 product areas are: SAPB course, workshops, blog/content, brand, and infrastructure.

L — Length:

At the end of this session I want agreed projects, milestones, and a starter backlog of tickets created in Linear.

E — Examples:

Reference what we built for the SAPB Linear project — projects, labels, milestones — and mirror that structure for the ABT team.

A — Audience:

This is a planning session for me. I need specific recommendations, not options. Lead with what to do.

R — Role:

Act as a senior product manager who understands revenue constraints, time-constrained founders, and product-led growth. Push back on scope creep. Connect every deliverable back to the primary goal.

That prompt gave Claude Code everything it needed to do more than generate content — it gave it enough context to make judgment calls.

What Claude Code Did Before Suggesting Anything

This is the part that matters.

Before making a single recommendation, Claude Code:

1. Queried the Linear ABT team via GraphQL to see exactly what existed (5 default onboarding tickets — nothing useful)
2. Read the repo’s core strategy documents (abt-core-project-instructions.md, abt-workflow-strategy.md)
3. Checked the existing content calendar and blog archive (80+ posts, calendar through W15)
4. Read the CLEAR method reference document and course structure

It built a picture of where we actually were before telling us where to go. That’s what a good PM does. Most AI tools skip that step and hallucinate context.

The Recommendations That Came Back

After reviewing the repo, Claude Code surfaced three things before touching Linear:

First: the honest prioritization. Given 10-15 hours/week and a $10k/month goal, only three workstreams matter in the next 30 days: register the LLC (blocking revenue), launch the AI Builder Sprint workshop (the direct path to cash), and close the SAPB Lab with a real CTA (the funnel). Everything else — AI-1001, brand work, AI-2001 — went to backlog.

Second: scope discipline. When the workshop concept started expanding (“what if we made it 60 days, added a community, awards, scholarships, team tracks…”), Claude Code flagged it: that’s a bigger time commitment than 10-15 hours/week can absorb. It proposed a refined structure that kept the ambition while fitting the constraint — and explained exactly why.

Third: revenue projection before committing. Before creating a single ticket for the AI Builder Sprint, we ran the numbers across three cohort scenarios. The math supported the direction. That’s what turns a good idea into a decision.

The 5-Project Structure That Emerged

AI Builder Sprint     → Lead product, target Jun 18 (end of cohort 1)
SAPB Lab              → Lead gen, target Apr 30
AI-1001 Course        → Backlog only — no action this cycle
Blog & Content        → Calendar refresh, Sprint announcement
Ops & Infra           → LLC/EIN, AWS baseline, Linear housekeeping

Each project got a color, a target date, and a description that explains what “done” looks like — not just what the project is.

What Made This Work

The CLEAR prompt did three things that a vague request can’t do:

It defined the constraint up front. “10-15 hours per week” is not a detail — it is the filter through which every recommendation has to pass. Without it, Claude Code would have produced a comprehensive plan that is impossible to execute alone.

It assigned a role with an opinion. “Push back on scope creep” is an instruction most people forget to give. Without it, AI is agreeable by default. With it, you get a thinking partner that tells you when you’re overcommitting.

It specified the output format. “Tickets created in Linear” is not the same as “a list of things to do.” Specifying the artifact forces the session to end with something real, not just a conversation.

The CLEAR method applies whether you’re writing a marketing email or running a planning session. The principle is the same: vague input produces vague output. Structured input produces structured output you can act on.

Next in This Series

In the next post: how we connected Claude Code directly to Linear’s GraphQL API — no plugin, no integration layer — and created 19 issues, 5 projects, and 6 custom views with a script we wrote in the same session.

---

The CLEAR method is part of ABT’s AI-1001 course curriculum. Full reference at alienbraintrust.ai.