AI-Generated Audio Summary

People often ask us why an education company that teaches teachers ended up building a proficiency platform. The honest answer is that we did not set out to build a platform. We set out to solve a problem for real students, one course at a time, and the platform is what that work became.

This is the story of how we got from a single bilingual certification course to Praxitude AI, and why the path made sense at every step.

It Started with Verónica, and She Was Never a Chatbot

For about four years, we have been building artificial intelligence directly into our online courses. The first thing we built was Verónica, our virtual tutor. From the beginning, we were careful about what she was.

Verónica is a tutoring system, not a chatbot. A chatbot answers whatever you type based on whatever it was trained on. Verónica is different. She is grounded in the actual content of the course a student is taking. She evaluates responses against a rubric, the same way a human instructor would. She gives feedback tied to evidence in the student’s own work, and she feeds into learning analytics so we can see where a cohort is struggling. The difference matters, because a tutor that invents plausible-sounding answers is worse than no tutor at all, especially in a certification course where accuracy is everything.

We wrote more about the thinking behind this in our post on how we are using AI to redefine teacher preparation. The short version is that future teachers need more practice with real feedback than any program can staff by hand, and a well-grounded AI tutor helps close that gap.

The BTLPT Was Our Proving Ground

Our anchor course has long been preparation for the Bilingual Target Language Proficiency Test, the BTLPT. Passing it is a requirement for bilingual certification in Texas, and it is a demanding exam that tests listening, reading, speaking, and writing in Spanish.

We did not build this in a vacuum. We worked with universities and Texas school districts to help both pre-service teachers and current teachers prepare and pass. Over time we assembled a full set of BTLPT resources, from practice items to speaking simulations to the evidence-based feedback Verónica provides. Working with real cohorts taught us something important: for a language exam, studying is not the same as performing. You can memorize vocabulary and still freeze when you have to speak under pressure. That lesson shaped everything that came next.

Then Texas Changed the Exam

In the middle of all this, the ground shifted. Texas is consolidating the Bilingual Education Supplemental (164) and the BTLPT (190) into a single new exam, the TExES 165. The new exam launches on September 1, 2026, and the older 164 and 190 exams sunset on August 31, 2027, with an overlap period in between so candidates can take either path for a while.

This is a significant change for every bilingual teacher candidate and every program that prepares them. We laid out the full timeline and what it means for candidates on our roadmap for Texas bilingual certification changes, which we keep updated as the state releases new information.

The important detail for this story is the shape of the new exam. It leans toward performance, toward showing what you can do, not just what you can recall. That is exactly the kind of assessment we had been building toward with Verónica and the BTLPT course.

The Change Pushed Us into ESL and Scenario-Based Practice

The consolidation also nudged us to broaden our work into the ESL Supplemental (154), since many of the same candidates need that certification too. As we expanded, we leaned harder into a method we already believed in: scenario-based practice.

Instead of asking a candidate to pick the right answer from a list, a scenario drops them into a realistic classroom challenge and asks them to respond. A parent conference that turns tense. A student who is not understanding a science lesson delivered in their second language. A moment that requires judgment, content knowledge, and the right language, all at once. We wrote about why this matters in our post on building the next wave of scenario-based learning. Multiple-choice tests measure recall. Scenarios measure whether you can actually do the job.

We Are Not Just Using a Chatbot

Here is the part we want to be clear about, because it is easy to assume otherwise. We are not simply dropping an off-the-shelf chatbot into a course and calling it AI.

We write real code. We work directly with the interfaces that language models expose, and we deliberately choose among several different models depending on the task, because the model that is best for grounded tutoring is not always the model that is best for generating a scenario or evaluating a spoken response. Four years of hands-on engineering with multilingual AI taught us where these systems are reliable and where they are not. That knowledge is the reason Verónica stays grounded in course content instead of improvising, and it is the reason we trust our scenario evaluation. Modern AI was multilingual from its foundation, which is a real advantage for bilingual education, but only if you build on it carefully.

So We Built Praxitude AI

All of this led to one conclusion. The scenario-based, evidence-grounded, dual-language approach we had been refining inside our courses deserved to be its own platform. That platform is Praxitude AI.

Praxitude AI measures two things at the same time. First, professional competency: can you perform the task correctly, with the right decisions and the right domain knowledge? Second, target-language proficiency: can you do it in your second language, with the right vocabulary and register? We call this dual-axis evaluation, and it is the heart of the platform. A program leader gets a dashboard that turns individual performances into insight about where a whole cohort needs support, not just a spreadsheet of pass and fail.

This reflects a belief that runs through everything we build, from Verónica to Praxitude AI: experts should design the training, and AI should power the practice. The rubrics, the standards, the scenarios, and the very definition of good performance all come from human expertise. The AI is what carries that expertise to every learner, available around the clock, with feedback grounded in real knowledge and real criteria. Humans stay on the loop, guiding how the AI is used and keeping it honest. That is the idea behind our name and our tagline: Human-designed professional training, AI-powered practice.

We built Praxitude AI first to serve teacher-preparation programs facing the TExES 165 transition, because that is the world we know best. But the same engine works anywhere a professional has to perform competently in a second language. Think of a nurse explaining a procedure to an English-speaking patient, or a hospitality professional handling a difficult guest. The competency changes, the vocabulary changes, the language changes, but the need to practice and be assessed in realistic scenarios stays the same. That is why Praxitude AI reaches into professions well beyond education.

Get Started

If you lead a teacher-preparation program navigating the TExES 165 change, or you work in another field where people need to prove they can perform in a second language, we would like to show you what dual-axis, scenario-based assessment looks like in practice. You can learn more and see the platform at praxitude.ai.

And if your focus is Texas bilingual certification right now, our TExES 165 roadmap and our BTLPT resources are the best places to start.

Praxitude AI is a proficiency and training platform developed by Enabling Learning LLC. It is designed to support educators and other professionals, not to replace them. As always, human judgment remains at the center of good teaching and good assessment.