Faster, More Data, and Less Expensive AI chatbots

Due to changes in our AI technology, we're now able to deliver faster AI chatbots, with more data. And we've changed our pricing to dollars. This implies you get more value for less money - While our chatbot technology is using less resources, resulting in less energy usage, and less constraints on the environment. Basically we're a "green AI chatbot vendor", at least as much as that it possible.

Previously an AI chatbot with 2,500 training snippets would spend roughly 10 to 15 seconds before answering. Now the same chatbot typically answers in 3 to 5 seconds, using much less resources while answering, allowing us to provide this value to our customers as a kick back bonus. We're therefor increasing how many training snippets each chatbot can handle, while simultaneously reducing the cost of our plans by roughly 10%. The price reduction comes from switching from EUROs to dollars in a one to one relationship.

Notice, for existing customers that are already paying in EUROs, both our professional plan and our enterprise plan have less support and customisation now, so we won't automatically downgrade your payment plan to dollars, unless you explicitly ask us about it, realising you will end up with less support.

If you get an email asking you if you wish to reduce your monthly cost by 10% by paying in dollars, please have in mind that we do provide less support in our professional plan, and we only provide email support. We do not provide private channel Telegram or WhatsApp support on our new pricing for the professional plan, and we only provide YouTube-based training. However, for existing customers already paying in EURO, we will honor the private PM support agreement of course, regardless of what plan you're on.

Speed improvements

We've measured these speed improvements, and for an AI model with 20,000+ training snippets, the chatbot will typically start answering after 3 to 4 seconds. Previously this would require 20 to 30 seconds. So for large models with a lot of data, these speed improvements are a really big deal, and we can therefor handle much more data than previously, while still using less resources.

Notice, you still get your own private database, and there is no shared database in our cloudlet technology. We do not use any commercial vector database vendors either, and we're still using our own VSS database technology. The only difference is that we've further optimised how our own VSS database actually works, allowing us to significantly improve its performance.

This significantly reduces electricity consumption too 😊

GDPR compliant

The above implies among other things that we're still 100% GDPR compliant, contrary to most other AI chatbot vendors. In addition we can also provide 100% custom solutions, with a local VPS provider of yours, ensuring no data ever leaves your country of choice. The latter is of course a 100% custom thing, and needs to be billed accordingly - But we're not using a commercial 3rd party vector database cloud service provider, so as far as we know, we're the only AI chatbot vendor out there that can do this.

For the public sector in Germany for instance, the latter point from above is a really big deal, allowing us to deliver 100% compliant solutions, also for the public sector in Germany. Other EU countries have similar rules and regulations, which we're able to comply by 100% if required.

AI Search

The above speed and performance improvements also affects our AI Search product. Similarly to how our AI chatbots are much faster, AI search will also be much faster. Technically, what we have done, is to significantly improve the way we retrieve matches from our VSS database as a query is phrased. This is at the core of both our AI chatbots and our AI search product, implying both products are gaining from these speed improvements, allowing us to for instance deliver AI search based upon a lot more data than previously.

The details

Below are the new storage values for our two primary plans.

Professional plan handles 10,000 training snippets and up to 5 projects now
Enterprise plan handles 30,000 training snippets and up to 25 projects now

This allows you to create up to 5 chatbots with our professional plan, containing a total of up to 10,000 training snippets - And up to 25 chatbots with our enterprise plan, containing a total of up to 30,000 training snippets.

If you're creating your chatbot by scraping your website, realise one web page is roughly 5 training snippets. This allows you to for instance create an AI chatbot by scraping 2,000 web pages and yet still (probably) be within our professional plan.

Same amazing service

We still provide the same amazing service, unless you sign up for a professional plan with our new $198 per month price, at which point we only give you email support as of now.

Our cost as a company isn't really hardware resources now, it's configurations, manual quality improvements, support, and customisations. We wrote about this the other day where we asked the question should you use our open source version.

The cost of our infrastructure isn't our largest cost. Don't get me wrong, we're paying a substantial amount for our Kubernetes cluster, CDN, etc - But the primary cost is "the human factor". Most of our customers don't see the amount of energy we spend making sure their AI chatbots and AI search widgets are performing. For them "it just works, every single day, 7 days of the week".

It's actually a surprising amount of manual labour required to ensure your automated support and sales AI chatbot is working as it should. Later today we will upgrade all your cloudlets to handle the increased sizes according to your plan, so before the end of the day, your quality of services will simply "silently significantly improve".