Volume and Velocity

Volume and Velocity

Volume and Velocity turns rough markdown notes, links, and updates into fast social posts using AI skills.

This folder contains the MVP design and a first prototype skill for generating publish-ready social drafts inside an AI session.

This is a process and skills need, not a product and UI need. The core building blocks are:

  • a lightweight markdown input
  • AI skills that guide polishing and publishing
  • reference materials for tone, products, and themes
  • optional scripts / APIs / CLIs for channel publishing

Getting Started

To try the current prototype:

  1. Start an AI session in a tool that supports skills, such as Codex CLI, Claude Code, or Claude Desktop with a compatible skills setup.
  2. Open this repo and choose one file from volume-and-velocity/examples/, or prepare your own rough markdown note.
  3. Load the skill at volume-and-velocity/skills/fast-social-orchestrator/SKILL.md.
  4. Give the skill the markdown input and ask it to turn it into fast social posts.
  5. Load all supporting context from volume-and-velocity/context/ by default, then add any relevant product or strategic docs on top.

The default fast path is:

  • produce an X post
  • produce a LinkedIn post
  • optionally produce a Bluesky variant
  • ask at most one clarifying question if a critical fact is missing

For the current prototype workflow, see volume-and-velocity/skills/README.md.

Design Vision

The full design vision has three layers:

  1. Capture Raw items get written into markdown by hand, or eventually arrive from sources like Discord, GitHub, voice notes, and YouTube links.

  2. Polish An AI skill reads the raw item, uses Datopian context and tone guidance, and produces channel-appropriate content quickly.

  3. Publish The same workflow either publishes directly through configured tooling or produces publish-ready output with a very small final approval step.

The intended experience is AI-native and conversational: the operator stays inside one thread, the skill asks only a few high-value questions, and the workflow is optimized for shipping rather than over-editing.

Core Principles

  • Speed first: default path should target draft-to-post in under 5 minutes, ideally under 3 minutes before final publish.
  • Publish, not perfect: bias toward short, good-enough posts over polished-but-never-shipped content.
  • Human in the loop: approval remains the default at first, but with minimal friction.
  • Markdown in, content out: the starting point is a simple markdown entry, not an application inbox.
  • Skills over software: prefer reusable AI skills, prompt structure, and lightweight tooling over building a custom app.
  • Context-aware by default: product voice, stream framing, and audience guidance should come from reusable reference docs, not repeated questioning.

End-State Flow

The fuller flow we want to support over time looks like this:

  1. A raw markdown item is created manually, or captured from another source.
  2. An orchestration skill classifies the item and chooses the right flow.
  3. The skill drafts social variants and, when requested, longer-form derivatives.
  4. The operator reviews quickly and makes small edits if needed.
  5. A publishing step posts to selected channels.
  6. The workflow records what was published so we can learn what performs.

This end-state may also support richer branches such as article drafting, visual briefs, or YouTube-related flows, but those are not the initial focus.

Prototype Stages

Stage 1: Fast Social MVP

Focus: go from a raw markdown item to publish-ready social posts extremely fast.

Scope:

  • input is manual markdown
  • default outputs are X, LinkedIn, and optionally Bluesky
  • the skill asks at most one clarifying question unless critical information is missing
  • the main experience is polishing short social content fast
  • publishing is part of the target flow, even if some publishing steps are initially semi-manual

Success criteria:

  • a usable post is produced in under 5 minutes
  • the ideal path is close to 3 minutes
  • people actually use it because it feels lighter than posting manually

Stage 2: Fast Publish

Focus: reduce the final friction from approved post to live post.

Scope:

  • connect channel-specific publishing tools, APIs, or CLIs
  • support credentials and account selection cleanly
  • keep the publish step inside the same AI-guided workflow where possible

Success criteria:

  • from approved text to posted content is near-instant
  • minimal context switching or browser use

Stage 3: Expand Mode

Focus: allow richer outputs without slowing down the default flow.

Scope:

  • optional short article or blog-stub generation
  • optional visual brief or image-generation prompt
  • optional campaign bundles or YouTube-related branches

Principle:

The richer flow must stay optional. It should never make the fast social path feel heavy.

Proposed Skill Shape

The likely MVP shape is one top-level orchestration skill with explicit modes:

  • Fast: social only, minimal questions, shipping bias
  • Expand: social plus short article/source draft
  • Campaign: richer bundle with visuals or extended packaging

The default should be Fast.

Over time, this can evolve into a top-level orchestrator plus smaller reusable subskills for polishing, visual generation, and publishing.

Open Design Questions

  • What reference docs do we need for tone, product context, and audience framing?
  • Which channels are essential in v1: X, LinkedIn, Bluesky, or others?
  • How much publishing can be automated reliably in the first prototype?
  • What is the right output format for raw inputs and generated outputs?
  • How should the workflow track what has already been published?
Edit this page
Built with LogoFlowershow