Skip to content

By playing, you agree to uListen's Terms and Privacy Policy and YouTube's Terms of Service.

How I AIHow I AI

How I run autonomous coding agents from my phone with OpenAI Symphony + Linear

Alessio Fanelli, founder of Kernel Labs and co-host of Latent Space podcast, walks us through two very different AI workflows: (1) a fully autonomous coding setup using OpenAI Symphony + Linear, where Linear acts as a state machine and…

Claire VohostAlessio Fanelliguest
Jul 6, 202635mWatch on YouTube

Summaries, chapters, transcripts, and FAQs on this page are made by uListen, not YouTube.

Great conversations run long. Their best ideas don't have to.

uListen is free on iPhone, iPad, Android, Chrome, and Claude.

See how

At a glance

WHAT IT’S REALLY ABOUT

Running cloud coding agents via Symphony, Linear, and mobile management workflows

  1. Alessio explains the shift from being an “agent prompter” to an “agent manager,” using Linear as the state machine and source of truth for long-running autonomous coding work.
  2. He demos a Symphony+Linear workflow that turns issues into Codex workpads, produces PRs, routes tasks through human review/rework, and can be managed from anywhere (including a phone).
  3. They discuss why orchestration is less the hard part than tool support and context-shaping—e.g., adding capabilities like visual testing/screenshot diffs and tracking token usage per task to diagnose friction.
  4. Alessio highlights maintenance pitfalls of “skills” and markdown instruction files, arguing for periodic purges and tight, minimal instructions to prevent compounding agent confusion.
  5. A second demo shows Codex doing business-value automation for a trading card shop: extracting PSA certificate numbers, hunting underpriced high-end cards on eBay in controlled batches, and envisioning real-time trade-show pricing assistance.

IDEAS WORTH REMEMBERING

5 ideas

Treat yourself as an agent manager, not a prompt typist.

Alessio’s core change is moving from direct back-and-forth prompting to managing a queue of issues and interventions, letting the agent execute plans and only stepping in at review/rework points.

Use Linear (or an equivalent) as the source-of-truth state machine.

By encoding status transitions (To Do → In Progress → Human Review → Rework → Done) in Linear, you get a durable workflow that’s easy to operate on desktop or mobile and keeps work organized across many runs.

Cloud-hosted runtimes make autonomous work actually usable day-to-day.

Moving agents off local machines to a VPS enables persistent execution and multi-channel control (phone, web, shell), removing the friction of babysitting hardware and sessions.

The real value is task history + context shaping, not “new capability.”

Symphony doesn’t magically add intelligence beyond the coding agent; it structures context (spec, plan, acceptance criteria, rework checklist, PR discussion) so you can understand failures and iterate systematically.

Track token spend per task to estimate cost and uncover bottlenecks.

A token ledger lets you see which tasks explode in cost (e.g., deployability refactors) and guides improvements in specs, checks, and tooling to reduce wasted cycles.

WORDS WORTH SAVING

5 quotes

What really clicked for me was, like, starting to move away from being a agent prompter to kinda be a agent manager.

— Alessio Fanelli

If you expect something... You should start to have, at some point, some idea of, like, how many tokens you think this will take. You know, is this like a 10 million token task? Is this like 100 million token task? And if the reality is very far away from your expectations, there's probably something in the tooling layer that you can do to improve.

— Alessio Fanelli

Everything is like, how do you shape the context, you know? Like Symphony is just a way to shape the context. It's not, it's not giving you any new capability that you wouldn't have by using the coding agents directly. It's just helping you wield it.

— Alessio Fanelli

This is my favorite positive outcome of AI, which is small business creation. Just the ability to, like, intersect the human world in a way that has been historically very inefficient has been a quality of life improvement for me.

— Claire Vo

You can actually use AI to save clock time for real people by doing these things autonomously.

— Alessio Fanelli

Read the available transcript
Symphony as an agent orchestration loopLinear as workflow/state machineCloud VPS setup for always-on agentsWorkpads, PR review, and rework loopsToken usage ledger and cost estimationSkills/markdown file drift and “purging”AI automation for inventory/market hunting (Pokémon cards)

Ask ChatGPT about this episode

Get its key ideas, plus a summary of each chapter linked to that point in the video, right in your chat.

  1. Add uListen AI to ChatGPT.

  2. Paste this prompt into a new chat.

    Use uListen AI to give me this episode's key ideas and a chapter-by-chapter summary, with timestamps: https://www.youtube.com/watch?v=KtmaWUVdnx4

SPEAKERS

  • Claire Vo

    host

    Product leader and host of the "How I AI" podcast.

  • Alessio Fanelli

    guest

    Founder of Kernel Labs and co-host of the Latent Space podcast.

High quality AI-generated summary created from speaker-labeled transcript.

Get more out of YouTube videos.

Read along on your phone, search it beside the video, or ask your favorite AI app what was said.