We found stuff AI is pretty good at

Episode Summary

Executive Summary: This VergeCast bonus episode explores practical, non-hype uses of AI in everyday life. The hosts and guests show AI working best as a planning, summarizing, and admin assistant for tasks like moving, calendars, search, bedtime routines, and game generation—while repeatedly stressing that outputs must be checked because AI can hallucinate or fail at shopping, pricing, and quotes.

Main Topics: AI for moving and financial planning (Priority: 5/5): Esther Cohen describes using ChatGPT to estimate take-home pay, taxes, utilities, gas, commute costs, and moving budgets after relocating from New York City to Miami Beach. The value came from reducing overwhelming logistical work and avoiding paid professional help. AI as a calendar and family logistics assistant (Priority: 5/5): Allison Johnson explains that AI is extremely useful for adding events to calendars from screenshots or emails, especially for complicated kid schedules like school events, while noting it was initially buggy but improved over time. AI for search and research with verification (Priority: 5/5): Jake Kastranakis argues AI chatbots can outperform keyword search when the query is nuanced or specific, but only if the user verifies citations and links because hallucinations remain a major risk. AI for brainstorming and project planning (Priority: 4/5): Jake also uses ChatGPT as a persistent brainstorming partner for apartment renovations and contractor estimates, treating it as a memory-keeping sounding board that helps catch omissions in plans and quotes. AI for generative admin in tabletop RPGs (Priority: 4/5): Travis Larson used Gemini to build a custom app that randomly generates game creatures and handles combat/admin rules for an original Pokémon-inspired RPG, showing AI can automate tedious rule execution while leaving creativity to the human. AI for bedtime stories and parenting hacks (Priority: 3/5): Allison shares a playful use case where AI generated custom bedtime stories, and also used AI to synthesize parenting advice into a clear bedtime 'playbook' to coordinate with her spouse. AI for astrology/self-soothing (Priority: 3/5): VSong describes using ChatGPT with saved natal chart data as a coping and reflection tool for astrology questions, explicitly framing it as a silly but personally helpful placebo-like habit rather than a serious belief system.

Key Arguments: AI is most useful when it removes tedious, rule-based labor rather than replacing human judgment or creativity. ChatGPT can be surprisingly accurate for structured calculations like taxes and take-home pay when prompts are specific enough. AI search is strongest for nuanced, natural-language queries that keyword search struggles to answer, especially when linked sources are available. Hallucination remains a real risk, so users must verify quotes, pricing, and factual claims before relying on results. AI works well as a personal sounding board or memory aid, especially for projects with many moving parts. Some of the best AI use cases are low-stakes and practical: calendars, budgeting, note synthesis, and admin automation. AI can support emotional coping or playfulness, but users should still maintain boundaries and not treat it as a substitute for professional help.

Data Points: Family size: 4 - Esther Cohen describes moving with her family of four from New York City to Miami Beach. Relocation: Upper West Side, New York City to Miami Beach, Florida - Used as the basis for AI-assisted budgeting and tax planning. First/second paycheck accuracy: Checked out down to the cent - Esther says the Florida take-home pay calculations were highly accurate after the move. Moving cost estimate range: $6,000 to $27,000 - AI gave a broad and unhelpful range when asked about moving a family of four in a three-bedroom apartment from Manhattan to Florida. Tabletop generator range: 1 to 4 spirits - Travis's Gemini-built app lets him generate one to four creatures at a time. Creature level range: 1 to 6 - The Crystal Spirits generator can produce creatures across six levels. Color-coded biomes: Visible light spectrum - Travis's RPG world uses biomes organized by the visible light spectrum. Creature database size: About 200 names - Gemini was fed a list of around 200 real animals and fictional creatures for randomized naming. Website index: Over 1 billion professionals - Mentioned in a LinkedIn Ads sponsor message describing the platform’s network size. Decision-maker count: 130 million - LinkedIn Ads sponsor message cites its own data on decision makers in the network. Wealthfront APY: 4% APY - Sponsor message for Wealthfront cash account. Wealthfront deposit bonus: $50 bonus with a $500 deposit - Sponsor message for Wealthfront first cash account offer.

Pivotal Quotes: "AI can be so open-ended, it's really hard for the average person to know what it's good for." — VSong: Opening framing for the episode’s premise about practical AI use cases. "The best scenarios I found for AI chatbots is just like thinking of it as me talking to myself in the mirror, but the mirror version of me has like a Google access and like a running list of everything that I said I wanted." — Jake Kastranakis: Describing AI as a searchable brainstorming partner rather than a magical answer machine. "the whole point is that I don't want the tools. I want ChatGPT to do the figuring out for me." — Esther Cohen: Explaining the moment she learned to prompt ChatGPT as a full decision-support assistant for move planning.

Implications: The episode suggests AI’s real value is in mundane, high-friction workflows: planning, searching, summarizing, and admin. For users and companies, the lesson is to build around verification, specificity, and augmentation—not hype or full automation.