Skip to content
UGConic
All work
AI AUTOMATION · CONTENT SYSTEMS

From AI News to Ready-to-Publish LinkedIn Content.

LinkedIn Post Studio streamlines the daily content workflow by combining news discovery, AI-assisted writing, branded image generation, and browser-based publishing into one local dashboard.

LinkedIn Post Studio dashboard showing generated post drafts
The challenge

Consistent LinkedIn publishing is a daily grind

Thought-leadership content means repeating the same research, writing, and design work every single day.

  • 1

    Finding relevant, recent AI news worth posting about.

  • 2

    Developing original angles for LinkedIn.

  • 3

    Writing posts that keep a consistent personal voice.

  • 4

    Creating branded visuals for each post.

  • 5

    Preparing and publishing selected content.

  • 6

    Maintaining a consistent posting routine.

The objective is to reduce this repetitive work — not to replace the human decision about what finally gets published.

The solution

One local studio, eight drafts per run

LinkedIn Post Studio brings research, writing, and design into a single local application.

Dashboard of generated LinkedIn drafts

Each run targets eight draft posts — four focused on GoHighLevel and four on major AI companies or developments in current news. Each draft can include a caption, hashtags, a suggested first comment, a short visual headline, and image direction.

  • Inspect, edit, and select posts in a local dashboard
  • Copy text and browse previous daily outputs
  • Prepare selected posts in LinkedIn to review and publish manually

Eight is the intended output per generation run — not a guarantee of successful output under every service or network condition.

Branded visuals generated per post

  • Generated LinkedIn post about AI employee and conversation AI
  • Generated LinkedIn post about personalized voicemail automation
  • Generated LinkedIn post about sentiment-driven offers
  • Generated LinkedIn post about AI music labeling

Actual generated post cards from the studio. Review and fact-check content before publishing.

The workflow

Discover → Develop → Write → Visualize → Review → Prepare → Save

A single run moves from fresh news to reviewed, ready-to-publish content — with you in the loop.

  1. 01

    Discover

    Read configured RSS feeds and identify recent AI-related news.

  2. 02

    Develop Angles

    Organize topics into the two intended content categories.

  3. 03

    Write

    Draft content following persona, tone, structure, and formatting rules.

  4. 04

    Generate Visuals

    Create imagery, add logos where appropriate, and apply branded headlines.

  5. 05

    Review

    Inspect drafts and images in the dashboard and select preferred posts.

  6. 06

    Prepare to Publish

    Prepare selected content in LinkedIn — you complete publication.

  7. 07

    Save & Revisit

    Store daily content locally so outputs can be reviewed later.

Capabilities

What the studio does

  • News Aggregation

    Reads configured RSS feeds and filters recent articles for relevant AI topics.

  • AI-Assisted Writing

    Generates ideas and full drafts guided by a defined persona and style rules.

  • Branded Visual Generation

    Combines imagery with company logos, headlines, gradients, and consistent styling.

  • Content Dashboard

    Review posts, edit captions, select favorites, copy text, and browse previous days.

  • Human-Reviewed Workflow

    Prepares selected posts in LinkedIn via Playwright; you review and publish.

  • Resilient Generation

    Retries, timeouts, provider fallbacks, and fallback image compositions.

  • Local Content Archive

    Daily text, images, and structured data in dated folders — no database required.

  • Portable Windows Setup

    Bundled Node.js runtime, Chromium, launch scripts, and config templates.

Architecture

How the pieces fit together

RSS feeds flow through filtering, AI writing, and image composition into a local archive and review dashboard — with provider alternatives as conditional paths, not every provider on every run.

Local review dashboard

A local Node.js dashboard where you inspect, edit, select, and copy generated posts.

  1. RSS Feedsrss-parser ingestion
  2. News FilterRecent AI topics
  3. AI WritingOpenRouter (+ fallback)
  4. Branded VisualsGemini (browser) + sharp
  5. Local ArchiveFiles & JSON
  6. LinkedIn PrepPlaywright + Chromium
  7. You PublishHuman approval

Provider alternatives (e.g. Pollinations for text, fallback image compositions) are used conditionally. Sessions and API keys stay private and local.

Technology stack

  • Node.js
  • Playwright + Chromium
  • HTTP server
  • rss-parser
  • OpenRouter
  • Pollinations (fallback)
  • Gemini (browser)
  • sharp
  • dotenv
  • Local files / JSON
Decisions & constraints

Why it's built this way

  • A lightweight Node.js architecture keeps the application relatively simple.

  • OpenRouter provides an alternative text-generation route when direct Gemini API access is unavailable in the user's environment.

  • Browser automation accesses the logged-in Gemini website for image generation instead of relying only on the Gemini API.

  • Retries and fallback providers help the application recover from some service failures.

  • Local files make it straightforward to retain and revisit generated content.

  • Human approval prevents the workflow from publishing selected content without a final user action.

  • Bundled runtime and browser files reduce installation requirements for the intended Windows setup.

Accessing the Gemini website through a browser was an environment-specific implementation decision for the original deployment — not a claim that any provider’s API is universally blocked or that browser access is guaranteed to work.

How it came together

Built iteratively, problem by problem

Each stage solved the next real constraint rather than chasing a perfect design up front.

  1. 01

    Establish the basic news-to-draft pipeline.

  2. 02

    Adapt the provider architecture to the original environment's API-access constraints.

  3. 03

    Add error handling, retries, and fallback generation.

  4. 04

    Build branded image composition.

  5. 05

    Introduce a dashboard for reviewing generated content.

  6. 06

    Add the human-reviewed LinkedIn preparation workflow.

  7. 07

    Package the application with launch scripts and portable dependencies.

Claude Code was used as an AI coding assistant during development. It is not part of the runtime product.

Results & status

A working internal content studio delivering:

  • A local dashboard for reviewing generated content
  • A target of eight draft posts per generation run
  • Branded image generation and composition
  • A local archive of daily outputs
  • A browser-driven LinkedIn preparation workflow with a final human action
  • A portable Windows-oriented project structure

No client names, engagement metrics, revenue outcomes, or guaranteed generation success are claimed.

Responsible automation

  • AI-generated posts should be fact-checked before publication.
  • News relevance, attribution, image rights, and company-logo usage should be reviewed.
  • Browser-driven workflows can break when external websites change their interfaces.
  • Third-party service limits, terms, pricing, and availability can change.
  • Authentication sessions and API keys must remain private.
  • Local browser profiles and credentials must not be shared in public project assets.
  • Users should comply with LinkedIn's applicable terms and avoid spam-like publishing.

This workflow is not endorsed by LinkedIn and does not use an approved LinkedIn API.

More automation

Related work

Make consistent content easier to produce.

UGConic builds practical AI-powered workflows that connect research, content creation, and business operations.