I test AI tools for a living. Every week I’m trying something new, and 90% of it ends up in the trash. But every once in a while, a tool just clicks — it solves a real problem, it fits into how I already work, and I can’t imagine going back to doing it the old way.
These are the five that stuck.
1. Claude (Anthropic) — My Thinking Partner
I know everyone says “use ChatGPT” but hear me out. Claude has genuinely become the tool I reach for when I need to think, not just generate. The context window is massive (I’m talking 100K+ tokens), which means I can paste in an entire research report and ask questions about it like I’m talking to someone who just read it.
How I use it: I paste my raw notes from a week of tool testing and ask Claude to identify patterns, contradictions, and the most interesting angles. It doesn’t just summarize — it synthesizes. That’s a different thing entirely.
Best for: Long-form thinking, analysis, writing feedback, research synthesis.
2. Perplexity Pro — Research Without the Rabbit Holes
I used to spend two hours on research and come out with 47 browser tabs and a headache. Perplexity changed that. It’s a search engine that actually reads the sources for you and cites them, so you can verify anything in one click.
How I use it: Every time I’m about to film a TikTok about a new AI tool, I run a Perplexity search first. It gives me the current state of the conversation — recent reviews, comparisons, founder interviews — in about 90 seconds.
Best for: Fast, cited research. Keeping your facts straight before you speak on something publicly.
3. Notion AI — The Notes App That Finally Talks Back
I’ve been using Notion for years, so adding AI to it felt like giving my brain a second brain. The killer feature isn’t the AI writing (it’s fine, not mind-blowing) — it’s the AI Q&A over your entire workspace.
I have thousands of notes in Notion. Now I can ask “what were my thoughts on AI video generators last quarter?” and it pulls from my actual notes. That’s not a small thing. That’s institutional memory.
How I use it: Meeting notes → instant action items. Research dumps → structured summaries. Old ideas → rediscovered and reused.
Best for: Anyone who already lives in Notion. The AI upgrade is basically free performance.
4. ElevenLabs — My Voice, Without Me Being There
I’ll be real with you: I’ve used ElevenLabs to create voiceovers for videos when I was traveling and couldn’t record. The voice quality is that good. But the more interesting use case is cloning your own voice to maintain consistency across content without always being in the studio.
Brands have started asking me about this for social content at scale. It’s a legitimate tool now, not a gimmick.
How I use it: Voiceover for explainer clips, A/B testing different narration styles, international content localization.
Best for: Content creators who need audio at volume.
5. Runway ML — The Video Editor I Didn’t Know I Needed
I’m not a video editor. I never have been. And Runway kind of removed that barrier. The Gen-3 model can extend video clips, remove backgrounds in motion, and generate B-roll that’s actually good enough to use.
Is it replacing professional video editors? No. But for a creator who needs to move fast? It’s a cheat code.
How I use it: Extending raw footage, cleaning up backgrounds, generating quick visual assets for thumbnails and overlays.
Best for: Solo creators who need video output without a full production team.
The Pattern I’ve Noticed
Every tool on this list does one thing: it removes a bottleneck I had in my workflow. Not adds features. Removes friction. The best AI tools aren’t trying to impress you — they’re trying to disappear into your process.
That’s the bar I use when I’m evaluating anything new. Does this make one step faster, better, or gone? If yes, it earns a spot. If not, it’s out.
Next week I’m looking at the new wave of AI agents — tools that don’t just help you do things, but do things for you. It’s getting interesting.
— Deco