DEV Community: nolan boxill

A scheduling workflow for solo founders: one draft, eight channels, no copy-paste chaos

nolan boxill — Mon, 26 Oct 2026 10:34:23 +0000

The solo founder scheduling problem is not writing posts. It is writing the same post eight times, in eight formats, and then finding out one channel rejected it silently. This post describes a workflow that fixes that, drawn from building Moqup, a social scheduling app at https://moqup.app. No metrics, no growth claims, just the workflow.

One source draft, per-channel adaptations

Start with one source draft, then adapt it per channel instead of copying it. Aspect ratios differ. Caption length limits differ. Link handling differs: some channels want the link in the post, others want it in the bio or a first comment. Hashtag norms differ, and a caption that reads well on one channel reads like spam on another. A good scheduler stores the source once and keeps the per-channel version beside it, so editing the source does not mean retyping eight posts.

Calendar for campaigns, queue for evergreen

Two scheduling shapes cover nearly everything. A calendar fits campaigns: launch week, event promotion, a dated announcement. A queue fits evergreen content: tips, reposts, and posts that stay true for months. Mixing them up causes the two classic failures: evergreen posts crowding out a launch, and campaign posts repeating after the date passed. Keep them in separate lanes and the calendar stays readable.

Failure handling is the unglamorous part that matters most

Tokens expire. Rate limits hit. A video gets rejected for format, or an image is the wrong size. The failure mode that hurts founders is silent failure: a post that simply never appears, discovered days later. Every scheduled post needs three states, clearly shown: scheduled, published, or failed with the reason. A failed post with its reason is a five minute fix. A failed post you never notice is a missed week.

A review step, even when you are the whole team

Solo founders skip review because there is nobody to review with. Keep a draft state anyway. Writing on Monday and reviewing on Tuesday catches typos, wrong dates, and tone problems that same-day posting never catches. The review step costs minutes and prevents the posts you end up deleting.

Automation notes: scheduling from a chat assistant

Moqup exposes its scheduling through an MCP server, and its ChatGPT plugin is published, which means drafting and scheduling can start from a chat assistant instead of a dashboard. The practical value is not novelty. It is that a founder can turn a conversation into a scheduled draft without opening another tab, and the draft still passes through the same review and per-channel adaptation steps above. Automation should shorten the path to a draft, not skip the review.

One honest note on channel availability

Publishing availability varies by channel and by platform approval status. For Moqup, Instagram, TikTok, and YouTube publishing run through approved connections, while Threads and Facebook publishing for public users is still going through Meta App Review, so for those two channels the workflow today is plan and draft, not publish. Any scheduler that claims every channel publishes today, for every user, without qualification, deserves skepticism. Build the workflow around draft states and the channel mix survives approval changes.

That is the workflow: one source draft, per-channel adaptations, calendar and queue kept separate, failures shown with reasons, a review step even for a team of one, and automation that feeds drafts instead of bypassing review. Moqup is at https://moqup.app if you want to try that shape of scheduling.

How AI website cloning actually works, from URL to editable code

nolan boxill — Mon, 26 Oct 2026 10:34:02 +0000

When people hear "website cloner," many picture a download button that saves a page. A real cloning tool does something different: it rebuilds a page as clean, editable code you can actually work with. This post explains the pipeline, where an AI model genuinely helps, and where it cannot help at all. I am writing from building Kloner, an AI website cloner at https://kloner.app, and everything below describes the general technique, not a secret sauce.

Step 1: render, do not just download

A static HTML download misses most of the modern web. Content loaded by JavaScript, lazy loaded images, and layout shifts after load all mean the raw HTML file is not the page the visitor sees. The reliable approach is to render the page in a headless browser, wait for the network to settle, and capture the final DOM. That rendered DOM is the source of truth for everything that follows. If a section only appears after scrolling or clicking, a single render pass will miss it, which is why thorough tools render more than the first viewport.

Step 2: capture styles, not just markup

Copying raw CSS from a production site rarely works. Production stylesheets carry unused rules, framework utility classes, minified names, and overrides stacked across files. What matters is the computed style of each element: the font, size, color, spacing, and layout values the browser actually applied. A cloning pipeline reads those computed values and writes fresh CSS from them, keeping what the eye sees and dropping the dead weight. Fonts, images, and icons are downloaded, compressed, and rehosted so the rebuilt page does not depend on the original server.

Step 3: rebuild structure as components

A flat dump of divs is technically a clone and practically useless. The useful step is mapping the DOM to components a developer recognizes: navigation, hero, feature cards, testimonials, pricing table, footer. Semantic HTML replaces nested wrapper divs where the structure allows it. Class names get rewritten to something human readable. Fixed pixel layouts get converted to responsive ones, because a clone that breaks on mobile is not a finished page.

Where the AI model actually helps

The model earns its place in three spots. First, it turns messy production DOM into clean structure, deciding where a component starts and ends when the original markup gives no hint. Second, it names things sensibly and converts inline styles and utility soup into readable classes. Third, it repairs common breakage, like a layout that only worked because of a global stylesheet the clone left behind. The model is doing structured cleanup, not magic.

Where it cannot help

An AI cloner cannot recover what the browser never received. Backend logic, databases, payment flows, user accounts, and anything behind a login wall are not in the page and cannot be cloned from it. Canvas and WebGL content, heavily animated scenes, and pages behind serious bot protection are partial at best. If a tool promises a pixel perfect copy of any site including its backend, that promise is not real.

A practical checklist for a clean clone

Start from a site you own or have permission to rebuild. Audit the sections before you clone, and decide what to keep, rewrite, or drop. Replace placeholder copy with your own text rather than shipping someone else's words. Test responsive breakpoints, because desktop perfect and mobile broken is the most common cloning failure. Check performance after rebuild, since unoptimized images are usually the biggest cost. Then treat the clone as a starting point to edit, not a finished product.

That is the whole technique: render the real page, capture computed styles, rebuild as components, and let the model handle cleanup while staying honest about what cannot be copied. If you want to try it, Kloner is at https://kloner.app, and there is a longer walkthrough at https://kloner.app/blog/how-to-clone-a-website.

I Built an AI Social Media Manager for Solo Founders (No Agency Required)

nolan boxill — Thu, 08 Oct 2026 12:35:23 +0000

Every solo founder knows the drill: you're building the product, answering support emails, jumping on sales calls, all while you're supposed to be "creating content" every single day. Hire an agency? That's $2,000+ a month most of us don't have. Post randomly when you remember? That's how social accounts go silent for six weeks.

I built Moqup to kill that dilemma. It's an AI social media manager designed specifically for solo founders and small businesses with small budgets.

It learns your brand voice from your website

The part I'm most proud of: Moqup studies your website and learns how you talk: your tone, your positioning, your audience. No generic AI slop. Every caption and script sounds like you wrote it, because it's trained on how you already write.

UGC-style videos, generated for you

Short-form video is where the attention is, but filming yourself every day is exhausting. Moqup generates UGC-style videos for your brand automatically - the kind of authentic, native-feeling content that actually performs on TikTok, Reels, and Shorts.

One calendar for 8 channels

Plan everything on a drag-and-drop content calendar, then schedule across 8 channels: Instagram, TikTok, YouTube, X, Threads, Facebook, and more. Write it once, publish it everywhere, without living inside six different apps.

Review ideas like swiping on Tinder

Content review is usually the boring bottleneck. Moqup turns it into a quick swipe-through: like the ideas you love, skip the ones you don't. A month of content reviewed in minutes, from your couch.

All your comments in one inbox

Engagement matters more than posting, but checking comments across 8 platforms is chaos. Moqup pulls every comment into a single inbox so you never miss a customer question or a warm lead again.

Plays nice with your AI stack

ChatGPT MCP plugin. Moqup has a ChatGPT plugin (via MCP) that lets you manage your content conversationally right inside ChatGPT - "schedule three posts about our new feature this week" and it's done. Your AI assistant becomes your social media manager.

Why I built it

Agencies are great, if you can afford them. Most solo founders can't, and shouldn't have to. The tools changed: AI can now do the heavy lifting of ideation, scripting, production, and scheduling. Moqup exists so a one-person company can have the social presence of a team of five, for a fraction of one agency retainer.

If you're a solo founder doing your own marketing, check out Moqup and tell me what you think. What would you automate first?