Dialpad Logo
AI RESEARCH LAB

Advancing Agentic AI for real-world impact

Our AI Lab builds agents tailored to your business DNA. By fusing linguistic expertise with your proprietary data, we deliver secure, ethical AI that operates with full context and measurable impact.


Areas of research


Real-Time Communications Intelligence

Understanding human conversation as it happens.

We create real-time language intelligence that listens, understands, and responds as conversations unfold. Grounded in data and linguistics, our models surface the right insights at the right moment—helping people communicate with clarity, empathy, and impact.


Learn more
Real-Time Communications Intelligence

Agentic AI Systems

Reason, retrieve, and act autonomously.

We’re creating AI collaborators—systems that reason, learn, and act from real data. Rooted in ethics and evidence, they grow through experience while keeping humans firmly in control.

Learn more
Agentic AI Systems

Universal Understanding

Enabling businesses to connect across languages and locales 

We’re teaching Dialpad AI to understand every voice, expanding across languages, dialects, and cultures. Grounded in linguistic science and real conversational data, our models capture nuance and context naturally, creating inclusive, accurate AI that listens globally and learns locally.

Learn more
Universal Understanding

Perceptive Systems

Expanding the science of understanding.

We’re exploring how people connect and anticipate what’s next. From analytics and forecasting to perceptual AI, our work blends science, ethics, and creativity to expand the boundaries of understanding.

Learn more
Perceptive Systems

Responsible & Evidence-Based AI

Ethics, privacy, and safety grounded in data and design.

Ethical AI is not an afterthought—it’s embedded in every experiment and product we create. Our systems are explainable, compliant, and privacy-preserving by design—grounded in scientific rigor around fairness, accountability, and safety. From GDPR and HIPAA to PII redaction, we innovate responsibly.

Learn more
Responsible & Evidence-Based AI

Featured papers

How Accurate are LLMs at Multi-Question Answers on Conversational Transcripts?

Dialpad’s new batch prompting framework lets AI answer multiple questions from a single conversation in one pass—cutting cost, latency, and complexity. With fine-tuned smaller models rivaling GPT-4o in accuracy, it shows how open, efficient AI can power real-world understanding at scale.

Read article
How Accurate are LLMs at Multi-Question Answers on Conversational Transcripts?

LLM Evaluate: An Industry-Focused Evaluation Tool for Large Language Models

Dialpad’s LLM Evaluate is an on-premise framework for testing large language models with rigor and privacy. It enables accurate, reproducible, and cost-efficient benchmarking across open and closed models, advancing trustworthy, data-driven AI development for real-world use.

Read article
LLM Evaluate: An Industry-Focused Evaluation Tool for Large Language Models

All papers

How Accurate Are LLMs at Multi-Question Answering on Conversational Transcripts?

Read article

DACP: Domain-Adaptive Continual Pre-Training of Large Language Models for Phone Conversation Summarization

Read article

Domain Adaptive Continual Instruction Pre-Training via Reading Comprehension on Business Conversations

Read article

LLM Evaluate: An Industry-Focused Evaluation Tool for Large Language Models

Read article

Can Post-Training Quantization Benefit from an Additional QLoRA Integration?

Read article

AI Knowledge Assist: An Automated Approach for the Creation of Knowledge Bases for Conversational AI Agents

Read article

Resolving Transcription Ambiguity in Spanish: A Hybrid Acoustic-Lexical System for Punctuation Restoration

Read article

Unsupervised emotional pattern recognition using rhythmic and vocal features

Read article

Tiny Titans: Can Smaller Large Language Models Punch Above Their Weight in the Real World for Meeting Summarization?

Read article

Query-OPT: Optimizing Inference of Large Language Models via Multi-Query Instructions in Meeting Summarization

Read article

Double Decoder: Improving latency for Streaming End-to-end ASR Models

Read article

AI Coach Assist: An Automated Approach for Call Recommendation in Contact Centers for Agent Coaching

Read article

Query-OPT: Optimizing Inference of Large Language Models via Multi-Query Instructions in Meeting Summarization

Read article

Predicting Customer Satisfaction with Soft Labels for Ordinal Classification

Read article

Building Real-World Meeting Summarization Systems using Large Language Models: A Practical Perspective

Read article

Are Large Language Models Reliable Judges? A Study on the Factuality Evaluation Capabilities of LLMs

Read article

BLINK with Elasticsearch for Efficient Entity Linking in Business Conversations

Read article

Developing a Production System for Purpose of Call Detection in Business Phone Conversations

Read article

An Auto Encoder-based Dimensionality Reduction Technique for Efficient Entity Linking in Business Phone Conversations

Read article

Punctuation Restoration in Spanish Customer Support Transcripts using Transfer Learning

Read article

An Effective, Performant Named Entity Recognition System for Noisy Business Telephone Conversation Transcripts

Read article

Entity-level Sentiment Analysis in Contact Center Telephone Conversations

Read article

Extracting Similar Questions From Naturally-occurring Business Conversations

Read article

Improving Named Entity Recognition in Telephone Conversations via Effective Active Learning with Human in the Loop

Read article

Improving Punctuation Restoration for Speech Transcripts via External Data

Read article