Your productivity toolkit probably hasn’t changed much in five years. You’re still using the same note-taking app, the same search engine, the same creative software. Meanwhile, AI tools have gone from interesting experiments to genuinely useful applications that can save you hours every week.
The problem? There are thousands of AI tools now, most of them mediocre clones or solutions searching for problems. I’ve spent the last three months testing dozens of them to find the ones actually worth installing.
The New Search Experience You’ll Actually Use
Perplexity has done something Google hasn’t managed in years: made search feel innovative again. Instead of ten blue links and a wall of ads, you get direct answers with sources cited inline, plus the ability to ask follow-up questions naturally.
I used it to research a complicated tax question last week. Where Google would have sent me down a rabbit hole of forum posts and outdated blog articles, Perplexity synthesized information from the IRS website and recent tax publications into a clear answer in about 30 seconds. The free tier gives you five AI-powered searches per day, which sounds limiting but is actually enough for most people.
The Pro version costs $20 per month and removes the search limit while adding access to GPT-4 and Claude. It also lets you upload documents and search within them, which is surprisingly useful for research projects.
What Perplexity Doesn’t Do Well
It struggles with very recent events from the last few hours, and occasionally cites sources that don’t quite support its claims. Always click through to verify critical information, especially for medical or legal questions.
The ChatGPT Upgrade Nobody Talks About
Claude from Anthropic deserves more attention than it gets. While everyone obsesses over ChatGPT, Claude has quietly become the better choice for many tasks, particularly anything involving large amounts of text.
The context window is massive—200,000 tokens, which translates to roughly 150,000 words or about 500 pages of text. I’ve used it to analyze entire business contracts, compare multiple research papers simultaneously, and even edit a 40,000-word novella in a single conversation. ChatGPT would choke on most of these tasks.
Claude also tends to be more nuanced and less prone to hallucination than GPT-4, though it can be overly cautious and refuse reasonable requests. The free tier is genuinely useful with no credit card required. Claude Pro costs $20 per month and gives you 5x more usage, priority access during peak times, and early access to new features.
Where Claude Falls Short
It refuses more requests than ChatGPT, sometimes infuriatingly so. Ask it to write anything involving conflict or mild controversy, and you’ll get a lecture about ethics. It also can’t browse the web or generate images.
Video Generation Finally Gets Useful
Runway released Gen-2 earlier this year, and it represents the first video generation tool that doesn’t just produce novelty garbage. You can create 4-second clips from text prompts or extend existing videos with AI-generated content.
A friend used it to create placeholder footage for a client presentation when stock video wasn’t cutting it. The results weren’t perfect—there’s still that slightly uncanny AI look—but they were good enough to communicate the concept and move the project forward. That’s the threshold where AI tools become actually valuable rather than just impressive demos.
Pricing starts at $12 per month for 125 credits, where each credit generates one second of video. The $28 per month plan gives you 625 credits. For professional use, you’re looking at $76 per month for 2,250 credits. It’s not cheap, but it’s considerably less expensive than hiring a videographer for every small project.
Runway’s Limitations
Generated videos max out at 16 seconds, the output resolution is 720p, and you have minimal control over camera movement or subject actions. Complex prompts often produce bizarre results, and human faces still land squarely in the uncanny valley.
The Meeting Note-Taker That Actually Works
Otter.ai isn’t the flashiest AI tool, but it’s one I use almost daily. It joins your Zoom, Google Meet, or Teams calls and produces accurate transcripts with speaker identification and timestamps.
More importantly, it generates summaries of action items, key decisions, and important moments. I stopped taking notes in meetings entirely about six months ago. Now I just stay present in the conversation and review the AI-generated summary afterward.
The free plan includes 300 monthly transcription minutes and 30 minutes per conversation, which covers most casual users. The Pro plan at $16.99 per month ups that to 1,200 minutes monthly and 90 minutes per conversation. The Business plan costs $30 per user per month and is overkill unless you’re in meetings all day.
What to Watch Out For
Accuracy drops significantly with heavy accents, background noise, or technical jargon. The action item detection is hit-or-miss and often flags things that aren’t actually actionable. You still need to review the output rather than trusting it blindly.
Design Without the Designer
Midjourney has competition now, but it’s still the best AI image generator for quality over quantity. Version 6 produces images that can genuinely pass for professional photography or illustration in many cases.
I’ve used it to create custom illustrations for articles, mockup concepts for client presentations, and even artwork for my apartment. The Discord-based interface is clunky, but the results justify the learning curve.
There’s no free tier anymore. The Basic plan costs $10 per month for about 200 images. The Standard plan at $30 per month is the sweet spot for most users, offering 15 hours of GPU time. The Pro plan costs $60 per month and adds stealth mode so your images aren’t public.
Midjourney’s Weak Points
The Discord interface is genuinely terrible for organization and privacy. Text rendering in images is still broken—forget about generating anything with readable words. And despite improvements, it still struggles with accurate hand rendering and specific poses.
Code Generation That Doesn’t Embarrass You
GitHub Copilot has evolved from autocomplete on steroids to something approaching a junior developer. It now offers chat-based assistance, can explain code in plain English, and generates entire functions from comments.
I’m not a professional developer, but I maintain several small projects. Copilot has probably doubled my coding speed for routine tasks while helping me avoid stupid syntax errors. It’s particularly excellent for boilerplate code, test writing, and translating between programming languages.
It costs $10 per month for individuals or $19 per user per month for businesses. There’s also a free tier for verified students and maintainers of popular open-source projects.
Where Copilot Fails
It suggests outdated or deprecated methods surprisingly often, especially for rapidly-evolving frameworks. Security vulnerabilities sometimes slip into suggested code. And it’s terrible at understanding larger architectural decisions—it only sees the immediate context.
Research Assistant That Reads Papers for You
Elicit is purpose-built for academic research and literature reviews. You ask a research question, and it searches through academic papers, extracts relevant findings, and synthesizes the results into summaries.
A researcher friend used it to survey 200 papers on a specific medical topic in an afternoon—work that would have taken weeks manually. It’s particularly good at identifying patterns across multiple studies and flagging contradictory findings.
The basic version is free with limited searches. The Plus plan costs $10 per month and adds more searches, better summarization, and the ability to export results. The Pro plan at $42 per month is designed for institutional research use.
Elicit’s Constraints
It works best with STEM fields and struggles with humanities research where interpretation matters more than data. The paper coverage is extensive but not comprehensive, and it sometimes misinterprets nuanced statistical claims. Always verify key findings in the original papers.
The Bottom Line
These tools won’t replace human expertise, and they all have significant limitations. But they’ve crossed the threshold from interesting to useful—they save time, reduce friction, and genuinely improve output quality when used thoughtfully.
Start with the free tiers. Test them on real work, not toy problems. And stay skeptical of the output until you’ve verified accuracy on your specific use cases. The AI tool landscape changes monthly, but these represent the current best-in-class options actually worth your time and money.
Disclaimer: Tool pricing and features change frequently. Always verify current information on official websites. Results vary based on individual use case.
Want More?
Ready to find your next favorite AI tool? Browse our full reviews at UntappedAI AI Tools.
Sources
perplexity.ai • claude.ai • runwayml.com • otter.ai • midjourney.com • github.com • elicit.org
