Yesterday's Top Launches: 3 Tools from November 20, 2026
Three creator-oriented tools launched on November 20, 2026, each condensing a multi-tab workflow into a single MCP connector, phone app, or browser page, with an honest first look rather than rankings.

Three products landed on November 20, 2026, and they share a quiet theme: each one collapses a workflow that normally sprawls across half a dozen tabs into a single surface. The surfaces differ though. One is an MCP connector, one is a phone app, one is a browser page. And none of the three came with a community ranking yet, so there is no vote tally to point at. What follows is closer to a friend’s honest read than a leaderboard.
If your feed is mostly new developer tools and agent plumbing, this batch sits nearer to creator work, but each product leans on infrastructure you may already be paying for.
SlidesLand
TikTok slideshows look cheap to make and are not. The good ones read like a person built them, which is exactly where most AI slideshow generators fall apart: stock-looking photos, the same template every time, captions running across someone’s face.
SlidesLand takes a different route. It is not really a site you visit. It is a connector at https://slides.land/mcp that you add to Claude, ChatGPT, Gemini, Cursor, Copilot or any other MCP-capable client, then ask in plain language for a slideshow. The agent writes the slides using one of 13 formats that already performed (Comments, Chat, Viral news, Ragebait GF, a couple of outfit layouts), searches photos slide by slide, renders each one, and looks at its own output before you do. In the demo on the site, the agent notices a soft photo on slide two and swaps it for a sharper one without being asked.
That self-check is the actual argument. AI writing inside an agent’s reasoning loop can spot a bad crop. A form-based generator cannot. SlidesLand also exposes the whole production chain as tools: replace_picture, find_people for keeping text off faces, identify_font for matching lettering from a screenshot, schedule_post, get_post_stats. Revisions are a sentence. “New photo on slide 3” moves slide 3’s photo and nothing else.
Worth weighing honestly: the site’s $193-a-month comparison against an eight-tool stack is marketing arithmetic, though the direction is right. The bigger catch is that the best version of this product requires an MCP client and a comfort with connectors. The web studio exists for people who would rather click, and SlidesLand itself admits it is a step behind the agent on quality.
Pricing runs $49 for 100 slideshows a month, $95 for 250, $195 for 750, with five free to start.
Runvia
The pitch from indie founder Fabio Barbieri is refreshingly specific: your watch records runs brilliantly and then tells you nothing about what to do next. Runners end up downloading a generic plan and quietly abandoning it two weeks in.
Runvia puts a coach named Via inside the app. You describe the race, the days you can actually train, road or trail, and Via proposes structured workouts onto a calendar. Miss a session and it rebuilds the rest of the week rather than leaving a dead plan rotting on the screen. The Today screen combines check-in, readiness and a race countdown, and the season view shows blocks building toward race day. Distance range runs 10K through ultras.
Running works without a watch. Phone GPS plus voice cues plus an SOS alert to a contact covers the basics, and the Wear OS companion is free. Post-run analysis gives map, pace, heart rate, elevation and written insights. There is also a strength builder with 90+ exercises.
The design decision worth flagging is bring-your-own-key. After the trial, Via runs on your own Google Gemini API key. That keeps chat history on your phone and no proprietary cloud in the middle, which privacy-minded runners will like. It also means creating a Google AI Studio key and pasting it in, which is genuine friction for anyone non-technical. The coach stays free with no mandatory subscription; Premium adds deeper race planning, Garmin/Suunto/Apple Health sync and load insights. Strava and FIT/GPX/TCX import sit on the free path.
Cliplato
Twitch streams run for hours. The clips that matter are buried in chat reactions, loudness spikes and the moments other viewers already clipped. Cliplato reads those signals, cuts 9:16 MP4s at 1080x1920, burns in word-by-word captions, finds the facecam scene by scene, and writes the TikTok caption and YouTube description in the video’s own language.
Paste a link from Twitch, Kick, YouTube, Dailymotion, Vimeo, Rumble, Facebook, X, Bilibili or Odysee, or upload up to 4 GB. A job page walks through Listen, Moments, Transcribe, Pick, Webcam, Render, Review, and you can close the tab while it finishes. Pro adds autopilot that watches a channel and clips every new stream unattended.
The credit model is the sharpest thing here. One credit clips one video regardless of length, so a six-hour stream costs the same as a twenty-minute clip. That contrasts with competitors metering uploaded minutes, where a single long stream can eat a month.
Two limits. Uploaded files have no chat replay, so moment detection falls back to sound and speech alone, which is weaker. And the 4 GB cap plus login-walled sources will bite some users. Pricing starts free at 3 credits a month, then $8.99, $13.99 and $49.88 for 20, 60 and 250 credits.