The Daily Catch: Friday, October 23

Five things people are getting AI agents to do, scored and ready to copy.

Prep sales reps with AI call briefs and coaching, adding $1.5M+ ARR in a quarter

Blaze Score 61 with Gumloop

Gusto's go-to-market team built Gumloop agents that score sales calls, write pre-call briefs from Salesforce, prepare partnership reviews and flag churn risks each morning. Gumloop reports over $1.5M in added ARR in three months and a 31% higher win rate on deals prepared with the briefs.

Write a pre-call brief for my meeting with [company]. Using the CRM notes and recent public news below, include: who they are in two lines, what they bought or discussed before, likely goals and objections, three discovery questions, and a suggested talk track. Keep it to one page. CRM notes: [paste]. News: [paste].

Compare prices and reviews before a big purchase

Blaze Score 60 with Perplexity

An agent shortlists products that fit your needs and budget, compares current prices and the most common complaints in reviews, and tells you which to buy.

I want to buy a {{product}}. Must have: {{must}}. Nice to have: {{nice}}. Budget: {{budget}}.
Shortlist 3 options. For each: current price at 2 stores with links and the date checked, the 3 most common complaints in owner reviews, and who it is best for. Then pick one for me and say why. Mark any price you could not confirm today.

Let an AI model play and finish Pokemon Blue live on a stream

Blaze Score 60 with Gemini

Software engineer Joel Zhang built a harness that let Google's Gemini 2.5 Pro play Pokemon Blue on a Twitch stream. The model finished the game after about 37 days, which Google CEO Sundar Pichai shared publicly.

Write a Python harness that connects a Game Boy emulator (PyBoy) to the Gemini API. Each turn, capture a screenshot and read the player's map location from memory, send both to the model with a short goal, and let it reply with a JSON list of button presses. Keep a notes file the model can append to and read back. Log every turn to a file.

Build a small automation without code by describing it

Blaze Score 60 with Make, Zapier

Describe the job in one sentence and let an automation tool's built-in agent wire the trigger, steps and apps, then test it with one real example.

When {{trigger}} happens in {{app_a}}, I want {{action}} in {{app_b}}.
Build this as a workflow. Show me each step and what data passes between them before turning it on. Add a step that emails me if any step fails.

Plan a trip with a day-by-day itinerary and a packing list

Blaze Score 60 with Gemini

Tell an agent your dates, budget and what you like, and it plans each day with travel times, booking deadlines and a packing list for the forecast.

Plan a trip.
Dates: {{dates}}
Places: {{places}}
Budget: {{budget}}
Pace: {{pace}}
We love: {{interests}}

For each day: morning, afternoon and evening, with travel time between stops and a rain backup. Flag anything that must be booked ahead and by when. Check current opening days and say if you could not confirm them. End with a packing list based on the typical weather for those dates and a rough daily budget.

Want the full prompt packs and the weekly build session? Join Blaze School.