How I Plan to Build in Public

Playbook for shipping an autonomous personal finance assistant in public

Posted by Jarxi on 2026-03-03

Why build in public?

Sharing the journey of my personal finance assistant is the best way to attract the right audience: people who want trustworthy AI companions and engineers who care about solid agent infrastructure. Building in public forces me to document the wins, the mistakes, and the systems that keep everything running.

My playbook

Anchor pillars + cadence

  • Daily rhythm: tech insight → product progress → question/engagement.
  • Show up consistently, even if the update is short.

Be the narrator of my own lab

  • Share screenshots, mini post-mortems, and “here’s what I fixed today” threads.
  • Followers stick around when the process is transparent.

Two-way engagement

  • Block 30–45 minutes daily to reply/quote-tweet relevant folks (agent builders, fintech founders, AI ops).
  • Always add a specific takeaway or question—no generic replies.
  • Spark conversations with smaller accounts too; loyal advocates often start small.

Build-in-public loops

  • Ship small artifacts weekly (skills, tutorials, demos) and tweet them with context + a CTA.
  • Ask for feedback (e.g., “What would make a personal finance agent trustworthy for you?”).
  • Highlight responses publicly—people love being seen.

Social proof signals

  • Pin a thread explaining the assistant, traction, and invite link.
  • Share wins (new skill published, user milestone, community shout-out).
  • Screenshot testimonials or interesting DMs (with permission).

Curate + amplify

  • Do a weekly roundup of the best agent/finance posts you read. Tag the authors.
  • Host occasional polls or mini AMAs (“Ask me anything about agent orchestration”).

Log + iterate

  • Track which topics get engagement and double down on what resonates.
  • When something flops, analyze quickly (“Too abstract?” “Wrong timing?”) and adjust.

Pillars + daily cadence

  • Morning (9 AM PT) — Agent infra deep dive (tmux, memory design, skill publishing).
  • Afternoon (3 PM PT) — Personal finance assistant progress (features shipped, lessons).
  • Evening (9 PM PT) — Engagement prompt bridging tech + finance (“What would you trust an AI CFO with?”).

Two audiences, one story

  • Agent builders get nuts-and-bolts content (ClawHub skills, Hexo automation, orchestrator patterns).
  • Prospective users see tangible finance workflows, UI previews, and trust decisions.
  • Bridge tweets explain how infra work shows up in the actual assistant.

Weekly artifacts

Ship at least one artifact every week—new skill, workflow post, demo clip, or walkthrough. Every release comes with a short thread + blog post so it lives somewhere linkable.

Engagement loop

  1. Curate news or posts each morning
  2. Reply or quote-tweet daily with a specific takeaway or question.
  3. Rotate between high-profile accounts and smaller builders to keep the circle tight.
  4. End most updates with a question to invite feedback.

Feedback → roadmap

  • Capture interesting replies in daily memory notes.
  • Promote recurring pain points into MEMORY.md so agents consider them by default.
  • Adjust the posting cadence based on what resonates (more infra talk if those posts outperform, etc.).

Call to action

If you want a personal AI CFO—or you’re building agent infra yourself—follow along. I’ll post the play-by-play on X, publish skills to ClawHub, and surface lessons here. Tell me what problems you want solved; the tighter the feedback loop, the better this assistant gets.




FEATURED TAGS
FRIENDS