SeedToScale | Curated by Accel

How fast can AI really talk back? For Smallest.ai, the answer is 45 milliseconds. In this episode, Akshat Mandloi, Co-founder, breaks down how they built one of the fastest and most efficient voice AI engines, proving that systems thinking, not scale, wins in the real world. He shares insights on continuous retraining, evolving architectures, and the hidden engineering layers that make Smallest AI a leader in real-time speech intelligence. Want to see how the future of AI sounds? Watch the conversation till the end. Chapters: 0:31 - Intro 0:37 - The secret sauce behind the world’s fastest TTS engine 1:43 - Different approach from the traditional auto-regressive model 2:41 - The process behind the parameters of efficiency and quality 4:00 - The one unique thing Smallest.ai is solving 4:57 - Lessons from calibrating the speech models 6:20 - Building different corpuses for different languages 7:29 - Solving the hallucination problems 9:33 - Mental model for building the initial team 12:17 - The most exciting part to take on 14:16 - Outro Follow SeedToScale on LinkedIn for more insights: linkedin.com/company/seedtoscale Follow Akshat on LinkedIn: https://www.linkedin.com/in/akshat-2503/ #AWS #InsideTheEngineRoom #LLM #AI #Innovation #Scale #AIModels #Voice #Language

What if your entire brand could run on autopilot? Sai Krishna, Co-founder, ShopOS, shares how AI agents will redefine the way brand-owners build, market, and scale businesses. From smarter catalogs and campaigns to fully automated brand orchestration, he explains why AI is the perfect draft engine, giving founders a head start on every creative and operational task. Sai breaks down the rise of ambient agents: AI that works quietly in the background to optimize pricing, discovery, and customer engagement at scale. Finally, he paints a bold picture of the shopping page of the future: dynamic, hyper-personalized, and tailored to every buyer’s unique context. Chapters: 0:26 Intro 0:39 House of Model’s journey and the problems it is solving 1:40 The core problem 4:27 The future of e-commerce with AI 7:27 How the idea is being enabled 13:10 A future with no clicks 14:13 Advice on fundraising 17:36 Outro Follow SeedToScale on LinkedIn for more insights: https://in.linkedin.com/company/seedt... Follow Sai on LinkedIn: / 5aitec #AI #Ecommerce #AICommerce #UnicornStartup #SeedToScale #Accel #AWS #InsideTheEngineRoom #Tech #Innovation #Startup #Founders

Integrations are the hidden bottleneck in modern engineering slowing teams down, draining time, and killing creativity. What if AI could automate that completely? Soham Ganatra, Co-founder & CEO, Composio, shares how his team built one of the largest hosted MCP platforms, where AI coding agents autonomously build and manage integrations end-to-end. From early experiments with LLMs to tackling agent reliability, tool design, and scaling through community-led GTM, the conversation dives deep into how Composio is transforming developer productivity. Chapters: 0:24 - Intro 0:35 - The problem statement Composio aims to solve 3:20 - The differentiator between Composio and others 5:34 - The secret behind high SWE-benchmark 7:29 - How is Composio powering rapid development & prototyping 8:45 - Identifying the initial customer base 13:23 - Were there any specific use cases that were surprising? 15:06 - Mental model around team building 19:37 - Outro Follow SeedToScale on LinkedIn for more insights: https://www.linkedin.com/company/seedtoscale Follow Soham on LinkedIn: https://www.linkedin.com/in/soham-ganatra/ #AI #AWS #InsideTheEngineRoom #Developers #Tools #Automation #ArtificialIntelligence For more insights on building AI-native startups and scaling developer ecosystems, watch the full episode of AWS Inside the Engine Room. #AI #AWS #InsideTheEngineRoom #Developers #Tools #Automation #ArtificialIntelligence

The fastest solution is not always the smartest one. Even the world’s fastest inference engine can’t win on speed alone. That’s why Amritanshu Jain, Co-Founder & CEO of Simplismart, emphasizes adaptability as the true goal. For enterprises, it’s not just about running models at lightning pace, but balancing cost efficiency, speed of execution, and scalability across dynamic environments. In a world where conditions evolve rapidly, adaptability is the edge that keeps businesses ahead. Chapters: 0:25 Intro 0:39 Core innovations in the engine’s architecture 4:07 Mental model around building modular blocks 6:26 Solving the gaps in MLOps 17:41 Approach with hiring 23:17 What kind of technology & research excites you? 28:42 Outro Follow SeedToScale on LinkedIn for more insights: https://in.linkedin.com/company/seedt... Follow Amritanshu on LinkedIn: / jainamritanshu About the series: The builders of India’s AI revolution are opening up their playbooks. From evolving stacks to scaling breakdowns and tough infra trade-offs, Inside The Engine Room, in collaboration with ‪@amazonwebservices‬, dives deep into the lessons every founder wishes they had on Day 1. #AI #AWS #MLOps #InferenceEngine #AIstartups #Amazon #AIInfrastructure #Founders #InsideTheEngineRoom #Accel #Simplismart #SeedToScale

In this engaging discussion, Harsh Sahai, Co-founder and CEO of Bridgetown Research, shares insights from his journey working at Amazon and McKinsey. He delves into how AI transforms board-level decision-making by autonomously gathering insights and enhancing research depth. Harsh highlights the contrast between consulting firms and tech companies regarding automation hesitations. Additionally, he discusses innovative hiring strategies at Bridgetown, focusing on adaptability over formal qualifications, and the vital role of strategic partnerships in niche AI markets.

Meesho is redefining the role of AI in e-commerce, driving innovations that scale to meet the demands of millions. From handling over 60,000 voice bot calls daily to managing 'trillions' of features, AI is embedded in every facet of Meesho’s operations. In this episode of Decoding AI, Anagh Prasad, Investor at Accel, sits down with Debdoot Mukherjee, Chief Data Scientist and Head of AI at Meesho, to explore the transformative impact of AI on one of India’s largest online marketplaces. Debdoot shares learnings from his early days of machine learning at Myntra to shaping Meesho’s AI-first marketplace. The discussion dives into Meesho’s use of AI for solving complex challenges, such as optimizing matchmaking between users and products, and developing scalable machine learning platforms. Debdoot also shares how open-source frameworks drive innovation at the company. This episode offers a compelling look at how AI is transforming industries and presents invaluable lessons for anyone looking to build and scale AI-driven solutions at large.

Manish Gupta, Senior Director at Google DeepMind, APAC, dives into the transformative potential of AI in India, focusing on healthcare and education. He discusses the challenges of language inclusivity in AI, particularly for Indian languages, and highlights initiatives like IndicGenBench. Gupta shares insights on how startups can leverage partnerships with DeepMind while navigating intellectual property concerns. He also emphasizes India's unique position in the global AI landscape and the potential for breakthroughs in scientific fields, including clean energy solutions.

India is making strides in artificial intelligence, blending academic rigor with real-world applications. In Episode 4 of Decoding AI, Prayank Swaroop, Partner at Accel, is joined by Professor Balaraman Ravindran of IIT Madras, a pioneering figure in Indian AI research and application. As the founding head of several prestigious AI-focused centers, including Wadhwani School of Data Science and AI, the Robert Bosch Centre for Data Science & AI, and the Centre for Responsible AI at IIT Madras, Professor Balaraman discusses how IIT Madras is driving innovation in AI research, language models, healthcare, and more. Key highlights include IIT Madras's work on Indian-language AI systems, such as translating NPTEL (National Programme on Technology Enhanced Learning) videos into multiple languages and collaborating with the Supreme Court to make legal documents accessible in regional languages. On the healthcare front, groundbreaking initiatives like Garbhini tackle maternal and fetal health challenges unique to India. Professor Balaraman also provides insights into how startups can engage with IIT Madras's vibrant research ecosystem. Looking ahead, he emphasizes the importance of solving uniquely Indian problems, urging startups and investors to focus on creating value within the country instead of catering to first-world markets.

Simplismart is an AI infrastructure company building the critical platform-agnostic middleware to deploy, train/customize, and serve open-source AI models in production. With a very small team and under $1M in funding, these two Bengaluru-based founders built the world's fastest inference engine. In Episode 3 of our Decoding AI series, Amritanshu Jain, Co-founder and CEO of Simplismart, joins Anagh Prasad to discuss the startup's remarkable pivot. Initially built as an AutoML platform, Simplismart's breakthrough came when they happened to discover the speed of their inference engine after a conversation with a potential customer. In this episode, Amritanshu also elaborates on the open-source versus closed-source debate, noting that the comparison is complex and requires significant education. He emphasises that it is not simply about declaring one model superior to the other but understanding the broader context and specific advantages each model offers. While open-source solutions can offer high customization and cost efficiency, the choice between open-source and closed-source models ultimately depends on the unique needs and constraints of the enterprise.

Vishwesh Pai, Senior Director of Product Management for AI Platform at ServiceNow, shares his journey leading AI initiatives from Amazon Alexa to ServiceNow. He discusses how generative AI evolves into a vital partner for enterprises, enhancing productivity while tackling ethical concerns. Vishwesh emphasizes AI safety and trust as cornerstones for enterprise success. The conversation also covers the opportunities for startups in the changing landscape and the importance of a customer-centric approach in SaaS, highlighting the need for collaboration and innovation.