Lang.ai's goal is to enable operators to control AI on their own. One of the biggest challenges we see when business operators deal with AI is the extended implementation time.
Traditional AI relies on a process that requires training data, labeling that data, long iterative cycles.....and we believe this type of AI is built by engineers for engineers.
With Lang.ai we've built a new way to exploit the power of AI with humans in mind, taking into account needs and requirements from real business operators. This drastically changes the process to begin building an enterprise-ready model:
- Upload your data
- Group the concepts that matter to you
- Deploy into production (in minutes versus months)
We're super excited to make two big announcements that get us closer to our vision:
1. We just released a new version of the AI that powers the Lang.ai product, which reduces the time to generate Lang.ai initial projects by up to 5x. The impact is particularly relevant for large customers as it has a bigger impact on projects with more than 150,000 records. For example in a production environment we measured the following improvement:
- 2x improvement in a project with 40,000 tickets. It now takes less than an hour to get started with a project of 40,000 historical Zendesk tickets vs 2 hours with our previous version.
- 10x improvement in a project with 200,000 tickets. For a project with 200,000 records, it took more than 6 hours in the previous version to process, whereas in the new version it's been under an hour.
This is huge for our customers because now the feedback loop to generate initial projects is drastically reduced. Our end goal for implementing new projects is that we can reduce our average implementation time from 2 workdays to 1 workday. This means that with Lang.ai you are now able to create a live tagging, routing and prioritization project with 100 tags in 4 hours or less after we connect your data.
2. Lang.ai intellectual property is now backed by U.S. patent number 10,977,446 for our "Unsupervised language-agnostic intent induction and related systems and methods", further validating our novel and differentiated approach and technology spearheaded by our Chief Data Scientist Henry Anaya.
We are super excited about the new algorithm versions that are coming from our Data Science team and how we are driving AI to be:
- Human-centric. Designed for and by operators, not for engineers.
- Flexible & Interpretable. Anyone can easily understand, interpret and modify any decision made by Lang.ai to adapt the process to the business perspective. Put simply, it is the opposite of a black-box.
- Reliable. Soon, Lang.ai will be able to detect potential errors in any manual process involving agents or customers selecting custom fields, as well as in automated processes that look only for keywords or rely on traditional AI approaches. By analyzing the semantics we are able to suggest potential errors to our users and calculate the true impact of relying on manual work for high-volume, repetitive processes.