Do More Newsletter

Stay ahead with the most recent breakthroughs—here’s what’s new and making waves in AI-powered productivity from the past week:

• Clockdiary: The latest buzz in AI-driven time tracking and productivity management. Clockdiary offers automatic time logging, activity tracking, and AI insights for both individuals and teams, helping optimize every minute of your workday and providing managers with robust oversight tools.

• Sublime: A fresh take on organizing your saved ideas, quotes, and highlights, Sublime employs AI to declutter and intelligently categorize your notes, making retrieval and action seamless. Its centralized system appeals to anyone looking to bring structure to creative chaos.

• Bluebird: This new AI-powered time tracker promotes focus by encouraging structured intervals and synchronizing tasks across Apple devices. Ideal for maintaining consistency and maximizing productivity, especially for freelancers and remote workers.

• Raycast Mobile: Bringing the popular productivity desktop tool to mobile devices, Raycast now supports AI snippets, text analysis, and centralized note management to support busy professionals on the move.

• Craft: Not just another note app, Craft introduces an aesthetic, AI-assisted workspace designed for organizing client work, collaborative mood boards, and shared documents, all with a uniquely visual twist.

Clockdiary

AI-Powered Time Tracking Redefined

In a world full of distractions and shifting priorities, managing your time accurately is key to doing more. Clockdiary—making headlines this week—reimagines how professionals and teams track, analyze, and leverage time through an AI-first approach.

What Sets Clockdiary Apart?

• Auto Time Tracking: Start the automatic timer and Clockdiary does the rest—logging your activities across projects with single-click accuracy.

• Activity Insights: The app’s built-in tracker periodically captures app usage and categorizes time entries. Its AI “learns” your work patterns over time, surfacing gaps and opportunities for productivity gains.

• Smart Prompts: Receive AI-driven reminders, time-overrun alerts, and categorization suggestions so you never miss logging a minute or misunderstanding where your time goes.

• Team Productivity Engine: Dynamic rule-based settings for managers enable real-time oversight without micromanagement. Admins can monitor staff activity, idle behavior, and project alignment for true operational clarity.

• Intuitive Payroll & Project Integration: Precise, automated timesheets mean faster payroll processing and streamlined project accounting.

Who Should Try Clockdiary?

• Remote workers needing a non-intrusive, reliable time tracker.

• Teams or managers seeking data-driven workload management.

• Founders and freelancers wanting to pinpoint productivity leaks.

Why It’s Generating Buzz

This week, reviewers highlight its balance between intelligent automation and user-friendly design. Its ability to translate real, messy workdays into clean, actionable insights stands out from the field of generic trackers. As AI becomes more central to workplace tools, Clockdiary is leading with transparency and actionable value—giving you time back each week to do more.

Since AI like ChatGPT first appeared, people have been wondering how you know when something was made with AI.

At first, people were concerned with knowing about whether text was made by AI. Teachers wanted to know whether essays were actually written by students and not an AI (which is really good at writing bland, functional reports on a topic). And people have come up with heuristics for detecting AI, such as overuse of the em dash – though that’s really useful and functional punctuation for everyone to use.

Still, for the average person, detecting AI-produced text isn’t of particular importance. If you see the statement, “I just saw bigfoot!” it’s not more likely to fool you whether that statement was written by a human or AI.

But if it’s a picture of bigfoot, that’s another matter.

You Can’t Just Count Fingers Anymore

The advent of AI images and video has led to a much bigger problems as traditionally we considered photos of something to be evidence of something. Now that AI can generate a photo of anything in seconds, it’s much harder to know what to believe.

Early on, it was pretty easy to tell if an image was made by AI as there were almost always weird artifacts. One of the most famous ones was hands having too many fingers (hands are the most versatile part of the human body, and it was hard for AI to do them correctly with just its usual pattern matching). Skin also tended to be too perfect and plastic-looking.

But the limitations didn’t last long. The current image generators pretty much always get hands right, now, and are great at imitating the look and feel of real photos. Maybe you can still tell something is wrong, but it’s now like those old “what’s wrong with this picture” in those Highlights magazines, where you’re carefully analyzing each image for something that’s wrong.

And it’s not just images anymore. AI video is advancing as well. Early on, it was also easy to spot as there was always something not quite right in each video, like limbs moving in impossible ways and things morphing into other things. But again, it’s advanced rapidly, and now it takes quite a lot of detective work to see if anything is off with modern (i.e., from the last couple of months) AI video.

Computers Rescuing Us from Computers

Tech problems require tech solutions, so many people are working on this problem from a number of different angles.

1. Fingerprinting at the moment of creation
  • Google DeepMind — SynthID. Every image generated by Imagen 3 or Gemini 2.5 Pro now carries an imperceptible frequency‑domain watermark that survives cropping, compression, and light editing. Anyone can drop a suspect file into the public SynthID Detector portal and learn in seconds whether the mark is present.
  • Content Credentials (C2PA). An open standard backed by Adobe, Microsoft, BBC, and others: each edit to a photo or video stores a cryptographically signed “manifest” in its metadata. If a watermark is stripped, the missing signature shows up as a broken chain. The spec reached version 2.2 in February 2025 and is already baked into Photoshop, Premiere, and Android 15 camera apps.
  • Platform labels. EU‑based services must label AI‑generated or significantly manipulated media under Article 52 of the 2025 EU AI Act; TikTok, YouTube, and Instagram now auto‑apply “AI” badges (or threaten takedown) when files lack provenance data.

2. Detecting after the fact
  • Reality Defender. An enterprise API that runs a “voting ensemble” of 30+ specialized models—some look at JPEG macroblocks, others at GAN fingerprint patterns, others at physiological inconsistencies—to flag deepfake audio, video, and text with an overall risk score. Banks and dating apps plug it in to catch impersonation scams in real time.
  • Intel FakeCatcher. Perhaps the cleverest trick: it treats a face video like a medical monitor. When a heart pumps, tiny color shifts in skin betray blood flow (remote photoplethysmography). FakeCatcher watches those shifts; if the pulse is missing or physics‑defying, the clip is probably synthetic. Intel demoed 96 % accuracy on 10,000 videos and now ships the model on its Arc GPUs so it can screen streams live.
  • Open‑source inspector kits. Journalists rely on the InVID & WeVerify browser plug‑in to run key‑frame reverse‑image searches, check video encoding trees for odd transcodes, and magnify ELA (error‑level analysis) heatmaps.

3. Authentic‑by‑design cameras
Smartphone vendors are starting to embed tamper‑proof hardware roots of trust. Truepic Lens (partnered with Qualcomm) signs a hash of every raw pixel along with time, location, and sensor ID, so an editor must break the signature to alter the file—a bright red flag for any verifier. 

4. Benchmarks and bake‑offs
Why believe the marketing? Public leaderboards such as Meta’s Deepfake Detection Challenge, NIST’s Medifor evaluation, and the 2025 “Open Forensics” competition pit detectors against brand‑new fakes every quarter. Accuracy numbers that looked great in 2023 (80‑85 %) now count as baseline; top systems score 98 %+ F1 on the latest FaceForensics++ “compression world” track—until the next wave of generative models drops.

Real Versus Artificial Intelligence

So does this mean the problem of fakes will be eliminated by tech?

No. Absolutely not.

It’s an arms race, with AI improving by the day, and all its output becoming increasingly difficult to distinguish from reality. And techniques that catch some images won’t catch others. For instance, AI image detectors might not tell if only part of the image was AI-generated (such as only a face being changed using a technique called inpainting). And even if all the major AI embeds watermarks, there will be other freely available ones where people will turn off that feature.

So what does that mean for images and videos? Well, now it’s the same as text.

If you saw online someone post, “I saw a t-rex in the park!” you wouldn’t just automatically believe it – especially if it’s someone you don’t know or don’t trust. Now, images and videos are no different than text. If someone posts an image or video or something amazing, you need to ask yourself, “Is this from someone I know and trust?” – no matter how realistic it looks. Because eventually everything can be faked, and it will just be up to common sense to know what to trust.

So the people who are inclined to get tricked by sketchy email forwards are also going to get tricked by AI images and video, and there isn’t really much people can do about it that they haven’t been trying with all the other scams out there. But the people who learned long ago to be skeptical on the internet are just going to have to spread that skepticism a little further. The most advanced tech will never replace common sense (at least not for another generation or two).

Partner Spotlight: Duet Display

Expand Your Workflow with Duet Display Transform your iPad or Android tablet into a wireless second screen for your Mac or PC with Duet Display. Duet unlocks multitasking, boosts visual workspace, and supercharges how you work—seamlessly and securely.

• Ultra-low latency, crisp visuals

• Supports touch, gestures, and Apple Pencil

Discover more at the Duet Display website.

Stay productive, stay curious—see you next week with more AI breakthroughs!