Watch AI Do the Interviewing — On-Demand

**Mike Valera:**

Hi, everybody. I'd like to welcome you all to our session on TalkScore, specifically on AI interviews: smarter assessments in action. I'm Mike Valera, the Market Growth Manager here at TalkPush. On behalf of the team, I'd like to thank you all for joining us. I see people from the Philippines, Colombia, Mexico, and Honduras. We're really excited to show you what we have in store. We're going to be moving fast today, so make sure to stick around till the end for a chance to try the assessment yourself and for one lucky person to win a thirty-dollar gift card. We'll have a Q&A session at the end, so feel free to send your questions in the chat box.

Let's talk about the current hiring reality. At TalkPush, we see every day how AI has changed hiring. Job postings that once drew a trickle of resumes now receive hundreds, even thousands, of applications. We've found there's a 60% candidate bailout because the process takes too long. Recruiters are drowning in volume, and most concerningly, we're losing the people we want the most because there are so many people to get through.

We try to fix this with manual assessments, but they are time-consuming, leading to fatigue and inconsistency. Scores can vary depending on the time of day or our mood, and interviewer bias inevitably sneaks in. On the flip side, pure automation hasn't cracked the code either. Candidates can use AI to generate rehearsed, scripted answers, and many automated tools can't probe deeper or ask follow-up questions, leading to shallow evaluations.

So we're caught between two extremes: manual methods that are slow and inconsistent, and automated tools that can feel shallow or be easily gamed. That's where TalkScore comes in. It combines the best of both worlds by automating the parts that should be fast—screening, engagement, and scheduling—while preserving the depth of a real conversation. With TalkScore, candidates complete short, natural interviews on any device. The platform listens, probes when needed, and evaluates responses in real-time, capturing how people think and adapt. It uses consistent criteria and built-in bias checks to deliver fair, objective results in minutes, not hours.

Now, I'd like to introduce our speakers: our TalkScore Product Manager, Beatrice, and a familiar face, Perry Santiago. I'll welcome Beatrice to the stage to share her insights.

**Beatrice:**

Hello, everybody. It's a pleasure to be here. So, what is TalkScore? It's TalkPush's next-generation AI-driven voice interviewing assessment tool. It goes beyond a simple English assessment to deliver comprehensive candidate screening for any job campaign's key requirements. It's designed for high-volume hiring, helping recruiters pinpoint the most qualified candidates from thousands of applicants within minutes.

TalkScore is bringing the best of both worlds. The AI provides an interview that is more than just a robotic feeling; it's captivating and understands your responses. It picks up key points to determine if you are right for the job. It's better than usual interviews because we bring the human side, and it's better than human-only interviews because we can manage thousands of applicants at any time, on any day, on any device—a laptop, a phone, even on a bus if you have five minutes.

It offers real-time scoring, so you get instant reports and scores for crucial hiring decisions. The interview is a human-like, engaging conversation, not a robotic one. The workflow is integrated and seamless, with no extra clicks that might cause a candidate to drop off. It's also multilingual, supporting up to fifty languages, and consolidates multiple screening steps into one interaction. With TalkScore, you don't need separate English, cultural, or behavioral assessments; you get it all in one full report.

Compared to other assessment providers, our completion time is 5-10 minutes versus their 20-45 minutes. Our content is spontaneous and conversational, while theirs is often scripted and easily cheated. Our platform is mobile-friendly, while others are often limited to desktop or onsite. This results in a completion rate higher than 75%, compared to their 10-35%.

Now let's show how this works in real time. Mike will be my candidate.

**Mike Valera:**

Yeah, I'm just nervous, to be honest. Okay, let me load up the page. So this is what the candidate experience looks like. We offer a landing page to make the process easier. I'll put in my information, select SMS for engagement, and add my LinkedIn profile.

**Beatrice:**

In the application part, the recruiter can test the candidate in many ways, with video questions, audio questions, or even script tests to check fluency and pronunciation, all before the interview even starts.

**Mike Valera:**

Exactly. As you can see here, there's an audio assessment built-in. Let me give this a try.

*(Mike answers an audio question about learning a new tool for a client)*.

Alright, I'm probably not going to get hired, but it gives you a good idea. You can also upload a resume. Now that I've finished the application part, I'm moved to the AI interview with Stephanie, my AI interview buddy.

**(Live AI Interview Demo)**

**Stephanie (AI):** Hello, Michael Valera. I'm Stephanie, an AI assistant from TalkPush. How are you today?

**Mike Valera:** I'm doing all right. Thank you for asking. How are you?

**Stephanie (AI):** I'm doing well, thank you. I just wanted to spend a few minutes getting to know you better. Nothing complicated, just a relaxed chat. Would that be all right?

**Mike Valera:** Yeah, that would be fantastic. I'd love to get started.

**Stephanie (AI):** Great. Can you describe a time you had to explain a complex problem to someone who was unfamiliar with it? How did you ensure they understood?

**Mike Valera:** I think an important part is making sure you understand the problem first. It's about pulling in the right resources internally to find a resolution as quickly as possible. Then, it's a matter of delivering the solution as simply as possible.

**Stephanie (AI):** Thank you for sharing that. It sounds like you put a great emphasis on understanding the problem first. Could you provide a specific example?

**(The interview continues with Stephanie asking follow-up questions based on Mike's answers about handling difficult customers, adapting to change, and learning new tools.)**

**Stephanie (AI):** You've shared some wonderful insights today. Before we wrap up, do you have any questions for me about the company or the role?

**Mike Valera:** I'd love to get a good understanding of the next steps. What should I expect?

**Stephanie (AI):** Great question. After this call, our recruitment team will review your responses. If everything aligns, you may be contacted for the next steps. Just to confirm, this call was recorded for our team to analyze. Thank you for your time today. Best of luck.

**Mike Valera:** Fantastic. Thank you.

---

**Beatrice:**

That was not an easy interview! As you can see, our questions are designed for an open-ended conversation, and the AI is prepared to adapt. The full assessment for the job requirements is completed within that conversation. Now, Mike will show his report, which is already live in our CRM.

This is the candidate report. You can see Mike's overall score, which was 4.5 out of 5—a very high score. Below that is the AI summary assessment, highlighting his strongest and weakest points. On the right, you see the scores for specific skills like communication skills, empathy, and problem-solving, with explanations for each score.

Further down, you can see the English proficiency analysis, which assesses grammar, fluency, vocabulary, and pronunciation. Finally, there's a sentiment analysis that shows how motivated or excited Mike was when answering, and a section for candidate feedback on the process.

**Mike Valera:**

As someone who did the interview, I have to say it was difficult because it was live. I couldn't read the questions in advance, so it felt like a real in-person interview, which leads to more authentic answers. But it's super convenient that I can do it anytime, 24/7, on my phone or laptop.

**Beatrice:**

Exactly. And to prevent cheating, we created a "reading detection" feature that can tell if a candidate is reading from a script based on their pace, pauses, and speech patterns. This feature is built into the interview.

TalkScore works by combining job-specific skills with a human-defined evaluation rubric, which is key to ensuring the scoring is unbiased. For English assessment, we use the CEFR framework to evaluate fluency, grammar, pronunciation, and more. All of this is customizable. You can build your own interview, modify questions, and even incorporate your own scoring rubric to match your needs.

We've already had a success story with Concentrix. We ran a test for the past month, and their completion rate was 98%. We streamlined what was a four-step process into a single landing page, which was a huge help for both the recruiters and the candidates.

Now, I'll pass it over to Perry to talk about setting up TalkScore.

**Perry Santiago:**

Hi everyone. I'll talk about setting up the traditional TalkScore as a language assessment tool. This will still be available for all our clients. It's very easy to set up. You can use three different question types: open-ended questions, "listen and repeat" to test pronunciation and fluency, and "listen and speak" to measure comprehension. Once a candidate completes these questions, their profile is updated with new attributes like their overall score and CEFR score. You can then use these attributes to filter candidates and set up automated workflows to move them to the next stage. With the new AI interview, this process is even more dynamic, allowing you to tailor questions specifically for the role you're hiring for.

**Mike Valera:**

Fantastic. Now let's address some questions from the chat. Daniel asks about different accents and languages.

**Beatrice:**

Yes, we can. We have up to 50 languages and can also incorporate specific accents, like a Filipino accent for English, if a client needs it.

**Mike Valera:**

Ethel asks if a candidate can ask for the question to be repeated.

**Beatrice:**

Yes, they can say, "Sorry, I didn't understand," and the AI will repeat the question. This is not a reason for rejection.

**Mike Valera:**

Abigail asks if managers can access the report.

**Beatrice:**

Yes, managers can also have access to the report.

**Mike Valera:**

Jimmy asks how many scenarios can be created.

**Beatrice:**

It's unlimited. You can create as many different assessment campaigns as you need for different job roles.

**Mike Valera:**

Alexander asks if the bot learns from recruiter feedback on past interviews.

**Beatrice:**

At this moment, no, but it's something we can consider for future development. However, all past assessments are attached to a candidate's profile, so recruiters can see their history. We don't have the current assessment influenced by the past one because a candidate could have upskilled in the meantime.

**Mike Valera:**

And a final question from Abigail: Is it possible for an applicant to have a friend answer for them?

**Beatrice:**

No, that is not possible. The AI will detect the difference in voice, and we also recommend including video questions during the application to cross-check the person's identity against any uploaded documents.

Thank you all so much for joining us today. We have so many exciting things planned for TalkScore. If you want to see how this can work for you, please connect with your account manager or reach out to us through the demo page on our website.

Oh, and congratulations to Ethel K., the winner of our thirty-dollar Amazon gift card! Please send us your email in the chat so we can send it your way. Thank you, and have a great rest of your day.