SlidesLand is a slideshow tool built for AI agents. Add it to Claude, ChatGPT, Gemini, Cursor, Copilot or any other app that supports MCP as a connector, then ask in plain words for a TikTok or Instagram slideshow. Your agent writes the slides; SlidesLand supplies 13 proven formats, searches for real photos slide by slide, lays out and renders every slide, checks each one, and can queue the finished post on your connected accounts. It is aimed at creators, social media managers and small teams who publish slideshows regularly and want the whole loop — writing, pictures, design, checking, scheduling and measurement — to happen inside one chat rather than across a stack of separate apps. A web app and studio are also available for people who would rather click.
Slideshow posts are everywhere on TikTok, and the ones that work share one thing: they look like a person made them. Most AI slideshow tools miss exactly that. They return stock-looking photos, the same template every time, and words running across someone's face. The alternative most creators fall back on is a stack of separate subscriptions. SlidesLand's own comparison prices that stack at roughly $193 a month across eight logins: an AI writing tool at $20, stock photos at $29, a design app at $15, a background remover at $9, template packs at $12, a slideshow maker at $29, a scheduler at $30 and an analytics dashboard at $49. On top of the money sits the time — six or more hours a week, eight apps and user interfaces to learn, and copy-pasting between tabs all day, with the creator acting as the glue between every tool.
SlidesLand ships formats that already went viral — the site names Comments, Chat, Viral news, Who's looking, Grief confessions, Fits: mix, Fits: flat-lay and Ragebait GF among them. Each one is a proven layout that the agent fills with new words and pictures, so the structure that earned attention is reused while the content stays yours. The agent can list the available formats and open one before it starts writing, which is exposed through tools called list_formats and get_format. A second route into the same result is copying a post you have already seen perform. Paste a TikTok or Instagram link into the chat and the agent reads the slides, matches the layout and type, and rebuilds the format around your page — fetch_tiktok_slideshow and create_custom_slideshow do that work. In the chat example on the site, a request for a viral-news slideshow about today's weirdest stories, with yellow on the shock words across four slides, comes back finished with the photo on slide two swapped for a sharper one.
Pictures are searched slide by slide rather than picked once for the whole post, and they can be swapped one at a time. The renderer is the other half of the promise: every slide is rendered and shown back to the agent, which looks at its own work and fixes blurry, off-topic or text-covered pictures before the creator ever sees them. The tools behind this are preview_slideshow and render_slide. On the site's example, the agent reports that the photo on slide two came back soft and that it swapped it for a sharp one before showing the result. Because the check happens inside the same conversation, a creator does not have to move to a design app to notice a bad crop — the correction is already part of the first draft. The site describes this self-checking loop as the reason it recommends the agent over the web app, where the built-in writer produces the slideshow in one pass and the creator checks each slide by eye.
Editing is done by asking. "New photo on slide 3", "make the hook punchier" or "make slide 3 punchier" change one slide, one picture or one line, and nothing else moves; the replace_picture and update_slide tools carry the change. Background removal is available in a single message: cut a person or a product out of any photo and place it on the slide's own colour. The agent also finds where people are on each slide and keeps text clear of them, or zooms in on a face on purpose, through a find_people tool — the site's promise is that words never cover a face. Typography is handled with more than 2,000 Google Fonts for captions, and any format or layout can use them; send a picture of lettering and the agent finds the closest match, using search_fonts and identify_font. In practice this means a creator can hold the look of a page steady from post to post without hunting for the same font file.
Posting is part of the same conversation. A slideshow can be queued on several connected accounts in one message, each at its own time and its own shape — the site shows the same content going out at 9:16 on one account at 9:00 and 4:5 on the same account at 9:05, with another account scheduled the following evening. A single instruction like "one post a day at 6pm next week, mix the formats, Tuesday's news at 9am" can fill a week of the calendar. schedule_post and get_post_stats are the tools involved. Measurement follows the same pattern: ask "how did it do?" and the agent pulls the views, likes, comments, shares and saves for the last seven days, covering both posts made with SlidesLand and other recent posts on the connected accounts, so the next request can repeat what worked.
Structurally, SlidesLand is a connector rather than a destination site. The MCP address is https://slides.land/mcp, and setup means adding a custom connector in the AI app's settings, naming it SlidesLand, signing in and pressing Allow. Supported apps listed on the site include Claude on web, desktop and mobile, Claude Code, ChatGPT, Gemini CLI, Cursor, Windsurf, VS Code with GitHub Copilot, Zed, Cline and n8n; apps that only run local servers can connect through mcp-remote. The reasoning behind the design is stated plainly: the best results came from a thinking model that could look at its own work and fix it, so instead of hiding the AI behind a form, SlidesLand was built for the agent a creator already uses. The words are written by that agent, with its own model, voice and reasoning, while SlidesLand contributes formats, photos, layout, rendering and the rules learned from reviewing finished slideshows one by one.
The practical benefit is that six or more hours a week of tab-switching and eight separate logins collapse into one chat. Writing, photo sourcing, layout, background removal, font matching, scheduling and analytics all happen where the brief was given, and revisions cost a sentence instead of a session in a design editor. Because the agent inspects each rendered slide, the first draft already arrives checked. Because the analytics come back in the same thread, the next slideshow can be shaped by what the last one did — the site's example ends with the agent offering three more posts in the same shape after a post beat the account's average.
Concrete prompts on the site show the range. A creator can ask for a six-slide chat slideshow about a situationship that ended over a playlist; copy a TikTok for a skincare brand by pasting a link; plan a week with one Comments post a day at 6pm on TikTok; redo slide four with a sharper photo of the same person; find the font in a screenshot and use it on the cover; or turn a last post into part two with the same names and a new twist. Other examples given include a who's-looking slideshow with old paintings and one line per slide, a grief-confessions post written in the same voice as the previous one, a ragebait slideshow about things he said, flat-lay menswear outfits, and a viral-news post with yellow on the shock words.
The plans are aimed at different posting volumes. Growth costs $49 a month for one creator posting most days and includes 100 slideshows a month. Scale costs $95 a month for several accounts or a small team and includes 250 slideshows a month. Enterprise costs $195 a month for agencies running many brands and includes 750 slideshows a month plus priority support. Every plan includes use from Claude, ChatGPT and any MCP app, the web app and full studio, scheduling on connected accounts, analytics for your accounts, and unlimited accounts and team members. Each slideshow counts once toward the monthly allowance whether the agent or the creator made it, however many times it is edited or re-rendered. New accounts get five slideshows free with everything included and no card required. Billing runs monthly through Polar as merchant of record, and cancellation is available any time under Settings → Billing.
Two other routes exist alongside the connector. The web app and studio let a creator pick a format, write a line and edit every slide by hand, with a calendar and analytics; the site describes it as quick but a step behind the agent on quality, and slideshows made by the agent appear there for fine-tuning. For Claude Code and other agents that load skills, there is a downloadable SKILL.md that teaches the whole loop and the quality rules before the first message, and that works alongside the connector rather than instead of it. On permissions, the site states that the agent can queue posts on accounts already connected in SlidesLand but cannot connect new accounts, delete anything, or see passwords or payment details; access can be removed under Settings → AI.
SlidesLand's value proposition is narrow and clear: one chat for the whole slideshow, start to stats. It hands an AI agent the formats, photos, renderer, checker and scheduler it needs to produce a social slideshow that looks like a person made it, and it leaves writing and judgement to the model the creator already trusts.