nthypes

After selling to mid-market customers, I've become convinced that multi-tenant SaaS architecture is the wrong model for anything beyond PLG at low price points.

Enterprise procurement (and some mid-market) wants dedicated infrastructure. Full stop. Security questionnaires are brutal on shared tenancy Compliance (SOC2, HIPAA, FedRAMP) is infinitely easier with isolation

"You share infrastructure with our competitors?" kills deals

K8s namespaces give you 80% of single-tenant benefits Resource isolation, blast radius containment, independent scaling Customer-specific customization becomes trivial Noisy neighbor problems disappear

Sales-led growth with FDE (forward deployment engineers) commands premium pricing Implementation fees offset infrastructure costs Retention is higher when customers feel they have "their own thing"

IMHO, where multi-tenant still makes sense:

Product-led growth at <$10K ACV True horizontal SaaS with zero customization You're optimizing for scaling to millions of users, not hundreds of mid-market customers

Am I off base here? What am I missing? Would love to hear from folks running B2B SaaS at scale.

I've been thinking about how vibrant web forums used to be in the 2000s and early 2010s - specialized communities with deep expertise, long-running threads, inside jokes, and real relationships. Places like Something Awful, vBulletin forums for every niche hobby, phpBB boards for gaming clans, even early Reddit. Now most of that energy has moved to Discord servers, subreddits, or died entirely. The few surviving forums feel like ghost towns.

What killed forum culture? Was this inevitable, or did we lose something valuable? Are there examples of forums that are thriving today?

I've been trying out Google's new AI Mode in Search and I'm genuinely curious about the technical architecture behind it. The response times are incredibly fast - often sub-second - and the quality of answers seems consistently high.

What's particularly impressive: - Speed: Near-instant responses even for complex queries - Quality: Accurate, well-sourced answers with citations - Integration: Seamlessly pulls from the knowledge graph and fresh web results

I'm wondering: - What model(s) are they running under the hood? - How are they achieving such low latency at scale? - Are they using some kind of speculative execution or caching strategy? - How does their infrastructure differ from standalone LLM APIs?

For those who've worked on similar systems or have insights into Google's approach, I'd love to hear your thoughts on what makes this possible.