💬 Assistants IA

Fabian's take — "Five stars, with a warning label. It's back after a brief government-ordered pause in June, and in experienced hands on the right problems it's remarkable — I use it to plan, audit, and think through hard problems, then route everything else to cheaper models. Everyone else will burn through their token limit with nothing to show for it. A specialist's tool, not a daily driver."

Fabian's take — "My favorite AI for most tasks. I use the standard Claude for conversations, and I get the most out of it in agentic setups: Claude Code for autonomous coding, Claude Co-work for longer workflows. This website is built and maintained entirely through Claude Code. Five stars, no contest."

Fabian's take — "Perplexity is indispensable. I use it for quick research on various topics, to learn something new, and to run deep research for better understanding of markets, potential customers, and competitors. I also have daily automated tasks running — like media monitoring. Five stars."

Fabian's take — "Gemini surprised me. Prompted right, it's a very good intellectual sparring partner, and it's the one I reach for when I need to interpret an image, write a prompt for another image generator, or generate images and videos directly. The custom Gems are what won me over — I used a set of them to plan, specify, and design a whole professional app, and over 200 tickets came out of that workflow. It does ask for effort up front, so if you want an assistant that just works from a lazy prompt, this isn't the one for you."

Fabian's take — "This aggregator has literally all the models. All the LLMs, all the thinking models, including the open source ones, the research models, all the image generators, video generators, audio models, and much more. All-in-one UI in one app and even accessible via API. For the price of a single subscription such as ChatGPT Pro. Minus points for bad user experience and UI that hasn't been improved at all in two years."

Fabian's take — "I cancelled my ChatGPT subscription because I found other models more compelling for my daily work. That said, I still use it from time to time — it remains a solid all-rounder with the broadest feature set. It's just no longer my first choice for any specific task."

Fabian's take — "Microsoft botched this and it's no wonder very few people use it. The problem: it's not productized well. It's not as configurable and personalizable and doesn't learn as cleverly as ChatGPT and the likes. It doesn't interact cleverly with other software, amazingly not even with Microsoft products. Claude is better at Excel and Powerpoint than Microsoft's own assistant. High time for Microsoft to catch up!"

Fabian's take — "Rovo is the AI assistant for the Atlassian suite. We use Confluence for product specifications, and Rovo helps turn those requirements docs into epics and tickets in Jira. It doesn't feel really smart yet, but within a small, well-defined context, it's quite helpful. A solid 3 stars — useful but not impressive."