What SaaStr 2025 taught us about AI-native GTM (and what’s already obsolete)
- BADideas.fund
- May 20
- 3 min read
If SaaStr 2025 had a headline, it’d be this: “AI isn’t a feature. It’s a forcing function.” From OpenAI to HubSpot, Snowflake to Mosaic, the smartest operators aren’t asking how to use AI — they’re rebuilding teams, playbooks, and products around it.
We attended seven of the most tactical sessions — and here’s the distilled playbook for founders, sales leaders, and GTM teams trying to stay ahead.
Delphi: Context > Content in the AI Era
AI is for building “digital minds” that reflect your team’s actual decision logic.
Key Takeaways:
Generic advice is dead. Delphi simulated 1M founder conversations — every insight depended on company stage, market, and customer type.
Vision + gut still matter. Great startups start with intuition, not dashboards.
AI-native UX = fewer buttons, more prompts. Fewer teams, more leverage.
Final mantra: “Your people are your IP.” Strategy can be cloned — context can’t.
HubSpot: AI Is Reshaping the Org, Not Just the Tools
Yamini Rangan made it clear: AI is production-grade at HubSpot. 95% of engineers use tools like Cursor daily.
“GTM now means human + agent collaboration — and seamless handoffs are the new trust currency.”
What’s changing:
Support: AI resolves up to 80% of tier-one tickets. Knowledge base quality is the #1 driver.
Engineering: Velocity is up — but so are expectations. Output = business impact.
Outreach: “Personalized” is dead. If your AI sends a generic email, it’s deleted in 10 seconds.
Mosaic.ai: Multimodal, Multilingual, Multi-Variant — By Default
Mosaic is building a video editing agent that handles global, dynamic content in one pass.
Why it matters:
Edit with natural language (“cut all clips with the guy in glasses”).
Instant localization via voice cloning + dubbing.
Branch-based editing: like Git, but for video assets.
The next generation of AI-native tools won’t offer “automation” — they’ll offer scale, leverage, and creative freedom.
Snowflake: From Data Warehouse to AI-Native GTM Engine
Snowflake’s real move? Enabling conversational access to any enterprise dataset.
“Customers don’t pay for databases. They pay for insights.”
The GTM stack they’re betting on:
Embedded co-sell motions (e.g., Observe built fully on Snowflake; customers “pay” in credits).
Usage-based sales comp — reps only win when data flows.
Reps as value architects, not just contract closers.
OpenAI: How to Actually Build a Sales Org
Maggie Hott broke it down for founders:
“One exceptional hire is better than three average ones. Every time.”
Top lessons:
Hire for impact, not enthusiasm. Passion is not a plan.
Founders must own sales until there’s a repeatable motion.
Avoid “logo hires.” Ask what they built, not where they worked.
Look for chaos translators — generalists who thrive in ambiguity.
Kyle Norton: Sales Is Getting an AI Teammate
Note-takers are the past. AI co-sellers are the future:
“It’s not just Digital Kyle — it’s a better one. Because it knows everything.”
What’s coming:
Agents that sit in Zoom, correct errors, answer questions, and augment reps live.
CROs who manage human + agent hybrid teams.
Always-on assistants that understand the product better than your team.
Jason Lemkin’s Tough Love Sermon
If you’re still running 2021 playbooks — you’ve already lost:
“Remote + AI ≠ less work. It means 6.5-day weeks with compounding output.”
Hard truths:
30–40% of SaaS teams aren’t re-skillable.
Moats are weaker, momentum wins.
VCs are chasing $10B+ outcomes. You’ll need 3x more proof, with 70% fewer funders.
🧠 Closing Frameworks for SaaStr 2025
Refound your company. If you were starting now, what would you build?
Don’t just adopt AI — operationalize it.
Measure outcomes, not engagement.
Make AI native — not an add-on.
Because in 2025, you’re not competing on product. You’re competing on execution speed, context depth, and whether your team can build with AI — not just talk about it.