Twin Opiniones 

Inizialmente tutto bello, poi, nonostante le istruzioni fossero sempre più stringenti ha prodotto report sempre meno interessanti spendendo tanti crediti inutilmente a volta in maniera esagerata. Inoltre per addestrare bene il modello ho dovuto spendere la metà dei crediti disponibili e nonostante tutto non esegue bene le istruzioni. Ho provato ad usarlo per un mese tutti i giorni e non rinnoverò più l'abbonamento. Ho sostituito con codex che fa le stsse cose ma meglio, incluso nell'abbonamento a chatGpt.

Excited about Twin.so a couple months ago, and planned to build it into my business. This week changed that. One run charged 557 credits for three reports, yet two came back byte-for-byte identical to an old report from three months earlier, not fresh data. Scraping seems to be the weak point, and runs cost far more credit than expected, more than building agents. Several mistakes in one week, and I'm seriously reconsidering renewing.

If I could I would give this website a negative. I found out about it on social media. It looked like it would be a great to help me improve my workflow in my company, I asked it before I started was it capable of producing an 80 page coloring book? I gave it the subject at first everything was going great. We ran through the styles and everything but at the end of the day it could not produce a sellable product. All I was given by the company is excuses as to why it couldn't perform.

I have asked him to help with photo editing and video processing tasks, but his performance in those areas was quite mediocre.

However, he is fairly capable when it comes to website API automation.

My assessment is that while he isn't particularly suited for repetitive file processing, he can handle automation scheduling tasks for workstations.

Here is what Twin.so said about itself...

The Twin AI assistant itself assessed this project's outcome as 2 out of 5. In doing so, the platform explicitly acknowledged that the key limitations — including the 8-second clip restriction, the unavailability of ffmpeg, and the inability to retrieve files across runs — should have been surfaced and communicated to the customer before credits were spent on approaches that were destined to fail. This self-assessment confirms that the platform fell well below acceptable standards for this engagement.

Avoid this platform, I run an AI agency working with experience working with fortune 500s. I paid for premium plan, tried running a simple task of extracting product images - the agent appeared to run, did not process the MD file I provided and then proceeded to try run the task.

After the UI freezing, I relogged, refreshed my browser and even tried to click around still nothing, it proceeded to rerun the same task over and over eating 1800+ credits in a single session. No way for me to stop it, see what the agent was doing.

After working with support they had no idea how to help fix the problem and proceeded to just inform me about credit spend. Special shoutout to Jose C. who didn't offer a solution.

I could care less about the $20 for the subscription - not to mention it says the free tier offers test credits (It didn't for me at least).

Im better off coding my own python script to complete the task - it is a pretty vibe coded garbage platform.

Hopefully they improve because the idea is great.

Franchement ce site est une grosse arnaque à fuir totalement !!! Lors de la finalisation de l'achat, sur la page paiement à l'abonnement à 20 euros il est clairement indiqué 10.000 crédits, mais une fois payé je me retrouve avec 2.000 crédits ! après les avoir contacté ils prétendent que c'est la plateforme Strype ! Mefiez vous car ils n'ont tjrs rien changé !