How are we using AI at TradeRange?

Thank you to those of you who have offered feedback about TradeRange or Tradely on any of our platforms! It is nice to hear from you. A thing we hear very often is "Is this written by AI?" or "Is this UI AI?". We understand your scepticism and would like to address it directly by announcing a few major updates and changes surrounding TradeRange.

Articles that are written by or with AI are now marked by a new AI Assisted tag. You can see them appear across articles already! This is here to offer clear communication from us to you regarding the origin of content. We had originally considered this showing what model was used, but with how heavily models vary even in the process of drafting a single article based of the goal in mind we have realised that articles involve between 2-5 models each, hence while we track this internally it is not displayed publicly to avoid cluttering the UI. This is a BETA feature and we are still assessing the SEO/GEO and human implications of using the tags publicly. If it becomes a hassle we may remove them, but to maintain our honesty we will continue tracking it internally, as we have been from the start.

Thank you to BitIlia for developing this feature, which is going to be explained in the next couple of lines. The MCP server is currently exclusively for internal use, but may be expanded with other features, which will be publicly available. Currently, the MCP Server allows us 3 main things:

We were hesitant to give AI Access to users through our platform, as there are plenty of stories on Reddit about people being billed thousands because they didn't safeguard their API Key enough. We are still working on AI Generated Summaries of Individual Portfolios and will continue doing so, as it is the most demanded feature. 

For now, we are offering a daily summary of the biggest news articles of the day directly on our website, via the home screen and via the posts page. This should make catching up with the daily news in the morning just a tiny bit easier! This is done by using the MCP Server with a scheduled task. The current model being used is Claude Haiku 4.5, which delegate to Sonnet 5 if it can't decide which articles are more important. We spent nearly a month refining the prompt to produce the kind of short summaries you see here today. We remain on the look-out for cheaper options that will produce a better result, but after trying all the cheapest frontline models we have decided to use Anthropic for the time being.

As mentioned earlier we will be working on offering you a customised experience, but currently this isn't even in Alpha for us due to the high security risks associated. Expect an update within around 60 days.

If you are interested about specific details you can reach out via our discord.