5 AI Tools I Actually Tested (And What the Internet Thinks About Them)
Kiswah Ashfaq
Blog

I have a confession to make. I spend an embarrassing amount of time testing AI tools. Not just reading about them, but actually using them, watching them fail, occasionally watching them do something that makes me put my coffee down and stare at the screen. It's become a bit of an obsession. And lately, five tools in particular have been living rent-free in my head.
Let me take you through them honestly — the good, the bad, and the not so ugly.
Manus or The One That Actually Does the Work
Manus (manus.im) is what I'd call the overachiever of the group. Unlike most AI tools that sit comfortably in the "chat and advise" lane, Manus actually executes. It browses the web, runs code, builds files, and hands you back completed work. Tell it to research a topic and come back in an hour — it'll have something waiting for you.
I've always believed the real test of any tool is whether it can work without you hovering over it. For deep research tasks and async analysis, Manus passes that test more often than not. Its Wide Research and Slides features alone make it worth exploring for knowledge-heavy work.
But here's where the story gets complicated. Manus runs on a credit system, and a single deep research task can quietly burn through 1,000+ credits. Worse, reliability wobbles badly when it hits paywalls or CAPTCHAs — sometimes it just loops in quiet desperation. The Reddit crowd is split right down the middle: "Most useful AI product I've tested" on one side, and "Spent $20 to watch it refresh a page for an hour" on the other. Both reviews are probably true, depending on the day.
Cto.new which makes me want to believe “Free Shouldn't Mean Compromised”
Then there's Cto.new which is doing something that feels almost suspicious: it's completely free. Unlimited access to GPT-5 Codex, Claude Sonnet 4.5, and Gemini Pro. No API keys, no credit card lurking in the background.
For developers, the GitHub integration alone is worth the price of entry (which, again, is zero). Assign an issue, come back to a pull request. The multi-agent setup handles both task planning and code review in a single flow, which is the kind of thing that previously required stitching together multiple paid tools.
Of course, "fair usage" limits do exist, and the browser-first UI isn't exactly mobile-friendly. And I'd be lying if I said I wasn't at least a little curious about the long-term sustainability of a fully free model. But the developer community seems unbothered. The dominant Reddit sentiment? "Use cto.new for background agents, keep Cursor for local work — saves hundreds a month." That's a pretty compelling argument.
chroniclehq.com or Decks That Don't Need a Design Degree
Chronicle (chroniclehq.com) caught my attention because it approaches presentations differently. Most AI slide tools give you bullet points dressed up in a template. Chronicle gives you structure. You know, the hook, problem, solution, proof. It thinks in narrative arcs before it thinks in slides.
The output quality reflects that. I've seen users report that their decks come out 85–90% client-ready, which is a genuine leap over competitors like Gamma. The Peek and Deep Hover features are genuinely clever for live presenting. It’s the kind of detail that suggests someone actually thought about the experience of standing in front of an audience.
That said, if you're a designer who likes to control every pixel (padding, spacing, the precise weight of a border) Chronicle will frustrate you. The canvas doesn't do granular tweaking. And the monthly token system, where unused tokens simply expire, is a model I've never loved. The Reddit verdict: "Way better at one-shotting a deck than Gamma," with a footnote that templates can feel a bit rigid for strict corporate brand standards. Fair.
Moda.app or in other words, Design That Understands Design
Moda (moda.app) is addressing a problem that has quietly irritated designers since AI image tools became mainstream: the outputs are flat. You get a pretty picture you can't actually edit. Moda gives you a real layered canvas instead.
What makes it interesting is that it was trained on actual graphic design principles — layout, hierarchy, spacing.Not just aesthetics. You can use it for slides, social posts, PDFs, and UI mockups. In theory, it replaces three or four other tools.
The credit system with no rollover is a recurring complaint, and as a younger product, feature gaps show up (limited animations, no API access yet). The community's main gripe is also revealing: the AI is better at layout than at understanding the actual content you give it. That's an honest limitation worth knowing before you commit. Still, for designers tired of re-editing flat images elsewhere, this feels like a step in the right direction.
Inconify AI: Icons in Seconds, But at What Cost?
IconifyAI is the most specific tool on this list, and it delivers exactly what it promises: 1024×1024 high-definition app icons in seconds, across eleven-plus styles from clay to photorealistic to metallic. Commercial rights included, no attribution required, and credits starting at around $0.08 per icon.
The catch? No free trial. You pay before you know whether it works for you. The output is PNG only. No SVG, no pre-packaged iOS or Android size packs. And quality is inconsistent enough that some users report unclean lines and generic results after spending $17 or more.
The community is mixed. Fans say it kills the freelancer bill for a quick icon. Critics say tools like Recraft or Iconikai give better value. It occupies a narrow but real use case. If that use case is yours, it's probably worth a small test purchase.
So What's the Pattern?
Looking across all five, I notice something. The tools that generate the most genuine excitement — Manus, cto.new, Chronicle — are the ones trying to remove a step from a workflow entirely, not just assist with it. The ones with the most mixed reviews like IconifyAI, and to some extent Moda are working in spaces where expectations are already well-established and quality bars are high.
AI tools aren't magic. But the right one, used for the right task, genuinely changes what's possible in a day's work. My hands are itching to test what each of these looks like six months from now.
Funny how a set of tools can open up that many possibilities.
