Blog
How we think about shipping AI agents — insights, strategy, and playbooks from the Amodal team.
Everyone's Deploying AI Agents. But Are They Actually Getting Any Work Done?
75% of companies are deploying AI agents, but 95% of generative AI initiatives show no measurable return. The gap isn't ambition, it's whether the agent was actually built to fit the business it serves.
The First AI Feature You Should Ship (And Why It's Probably Not a Chatbot)
Most teams reach for a chatbot when adding AI. Here's why your first feature should solve a smaller, sharper problem, and five patterns that consistently win.
Strategy
What It Actually Takes to Ship an AI Agent
An agent isn't a feature. It's an application stack. Here's the stack, and what each "shortcut" actually covers.
Insight
Why We Open Sourced 90% of Our Agent Runtime
The reasoning behind giving away the runtime and charging for the platform.
Insight
Build vs Buy AI: What Engineering Leaders Get Wrong
The four options for adding AI to your product, and why most teams pick the wrong one.
Insight
The Conversational Layer Is Coming for Every SaaS Product
Railway's AI configured a database in 90 seconds. Then it tried to delete it. What that means for the 30,000 SaaS companies that need to ship the same thing.
Insight
Why Configuration Beats Intelligence
Even infinitely smart models need methodology, access, knowledge, guardrails, and presentation. That's not a model problem — it's a configuration problem.
Insight
Skills Are the New Code
The evolution from prompts to routers to multi-agent to skills mirrors the evolution from assembly to configuration. We're at the Rails moment for AI.
Insight
5 Mistakes That Kill Enterprise AI Agent Projects
Letting the LLM compute, skipping guardrails, monolithic context, dashboard thinking, and waiting for perfect models.
Insight
The Business Case for AI Agents: Numbers Your CFO Will Believe
ROI math, real stats, comparison to alternatives. Everything you need to justify the investment.
Strategy
Human in the Loop Is Not Optional
Confirmation dialogs, knowledge proposal review, admin approval flows. HITL isn't a feature you add. It's the default behavior.
Insight
Why Your AI Agent Needs Guardrails, Not Just Intelligence
Intelligence without guardrails is a liability. Confirmation rules, rate limits, role filtering, audit — these aren't afterthoughts.
Insight
The Embedded AI Playbook for SaaS Companies
Your board wants an AI story. Your customers want an assistant. Your engineering team is already stretched. Ship AI features in weeks, not quarters.
Playbook