TickerTain — Make Analysis Fun Again

←

▶ 8:54 · AI Infrastructure · Production model-serving infrastructure (routing, fallbacks, evals) is a prerequisite for effective post-training

now playing · AI Infrastructure

AI Applicationstailwindscore 9/10gabe pereyra

Application-layer AI companies can rival frontier labs by orchestrating the frontier ecosystem

Gabe argues that vertical application companies no longer need to build everything in-house; by leveraging open-source base models, NeMo lab partnerships, synthetic data pipelines, and mode…

Synthetic Data & AI Trainingtailwindscore 8/10gabe pereyra

Domain-expert-guided synthetic data solves the vertical AI data privacy bottleneck

Vertical AI companies blocked from training on sensitive customer data can use domain experts (lawyers, doctors) to guide synthetic data generation via coding models, then scale with platfo…

AI Infrastructuretailwindscore 8/10gabe pereyra

Domain-expert-guided synthetic data generation unlocks training for data-sensitive verticals

In domains like legal where customer data is privileged, using domain experts to guide synthetic data creation (via tools like Mercor, Snorkel) solves the cold-start problem and enables rig…

Open Source AItailwindscore 8/10gabe pereyra

Post-training open-source models has become viable and cost-effective for domain-specific frontier intelligence

With strong open-source base models (GLM, Kimi, NeMo-Megatron) and accessible post-training APIs (Fireworks, Baseten, Tinker), application companies can now build specialized models that co…

Open Source AItailwindscore 7/10gabe pereyra

Open-source base models now competitive enough for domain-specific post-training to reach frontier performance

Models like Kimi 3, GLM 5.2, NeMo-Megatron, and Inkling have reached sufficient base capability that post-training on domain-specific data can yield frontier-level performance on narrow tas…

AI Infrastructuretailwindscore 8/10gabe pereyra

Production model-serving infrastructure (routing, fallbacks, evals) is a prerequisite for effective post-training

Before investing in post-training, companies must build robust model-serving infrastructure including multi-model routing, automated/human evaluation gates, A/B testing, and production moni…

AI Economics & Business Modelsmixedscore 7/10gabe pereyra

Continual learning on private customer data—without training on it—is the endgame for vertical AI

The ultimate product is not a single best model but enabling each enterprise to continuously customize models on their private work streams while preserving data privacy, requiring breakthr…

Enterprise AI Adoptiontailwindscore 7/10gabe pereyra

Vertical AI wins by orchestrating organizational productivity, not individual productivity

Horizontal tools like Coda focus on individual workflows, but vertical AI (e.g., legal) must solve organizational coordination—managing multi-month projects with 20-30 person teams, resourc…

Enterprise AI Adoptiontailwindscore 8/10gabe pereyra

Vertical AI wins by going hyper-vertical into organizational workflows, not individual productivity

Horizontal tools (Coda, Notion) optimize individual productivity; vertical AI wins by orchestrating organizational workflows — multi-person, multi-month projects, resource allocation across…

←

▶ 8:54 · AI Infrastructure · Production model-serving infrastructure (routing, fallbacks, evals) is a prerequisite for effective post-training

now playing · AI Infrastructure

AI Applicationstailwindscore 9/10gabe pereyra

Application-layer AI companies can rival frontier labs by orchestrating the frontier ecosystem

Gabe argues that vertical application companies no longer need to build everything in-house; by leveraging open-source base models, NeMo lab partnerships, synthetic data pipelines, and mode…

Synthetic Data & AI Trainingtailwindscore 8/10gabe pereyra

Domain-expert-guided synthetic data solves the vertical AI data privacy bottleneck

Vertical AI companies blocked from training on sensitive customer data can use domain experts (lawyers, doctors) to guide synthetic data generation via coding models, then scale with platfo…

AI Infrastructuretailwindscore 8/10gabe pereyra

Domain-expert-guided synthetic data generation unlocks training for data-sensitive verticals

In domains like legal where customer data is privileged, using domain experts to guide synthetic data creation (via tools like Mercor, Snorkel) solves the cold-start problem and enables rig…

Open Source AItailwindscore 8/10gabe pereyra

Post-training open-source models has become viable and cost-effective for domain-specific frontier intelligence

With strong open-source base models (GLM, Kimi, NeMo-Megatron) and accessible post-training APIs (Fireworks, Baseten, Tinker), application companies can now build specialized models that co…

Open Source AItailwindscore 7/10gabe pereyra

Open-source base models now competitive enough for domain-specific post-training to reach frontier performance

Models like Kimi 3, GLM 5.2, NeMo-Megatron, and Inkling have reached sufficient base capability that post-training on domain-specific data can yield frontier-level performance on narrow tas…

AI Infrastructuretailwindscore 8/10gabe pereyra

Production model-serving infrastructure (routing, fallbacks, evals) is a prerequisite for effective post-training

Before investing in post-training, companies must build robust model-serving infrastructure including multi-model routing, automated/human evaluation gates, A/B testing, and production moni…

AI Economics & Business Modelsmixedscore 7/10gabe pereyra

Continual learning on private customer data—without training on it—is the endgame for vertical AI

The ultimate product is not a single best model but enabling each enterprise to continuously customize models on their private work streams while preserving data privacy, requiring breakthr…

Enterprise AI Adoptiontailwindscore 7/10gabe pereyra

Vertical AI wins by orchestrating organizational productivity, not individual productivity

Horizontal tools like Coda focus on individual workflows, but vertical AI (e.g., legal) must solve organizational coordination—managing multi-month projects with 20-30 person teams, resourc…

Enterprise AI Adoptiontailwindscore 8/10gabe pereyra

Vertical AI wins by going hyper-vertical into organizational workflows, not individual productivity

Horizontal tools (Coda, Notion) optimize individual productivity; vertical AI wins by orchestrating organizational workflows — multi-person, multi-month projects, resource allocation across…