Archive
The full Signal archive
Every dispatch, ordered by date. Filter by section above, or use the cluster index below to browse by topic.
The Post-SaaS Era: Why Vertical AI is Eating the Horizontal Giants
Seat-based pricing is fracturing. Agentic AI is replacing manual workflows, and outcome-based contracts are reshaping the next decade of business software.
Modern Web Application Architecture in 2026
The default stack has shifted. Edge, streaming, server components, and typed RPC are now mainstream.
The Invisible Layer: How LLM Middleware is Capturing AI Value
Beyond foundational models, a new class of orchestration software is defining the unit economics of generative AI.
MVP Scope Discipline: The 4-Week Rule
If you cannot ship the first cut in four weeks, your MVP is not minimal.
Edge Compute vs Region Compute: A Practical Decision Framework
When the edge actually wins, and when it just adds latency to your database round trip.
Agentic Workflows in Production: What Actually Breaks
Agent demos are easy. Agent reliability is not. A field report on what fails first.
The End of the Seat-Based Licensing Model
Usage-based and outcome-based pricing are no longer outliers — they are the survival mechanism.
The Founder-to-Product-Team Handoff
When the founder stops being the product manager — and how to do that without losing the thread.
Postgres at Scale: The 2026 Playbook
Read replicas, connection pooling, partitioning, and the moment you actually need to shard.
RAG Architecture Patterns That Actually Scale
Naive retrieval-augmented generation falls apart at moderate scale. Here's what production RAG looks like in 2026.
Dry Powder and the Efficiency Mandate
The metrics VCs actually care about in the post-growth-at-all-costs era.
Scaling an Engineering Team Without Breaking It
The well-documented hiring trap from 8 to 30 engineers, and how to avoid the rebuild.
Monorepo vs Polyrepo: The Honest Tradeoffs
A pragmatic framework for choosing, with the operational costs nobody talks about.
Prompt Injection Is the New SQL Injection
And we are about ten years behind where we should be on defenses.
The Burn Multiple Is the Metric That Won
Why a single ratio has displaced growth as the dominant late-stage financing signal.
When to Migrate Off No-Code
A pragmatic framework for founders running production software on no-code tools.
Kubernetes: When Not To
K8s solves real problems. Most early-stage teams do not have those problems.
Why Small Models Are Quietly Eating Large Ones
Frontier capability is converging. Cost and latency are not. The economics now favor specialization.
Net Dollar Retention: What 120% Actually Means
The metric is real. The way it is reported is frequently not.
Outsourced Development: When It Actually Works
The honest pattern of successful engagement, and the patterns that fail every time.
End-to-End Type Safety: Worth It or Cargo Cult?
tRPC, GraphQL codegen, server functions — the patterns and where they pay off.
AI Coding Assistants in 2026: Honest Productivity Numbers
Cutting through the demo theater to look at what AI tooling actually does to engineering velocity.
Early-Stage Pricing Experiments That Did Not Backfire
How a handful of seed-stage companies tested aggressive pricing without burning their pipeline.
The Onboarding Math That Decides SaaS Conversion
Why time-to-first-value is the only activation metric that consistently predicts retention.
Observability for Small Teams Without an SRE
The minimum viable observability stack — logs, metrics, traces — for product teams without dedicated platform engineers.
Open Source LLMs in the Enterprise: A 2026 Field Guide
When self-hosting pays off, when it doesn't, and what the real total cost looks like.
AI Cost of Goods Sold: The New SaaS Margin Problem
Why AI-first companies are reporting gross margins SaaS investors have not seen in fifteen years.
Five Pricing Page Mistakes That Tank Conversion
A practical audit of the patterns that demonstrably reduce trial starts and paid conversions.
Multi-Tenant Architecture: Pool, Silo, or Hybrid
The three canonical models, what they cost to operate, and the one decision you have to make early.
Evaluating LLM Quality Without Vibes
Why most AI product teams ship blind, and the eval discipline a few have built.
SaaS Valuation Multiples: The Reset Is Permanent
The 2021 multiples were not a baseline. They were an anomaly we are not returning to.
From Founder-Led Sales to a Real Go-to-Market
The transition is harder than the hire — and the hire is hard.
Modern Background Jobs: Beyond cron and Redis Queues
Durable execution, workflow engines, and what changed when 'idempotent' became table stakes.
Fine-Tuning vs Prompting: The 2026 Verdict
When fine-tuning is worth the operational tax — and when prompting plus retrieval will outperform it.
The Rule of 40 Revisited
The classic heuristic still works — but its components have shifted in importance.
Developer Marketing That Does Not Insult Developers
The patterns that work for developer-first companies in 2026.
Feature Flag Discipline for Fast-Moving Teams
How the highest-velocity teams treat feature flags as a release primitive, not a hack.
The Pricing Crisis in AI-First Products
Token costs are variable. Customers expect predictability. The math is breaking margins.
Secondary Markets and the Retention Problem
How structured secondaries are quietly solving the equity-retention problem at growth-stage companies.
Product-Market Fit Signals You Can Actually Measure
Beyond the Sean Ellis test: the operational signals that track with PMF.
Authentication in 2026: Passkeys, Magic Links, OAuth
The pragmatic choice for new applications, and the migration path off legacy passwords.
AI Agents in Customer Support: The Real Performance Numbers
What three years of production deployment tells us about deflection rates, escalations, and CSAT.
RSUs vs Options at Late-Stage Startups
Why an increasing number of growth-stage companies are switching equity instruments mid-life.
Running a Design Partner Program That Actually Helps
How early-stage teams structure paid pilots to learn fast without selling the company short.
Serverless vs Containers: The 2026 Calculus
Cold-start improvements and pricing changes have reshuffled the deck since 2022.
Voice AI's Second Wave Is Real
Latency, accuracy, and naturalness have all crossed thresholds that change what's buildable.
Closing Your First Real Enterprise Deal
The procurement and security work that happens after the buyer says yes.
The CDN Is Now an Application Runtime
What changes when your edge platform can execute code, mutate responses, and hold state.
AI-Native Coding vs Low-Code: Two Trends, One Question
Both promise to compress the build cycle. They are not converging on the same answer.
Support as a Moat
Why the fastest-growing SaaS companies treat support as a strategic asset, not a cost center.