3 Applied AI Pro Tips: Explore With Frontier Models, Scale With Small Ones
brandon_galang · x · 2026-09-30
Three increasingly niche applied-AI tips:
- Preserve your bandwidth with frontier models: intelligence and correctness compound — a smarter model wastes less time in circles. If token budget allows, taking an extra hour with a smaller model is a bad call; always start as large as you can when mapping the problem space.
- Productize with small pareto-optimal models: when scaling, flip the logic — get 80% of frontier performance at 1% of cost, paired with guardrails established during the frontier-model phase.
- Exploit the algorithm instead of letting it exploit you: ruthlessly curate your social feed with likes, comments and "not interested" so the recommender feeds you model testing and workflow content — the author credits 90% of their skills and current job to this habit.
More from coding & agent
- Structure-only AI audit: Jev predicts outcomes of 2,029 real calls at AUC 0.78 for $3 — alexcovo_eth · 2026-09-30
- Replicas V3 launches: run Claude Code and Codex in cloud VMs with BYO keys — KlausCodes · 2026-09-30
- Agno adds first-class MCP publishing for agent components — pritisinghhhh · 2026-09-30
- Dev on AI coding: don't grieve the craft — time was always the constraint — jh3yy · 2026-09-30
- Perplexity launches Automations in Perplexity Computer with event triggers and schedules — perplexity_ai · 2026-09-30
- Building a Codex-to-Flux 3 Pipeline That Autonomously Generates Ancient Greece Memes — venturetwins · 2026-09-30