lilAI AntiSpam vs The Bot Market
A full audit of our performance against our closest competition
How lilAI works
lilAI has utilised machine learning and Artificial Intelligence to provide a better solution, keeping chats clean, on-topic and safe. Our solution requires ZERO setup, other than some initial, generic information from the project. lilAI AntiSpam across the board, has provided significant benefit to customers across gaming, RWA, technology, DeFi and centralized exchanges.
Comparing our competition
We recently performed a message audit to measure how lilAI’s AntiSpam bot performs against competitor offerings, in real world environments. lilAI AntiSpam outperformed every other bot that was present in the communities we service, in some cases by up to 6,400%!
Methodology
Over a period of 5 days, we analysed a total of 17,000 messages from 10 of our more active communities. We looked for instances where lilAI deleted messages, and instances where other bots had deleted messages, looking for areas where performance was better or worse.
Who were we up against? Our competitors in the audit were: Rose Bot, Combot, Guardian, Protectron and Safeguard.
The results of the aforementioned audit were simply staggering 🤯 For us, our full audit revealed our AntiSpam service correctly deleted a total of 665 messages — meaning that nearly 4% of all messages received were scam and spam attempts. During this period, we did not receive a single ‘false negative’ or ‘false positive’ flag from our communities, suggesting that we did not miss any messages that needed to be deleted, or deleted messages that weren’t spam.
During these 5 days, we observed that lilAI AntiSpam removes up to 64 times more spam than competitors, and removed at least 27% more spam than the leading competitor claiming to use AI for anti-spam. Not bad for a no-setup solution. Users simply drop it into their chat, and get industry leading performance, straight out the gate!
Our competition in action
Let’s examine some of the most interesting observations, shall we? 🧐
Firstly, we see that Protectron mostly reacts to any message which contains any type of link. While this clearly removes spam and scam attempts, it undoubtedly has a negative effect on the community it has been deployed into.
Next, another interesting observation was the varied performance of Rose bot. Rose bot is a very versatile tool, but requires users to manually set up every aspect of the bot. Rose responds to ‘bad words’ which Rose users need to specify, carefully balancing removal of spam without damaging the real users within the group. This presents a particular problem for Web3, where slang and industry terms are often shared within normal messages and scam attempts. Anecdotally, we’ve heard some of our customers have spent over a full calendar year trying to configure Rose, just to get the right balance. Since moving to lilAI, all of these customers no longer rely on Rose for bad word removal; they simply use Rose for the other moderation tools that it provides.
Some anti-spam tools utilise an entry ‘portal’ to make people take extra steps to prove they are a human. This often requires people to verify their location by IP address, and typically creates a barrier to entering communities. Even with this extreme measure, we observed a severe lack in real world performance compared to lilAI, with competitors such as Safeguard removing only 12% of spam messages that lilAI acted on.
Bots present in our communities
Many of our customers continue to keep competitor bots in their chats. Sometimes because the bot offers other functionality, and sometimes because the bot has functioned in their chat for a while and has become a safety blanket. Since adding lilAI AntiSpam, many projects have removed other bots, or have at least disabled their admin rights. Why? It’s because they have found lilAI AntiSpam to be superior in terms of product, service and support.
It was observed that in addition to lilAI AntiSpam, several projects utilised more than 1 bot for moderation. Rose and Combot were often seen together in a channel, as well as Rose and Guardian, suggesting one of these bots on their own provides insufficient performance.
Get in touch with lilAI today!
The results are in… and the data doesn’t lie. If you’re looking for an AntiSpam solution to better moderate your communities, we are here to help! Currently our services cover both Telegram and Discord.
Feel free to DM us on X directly, or fill out a quick form to get in touch 💪🤖
Bot Glossary
Rose bot. A very popular choice for general chat moderation, provided a wealth of tools and features, that a community manager can program as they wish. Rose requires users to specify ‘bad words’ which are used to delete or respond to messages. Some of our customers claim to have spent over a year trying to get Rose to perform well, and have not succeeded.
Combot. One of the most widely used moderation bot on Telegram. Combot utilises simple tools like Rose, but provides a database of bad actors via their ‘CAS Ban’ tool.
Guardian. This bot utilises entry ‘portals’ to shield the actual chat from bad actors entering. The need for ‘human verification’ is aimed to stop algorithms joining accounts via the usual means. Guardian aims to only let humans into the community chats, though scammers are learning how to get around these portals, as we will show.
Protectron. The only functioning bot claiming AI anti-spam. Protectron uses various filters, such as looking for URLs to remove potential spam/scams.
Safeguard. Another portal and simple moderation tool. Combines features from other bots such as Rose and Guardian to provide protection to users.