Autonomous agents inherit the persona
When an agent runs unattended, it acts as a persona. The audit trail names the persona, not 'the AI'. The persona is the answerable party.
ai
My biggest AI mistakes of the past 12 months
An honest retrospective. The mistakes I made with AI in 2025-26, over-trusting model output, under-investing in evals, picking the wrong abstractions, skipping verification. The lessons I actually learned, written down so I don't have to learn them again.
AI in the news: week of April 26, 2026
Google Cloud Next launches the Gemini Enterprise Agent Platform. GPT-5.5 ships. Google commits $40B to Anthropic on Friday. DeepSeek drops a 1.6T MIT-licensed V4 the same day. April Challenger lands ugly. The frontier-lab market is now hyperscaler-vs-hyperscaler.
Embeddings need personas too
A semantic search is too helpful by default. It will find the personal note that shares a name with the work doc. Scope the index.
Personas, all the way down to the database
Persona isolation that stops at the application layer isn't isolation. The DB has to know the persona too, row-level filters, audit logs, the whole thing.
Tenant-scoped policies without tenant-scoped code
Policies vary per tenant. Code does not. The architectural payoff of tenant-aware policy bundles. OPA-style, that don't require a tenant-aware codebase. Bundle resolution, defaults, overrides, and the audit story that makes 'what happened for tenant X' answerable without grep.
Memory isolation is the whole point
Long-term memory has to be persona-scoped from day one. Otherwise the AI brings up the wrong thing at the worst possible moment, and you stop trusting it.
AI in the news: week of April 19, 2026
Anthropic ships Claude Opus 4.7 with a tokenizer change that quietly raises the bill ~15-25% at the same headline price. OpenAI fires back with GPT-5.4-Cyber. MiniMax open-sources a self-evolving agent model that ends the hosted-frontier argument. Q1 layoff data lands ugly.
MCP, but persona-aware
The tool-routing layer between an AI and the rest of your life has to know which persona is asking. Otherwise the wrong tool always shows up.
Personas at the model layer: prompt, tools, memory, all at once
When the persona switches, four things move at the same time: the prompt, the tool surface, the memory, and the identity doing the signing. All four. Atomically.
Cosmology for multi-tenant: universe, system, environment
The standard tenant / sub-tenant framing for multi-tenant is too thin. The cosmological metaphor, universe, galaxy, solar system, environment, gives you a layered model that maps onto the scopes that actually matter: compliance, billing, isolation, hierarchy.
Projects: the second axis of persona scope
Personas are the top-level containers. Projects are the next layer in. Here's why projects belong nested inside personas, not floating around as flat tags.