Uploaded July 2026 | Updated September 2026, 2 weeks ago
Every AI agent starts every session with zero memory. No context, no history, just a blank slate that has to rediscover its environment from scratch, and that costs you real time and money.
In this episode of Inside the Stack, Jase Lindgren, Principal P4 User Advocate at Perforce, shows a simple fix: instead of re-prompting your agent every time, capture what it learns in a markdown file, then go one step further and turn that into a custom script. The result, tested across 15 runs: lower cost, faster execution, and better accuracy at catching real issues.
00:00 The problem: AI agents have no memory
01:30 Naive prompting run
03:00 Adding a markdown instructions file
05:00 Building a custom script
07:00 Results across 15 runs
Perforce | Inside the Stack
Learn More & Follow Us:
➡️ perforce.com
➡️ linkedin.com/company/perforce
#AI #AICosts #DevOps #Perforce
Every AI agent starts every session with zero memory. No context, no history, just a blank slate that has to rediscover its environment from scratch, and that costs you real time and money.
In this episode of Inside the Stack, Jase Lindgren, Principal P4 User Advocate at Perforce, shows a simple fix: instead of re-prompting your agent every time, capture what it learns in a markdown file, then go one step further and turn that into a custom script. The result, tested across 15 runs: lower cost, faster execution, and better accuracy at catching real issues.
00:00 The problem: AI agents have no memory
01:30 Naive prompting run
03:00 Adding a markdown instructions file
05:00 Building a custom script
07:00 Results across 15 runs
Perforce | Inside the Stack
Learn More & Follow Us:
➡️ perforce.com
➡️ linkedin.com/company/perforce
#AI #AICosts #DevOps #Perforce
