StarZero All shows Try the podcast clipper →

Feb 18, 2026 · 1h 0m · saastr

The 90/10 Rule for AI Agents: What to Vibe Code, and What To Just Buy

Jason Lemkin · 14m spoken
0:00 / 0:00
▶ Watch on YouTube →

gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions

In this live SaaStr session, Jason Lemkin and Amelia present the '90/10 Rule' for software adoption, guiding founders on when to purchase commercial SaaS versus when to vibe code custom AI agents while addressing maintenance and security realities.

How this conversation actually went

Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Jason holds 25.8% of the talking time here. How this is scored →

Jason as informed peer 5.9 Guest teaching 1.0 Guest disagreement 0.1 Jason pushing back 0.5
05100:0015:0030:0045:001:00:001:38–8:01 · Jason as informed peer 5/10 Announcement: SaaStr Annual and AI Summit 2026 Jason introduces the event and context before handing off to Amelia, who establishes the 90/10 rule for building versus buying software. Jason chimes in with domain analysis on niche 'N equals one' vertical apps falling behind in AI innovation.8:02–15:32 · Jason as informed peer 4/10 Evaluating Build vs. Buy and Vibe Coding Priorities Amelia leads a presentation on why building an app yourself is risky without boundaries like SSO limits and reviews survey data on common vibe-coding use cases. The dynamic is collaborative with Amelia presenting internal workflow data.15:32–24:03 · Jason as informed peer 7/10 The Catalyst of Claude Cowork and Zero-AI SaaS Replacement Amelia details using Claude Cowork to replace legacy SaaS tools lacking AI. Jason forcefully asserts that any SaaS company lacking a jaw-dropping AI feature in 2026 should be embarrassed and faces imminent disruption.24:03–28:26 · Jason as informed peer 3/10 Case Study 1: Building the Internal AI VP of Marketing Amelia explains the technical architecture and day-to-day operations of their internal AI VP of Marketing built on Salesforce, Zapier, and Replit. The exchange is an informative walkthrough of operational mechanics.28:26–38:25 · Jason as informed peer 4/10 Case Study 2: Launching the Customer-Facing SaaStr Sponsors Portal Amelia breaks down the step-by-step process of using Claude Cowork and Clerk to build their external sponsor portal, automating contract parsing and ticket code generation that previously required human labor.38:26–46:43 · Jason as informed peer 7/10 The True Cost and Maintenance Dilemma of Vibe Coding Amelia and Jason discuss the hidden maintenance drag of custom vibe-coded apps. Jason calculates the economic valuation of their time at $1,000 to $2,000 per hour, concluding that software cost savings alone never justify building internal tools.46:43–54:40 · Jason as informed peer 9/10 Q&A: SaaS Market Valuation Realities and Monaco AI SDR Addressing audience Q&A, Jason delivers an in-depth analysis of public SaaS valuation contractions (Shopify, Atlassian trading at 4x ARR) and explains why companies must capture AI budgets with jaw-dropping utility to survive.54:40–59:11 · Jason as informed peer 8/10 Q&A: Autonomous Guardrails, Security Hardening, and Bandwidth Limits Jason dismisses hype around unconstrained autonomous agents like OpenClaw in favor of strict guardrails, and shares pragmatic security advice on utilizing LLM audits and managing team operational bandwidth.1:38–8:01 · Guest teaching 1/10 Announcement: SaaStr Annual and AI Summit 2026 Jason introduces the event and context before handing off to Amelia, who establishes the 90/10 rule for building versus buying software. Jason chimes in with domain analysis on niche 'N equals one' vertical apps falling behind in AI innovation.8:02–15:32 · Guest teaching 2/10 Evaluating Build vs. Buy and Vibe Coding Priorities Amelia leads a presentation on why building an app yourself is risky without boundaries like SSO limits and reviews survey data on common vibe-coding use cases. The dynamic is collaborative with Amelia presenting internal workflow data.15:32–24:03 · Guest teaching 1/10 The Catalyst of Claude Cowork and Zero-AI SaaS Replacement Amelia details using Claude Cowork to replace legacy SaaS tools lacking AI. Jason forcefully asserts that any SaaS company lacking a jaw-dropping AI feature in 2026 should be embarrassed and faces imminent disruption.24:03–28:26 · Guest teaching 2/10 Case Study 1: Building the Internal AI VP of Marketing Amelia explains the technical architecture and day-to-day operations of their internal AI VP of Marketing built on Salesforce, Zapier, and Replit. The exchange is an informative walkthrough of operational mechanics.28:26–38:25 · Guest teaching 1/10 Case Study 2: Launching the Customer-Facing SaaStr Sponsors Portal Amelia breaks down the step-by-step process of using Claude Cowork and Clerk to build their external sponsor portal, automating contract parsing and ticket code generation that previously required human labor.38:26–46:43 · Guest teaching 1/10 The True Cost and Maintenance Dilemma of Vibe Coding Amelia and Jason discuss the hidden maintenance drag of custom vibe-coded apps. Jason calculates the economic valuation of their time at $1,000 to $2,000 per hour, concluding that software cost savings alone never justify building internal tools.46:43–54:40 · Guest teaching 0/10 Q&A: SaaS Market Valuation Realities and Monaco AI SDR Addressing audience Q&A, Jason delivers an in-depth analysis of public SaaS valuation contractions (Shopify, Atlassian trading at 4x ARR) and explains why companies must capture AI budgets with jaw-dropping utility to survive.54:40–59:11 · Guest teaching 0/10 Q&A: Autonomous Guardrails, Security Hardening, and Bandwidth Limits Jason dismisses hype around unconstrained autonomous agents like OpenClaw in favor of strict guardrails, and shares pragmatic security advice on utilizing LLM audits and managing team operational bandwidth.1:38–8:01 · Guest disagreement 0/10 Announcement: SaaStr Annual and AI Summit 2026 Jason introduces the event and context before handing off to Amelia, who establishes the 90/10 rule for building versus buying software. Jason chimes in with domain analysis on niche 'N equals one' vertical apps falling behind in AI innovation.8:02–15:32 · Guest disagreement 0/10 Evaluating Build vs. Buy and Vibe Coding Priorities Amelia leads a presentation on why building an app yourself is risky without boundaries like SSO limits and reviews survey data on common vibe-coding use cases. The dynamic is collaborative with Amelia presenting internal workflow data.15:32–24:03 · Guest disagreement 0/10 The Catalyst of Claude Cowork and Zero-AI SaaS Replacement Amelia details using Claude Cowork to replace legacy SaaS tools lacking AI. Jason forcefully asserts that any SaaS company lacking a jaw-dropping AI feature in 2026 should be embarrassed and faces imminent disruption.24:03–28:26 · Guest disagreement 0/10 Case Study 1: Building the Internal AI VP of Marketing Amelia explains the technical architecture and day-to-day operations of their internal AI VP of Marketing built on Salesforce, Zapier, and Replit. The exchange is an informative walkthrough of operational mechanics.28:26–38:25 · Guest disagreement 0/10 Case Study 2: Launching the Customer-Facing SaaStr Sponsors Portal Amelia breaks down the step-by-step process of using Claude Cowork and Clerk to build their external sponsor portal, automating contract parsing and ticket code generation that previously required human labor.38:26–46:43 · Guest disagreement 1/10 The True Cost and Maintenance Dilemma of Vibe Coding Amelia and Jason discuss the hidden maintenance drag of custom vibe-coded apps. Jason calculates the economic valuation of their time at $1,000 to $2,000 per hour, concluding that software cost savings alone never justify building internal tools.46:43–54:40 · Guest disagreement 0/10 Q&A: SaaS Market Valuation Realities and Monaco AI SDR Addressing audience Q&A, Jason delivers an in-depth analysis of public SaaS valuation contractions (Shopify, Atlassian trading at 4x ARR) and explains why companies must capture AI budgets with jaw-dropping utility to survive.54:40–59:11 · Guest disagreement 0/10 Q&A: Autonomous Guardrails, Security Hardening, and Bandwidth Limits Jason dismisses hype around unconstrained autonomous agents like OpenClaw in favor of strict guardrails, and shares pragmatic security advice on utilizing LLM audits and managing team operational bandwidth.1:38–8:01 · Jason pushing back 0/10 Announcement: SaaStr Annual and AI Summit 2026 Jason introduces the event and context before handing off to Amelia, who establishes the 90/10 rule for building versus buying software. Jason chimes in with domain analysis on niche 'N equals one' vertical apps falling behind in AI innovation.8:02–15:32 · Jason pushing back 0/10 Evaluating Build vs. Buy and Vibe Coding Priorities Amelia leads a presentation on why building an app yourself is risky without boundaries like SSO limits and reviews survey data on common vibe-coding use cases. The dynamic is collaborative with Amelia presenting internal workflow data.15:32–24:03 · Jason pushing back 1/10 The Catalyst of Claude Cowork and Zero-AI SaaS Replacement Amelia details using Claude Cowork to replace legacy SaaS tools lacking AI. Jason forcefully asserts that any SaaS company lacking a jaw-dropping AI feature in 2026 should be embarrassed and faces imminent disruption.24:03–28:26 · Jason pushing back 0/10 Case Study 1: Building the Internal AI VP of Marketing Amelia explains the technical architecture and day-to-day operations of their internal AI VP of Marketing built on Salesforce, Zapier, and Replit. The exchange is an informative walkthrough of operational mechanics.28:26–38:25 · Jason pushing back 0/10 Case Study 2: Launching the Customer-Facing SaaStr Sponsors Portal Amelia breaks down the step-by-step process of using Claude Cowork and Clerk to build their external sponsor portal, automating contract parsing and ticket code generation that previously required human labor.38:26–46:43 · Jason pushing back 1/10 The True Cost and Maintenance Dilemma of Vibe Coding Amelia and Jason discuss the hidden maintenance drag of custom vibe-coded apps. Jason calculates the economic valuation of their time at $1,000 to $2,000 per hour, concluding that software cost savings alone never justify building internal tools.46:43–54:40 · Jason pushing back 0/10 Q&A: SaaS Market Valuation Realities and Monaco AI SDR Addressing audience Q&A, Jason delivers an in-depth analysis of public SaaS valuation contractions (Shopify, Atlassian trading at 4x ARR) and explains why companies must capture AI budgets with jaw-dropping utility to survive.54:40–59:11 · Jason pushing back 2/10 Q&A: Autonomous Guardrails, Security Hardening, and Bandwidth Limits Jason dismisses hype around unconstrained autonomous agents like OpenClaw in favor of strict guardrails, and shares pragmatic security advice on utilizing LLM audits and managing team operational bandwidth.

speaking balance: gold is Jason, purple is the guest (3 minute bins)

0:00 · Jason 62.4% · guest 37.6%0:00 · Jason 62.4% · guest 37.6%3:00 · Jason 0% · guest 100%3:00 · Jason 0% · guest 100%6:00 · Jason 46.8% · guest 53.2%6:00 · Jason 46.8% · guest 53.2%9:00 · Jason 0% · guest 100%9:00 · Jason 0% · guest 100%12:00 · Jason 0% · guest 100%12:00 · Jason 0% · guest 100%15:00 · Jason 0% · guest 100%15:00 · Jason 0% · guest 100%18:00 · Jason 0% · guest 100%18:00 · Jason 0% · guest 100%21:00 · Jason 34.6% · guest 65.4%21:00 · Jason 34.6% · guest 65.4%24:00 · Jason 0% · guest 100%24:00 · Jason 0% · guest 100%27:00 · Jason 0% · guest 100%27:00 · Jason 0% · guest 100%30:00 · Jason 0% · guest 100%30:00 · Jason 0% · guest 100%33:00 · Jason 0% · guest 100%33:00 · Jason 0% · guest 100%36:00 · Jason 0% · guest 100%36:00 · Jason 0% · guest 100%39:00 · Jason 0% · guest 100%39:00 · Jason 0% · guest 100%42:00 · Jason 48.9% · guest 51.1%42:00 · Jason 48.9% · guest 51.1%45:00 · Jason 69.2% · guest 30.8%45:00 · Jason 69.2% · guest 30.8%48:00 · Jason 100% · guest 0%48:00 · Jason 100% · guest 0%51:00 · Jason 35.3% · guest 64.7%51:00 · Jason 35.3% · guest 64.7%54:00 · Jason 63.9% · guest 36.1%54:00 · Jason 63.9% · guest 36.1%57:00 · Jason 68.5% · guest 31.5%57:00 · Jason 68.5% · guest 31.5%1:00:00 · Jason 0% · guest 100%1:00:00 · Jason 0% · guest 100%
Sharpest disagreement ▶ 45:15 Amelia and Jason rejecting the idea of white-labeling internal apps

When audience members suggest commercializing and white-labeling their custom portal, Amelia and Jason firmly reject the idea, emphasizing the immense customer support burden.

 Hardest push from Jason ▶ 54:53 Jason rejecting hype around unconstrained autonomous agents

Jason refuses the premise that rogue autonomous agents are useful for GTM teams, emphasizing that rigorous guardrails are far more critical than uncontrolled experimentation.

 Biggest teaching moment ▶ 32:20 Amelia explaining SSO limitations and authentication workarounds

Amelia educates listeners and the room on why naive vibe coding fails at persistent authentication, walking through the technical necessity of integrating Clerk for true SSO.

 Jason holds their own ▶ 47:40 Jason breaking down public SaaS multiple compression

Jason showcases deep domain knowledge by analyzing Shopify's post-earnings drop despite 30% growth and Atlassian's compression to 4x ARR to prove why incremental SaaS products face extinction.
the scores for every segment, with the reasoning behind each
ChapterTopicJason as informed peerGuest teachingGuest disagreementJason pushing backWhy
Announcement: SaaStr Annual and AI Summit 2026 5100 Jason introduces the event and context before handing off to Amelia, who establishes the 90/10 rule for building versus buying software. Jason chimes in with domain analysis on niche 'N equals one' vertical apps falling behind in AI innovation.
Evaluating Build vs. Buy and Vibe Coding Priorities 4200 Amelia leads a presentation on why building an app yourself is risky without boundaries like SSO limits and reviews survey data on common vibe-coding use cases. The dynamic is collaborative with Amelia presenting internal workflow data.
The Catalyst of Claude Cowork and Zero-AI SaaS Replacement 7101 Amelia details using Claude Cowork to replace legacy SaaS tools lacking AI. Jason forcefully asserts that any SaaS company lacking a jaw-dropping AI feature in 2026 should be embarrassed and faces imminent disruption.
Case Study 1: Building the Internal AI VP of Marketing 3200 Amelia explains the technical architecture and day-to-day operations of their internal AI VP of Marketing built on Salesforce, Zapier, and Replit. The exchange is an informative walkthrough of operational mechanics.
Case Study 2: Launching the Customer-Facing SaaStr Sponsors Portal 4100 Amelia breaks down the step-by-step process of using Claude Cowork and Clerk to build their external sponsor portal, automating contract parsing and ticket code generation that previously required human labor.
The True Cost and Maintenance Dilemma of Vibe Coding 7111 Amelia and Jason discuss the hidden maintenance drag of custom vibe-coded apps. Jason calculates the economic valuation of their time at $1,000 to $2,000 per hour, concluding that software cost savings alone never justify building internal tools.
Q&A: SaaS Market Valuation Realities and Monaco AI SDR 9000 Addressing audience Q&A, Jason delivers an in-depth analysis of public SaaS valuation contractions (Shopify, Atlassian trading at 4x ARR) and explains why companies must capture AI budgets with jaw-dropping utility to survive.
Q&A: Autonomous Guardrails, Security Hardening, and Bandwidth Limits 8002 Jason dismisses hype around unconstrained autonomous agents like OpenClaw in favor of strict guardrails, and shares pragmatic security advice on utilizing LLM audits and managing team operational bandwidth.

Statements from this episode (6)

Prediction Not checkable as stated
Lemkin: Mediocre Vertical SaaS Apps Risk Losing Customers Permanently to Vibe Coding
“I think there is a risk for folks building vertical apps, building niche apps. You should not build an agent if you can buy it or an app, but if you fall behind too far folks may build it out of frustration. And that's more what we did rather than try to cance…”
Jason Lemkin Feb 18, 2026 ▶ 7:30
Prediction Not checkable as stated
Lemkin: Software Companies Lacking Magical AI Will Be Quickly Disrupted
“Cuz there's really no excuse today and you're gonna be quickly disrupted and churned if you can't build something magical with your AI app today.”
Jason Lemkin Feb 18, 2026 ▶ 22:28
Insight
Lemkin: Vibe Coding Apps to Save SaaS Fees Makes No Economic Sense
“So there's no economic justification for software costs to building these apps, right? You can buy them for 10 grand, like it makes no economic sense. So you have to really need something that you can't buy today.”
Jason Lemkin Feb 18, 2026 ▶ 44:05
Assertion Supported
Lemkin: Atlassian Re-Accelerated at $5B Revenue but Trades at 4x ARR
“Atlassian beat the quarter. It's re-accelerated at five billion in revenue and it trades at Forex ARR.”
Jason Lemkin Feb 18, 2026 ▶ 48:21
Insight
Lemkin: AI Agents Perform Better Security Audits Than Outsourced Dev Shops
“What you should do with the agents now is they're pretty good. You can do this in Claude code. You can do it in Replit. You can do it in Lovable is ask them to do the deepest security audit possible on your app at least once every month or two. And they're pre…”
Jason Lemkin Feb 18, 2026 ▶ 56:50
Opinion
Lemkin: Only 1% to 5% of Employees Can Effectively Manage AI Agents
“I think honestly, you're probably lucky a five to one percent to five percent of your team could We could offload the management we have.”
Jason Lemkin Feb 18, 2026 ▶ 58:39
Made with StarZero

Turn any episode into a week of clips.

This entire site, over 400 episodes transcribed, diarized, checked and made playable, runs on the StarZero media pipeline. Drop in your own episode and the podcast clipper finds the moments worth sharing, cuts them, captions them, and reframes them for every feed.

Try the podcast clipper Learn more