Activity
Mon
Wed
Fri
Sun
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
What is this?
Less
More
Memberships
businessXP
392 members • Free
1 contribution to businessXP
7d •
I use 4 paid AI platforms — here’s how they compare
I use 4 paid AI platforms — here’s how they compare in my workflow (yours will differ): ChatGPT — Knows me best (persistent memory) and aligns responses to my profile. — My go-to for writing (I use a dedicated “Writing” project to maintain context). — Missing a native browser sidebar (Mac-only custom browser exists). — Agentic features are useful, but take too many clicks to get started. — The first AI I subscribed to (because it was the first out of the gate) Gemini — Strong for image generation. — Recently added a Chrome sidebar — very useful. — Lacks agentic features, so I can’t rely on it end-to-end. — Google made me pay for it via our Workspace subscription. Perplexity — First (that I used) with a browser sidebar via its custom browser, very convenient. — Seems strong at research and comparison-style queries. — Agentic features are easy to start. — But it loses context too often, so I don’t fully trust it. — New “computer” agent costs extra (haven’t tried it). — I was given 1 year free access via my cell phone provider. Claude — Desktop app can access local files, unique and useful. — Has a browser sidebar. — But I regularly hit credit limits and pauses — frustrating. — Purchased it recently to see what all the hype was about. Bottom line: I’d prefer to use just one — but for now, I'll keep jumping around to find the best tool for the job. And yes, they all make big mistakes. This is as of March 2026 — I'm sure it will all change soon.
1 like • 6d
@Mathew Georghiou Are you using it from a browser (i.e claude.ai) or via an IDE (cursor, VSCode, ...)? Main diff is that when you use the browser, the model can't "see" the entire project (local files, errors, libraries, ...) and hallucinations are common due to lack of context. That's why most devs use AI from an IDE (native or as an extension). You'll still be using prompts but you won't be "chatting" anymore, but sending API calls and letting the model "see" the entire project (then you decide whether you allow it to edit the code directly or to just show it on the prompt panel). Also the biggest difference is that when used from the IDE, the model "takes control", plans the tasks to accomplish the project, troubleshoots, selftests, submits PRs, ... I'm using extensively Claude (Haiku 4.5, Sonnet 4.6, Opus 4.6) and Codex and it's impressive. You still need to be able to understand what it does, but I would not go back to the old days at all. Suggestion: consider AI not as a consultant (to whom you make questions expecting magic answers) but as a peer that sits next to you, sees your project and codes with you.
1 like • 5d
@Mathew Georghiou I think there’s a slight mix-up on the terminology. When I mentioned 'APIs' earlier, I didn’t mean you should have to work on or find information about APIs yourself. Indeed, I've never done it! I was simply referring to the background plumbing that an IDE (like ClaudeCode, Cursor, or VSCode with the GHCP extension) uses to 'see' your local files and understand the project context. When you prompt, the IDE automatically handles all the technical communication (API calls, agents, indexing), which is how it avoids the 'guessing' you ran into and produces the right code. Hope it helps!
1-1 of 1
Active 4d ago
Joined Feb 2, 2024
Powered by