ChatGPT Is Not Enough Why AI Workflows Matter for Business | Ep 273 | DevReady Podcast

·33 min·2 clips

A strata workflow turns 50 or 60 documents into a report in 10 to 12 minutes.

1. DevReady Podcast episode 273 focuses on ChatGPT, AI workflows, and business automation. 2. Anthony hosts the conversation, and Nikhil Singh joins as a computer scientist with 20-plus years in software development and startup work. 3. The episode asks whether businesses need more than ChatGPT, and Nikhil argues that workflows, people, and processes matter more than a chat box. 4. Nikhil describes Decipher as a product for audio, video, and large documents that can generate summaries, quotes, blog posts, show notes, and audio reels. 5. He says the product started before the ChatGPT era, when the team wanted a way to decide what to read or listen to and then go deeper. 6. He explains that the early system used spaCy, NER models, knowledge graphs, clustering, and topic analysis before OpenAI APIs arrived. 7. Nikhil says the team later got early access to OpenAI APIs and moved to a hybrid model that combined GPT with spaCy and entity extraction. 8. He points out that AI has existed since the 1950s or 1970s, long before the current ChatGPT wave. 9. He also mentions earlier computer-vision work on contractor job photos, OpenCV, a fake Mars robot, and a recycling conveyor belt. 10. The recycling example used an overhead camera to identify bottles, caps, and translucency for better recycling analytics. 11. Nikhil says customer interviews with around 250 to 300 businesses shifted the work toward workflow customization and integrations. 12. He explains that many businesses need an AI layer around Decipher, CRM, ERP, or other systems instead of a standalone SaaS product. 13. He says discovery work now focuses on the onboarding journey, service journey, repetitive tasks, and two processes such as customer support and onboarding. 14. Nikhil says LLMs can add discretionary decision logic, not just rule-based steps, and that the workflow can build a knowledge base for the next use case. 15. He stresses that people, processes, and technology must align, because bad processes or the wrong attitudes make technology fail. 16. He compares AI adoption to training an employee, saying the system must be tuned to the business instead of working only from its general training. 17. The tone is practical and technical, with Anthony challenging claims about ChatGPT, Copilot, and so-called agentic workflows. 18. The format stays conversational, with back-and-forth examples from LinkedIn, customer support, coding agents, and video generation. 19. People building business automations, AI consultants, and operations teams will get the most from this episode. 20. People looking for hype about autonomous agents without process design may skip it.

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from DevReady Podcast