blog8 posts3 categories8 active topicsfiltered: engineering

Notes from the field.

Working notes on the craft — engineering at scale, the AI-era developer experience, frontend that most teams leave on the table, and whatever else has my attention this week.

Building tablesalt: a CSV agent where the answer IS the UI
tablesalt is an in-browser data agent. You drop a CSV, ask a question, and the agent renders the answer as the right kind of UI: a chart, a stat card, a table, or a list. No chat bubbles.
ai
generative-ui
vercel
evals
open-source
May 14, 20263 min
Building grant-pilot: a multi-turn agent that orchestrates three sub-agents over real federal data
A multi-turn agent that finds federal grants for small businesses and nonprofits. A planner orchestrates three sub-agents (discovery, eligibility, drafter) over real federal data, with a hard daily budget cap, atomic spend reservation, and a planner that never throws. Here's what I built and what shipping it taught me.
ai
agents
sub-agents
orchestration
fde
open-source
May 10, 202615 min
Building fedbench: an LLM eval harness for grounded Q&A
Most LLM agents that read documents and answer questions fail in three specific ways. fedbench measures all three. Here's what it does, why I built it, and what I learned shipping it.
ai
evals
rag
open-source
May 9, 202614 min
Building fieldops-mcp: designing the tool surface an agent actually uses
I built a small MCP server that turns a fictional dispatcher's workflow into agent tools. The interesting part wasn't the code. It was the design choices that decide what the agent can do at all. Here's what I built and what shipping it taught me.
ai
agents
mcp
tool-design
open-source
May 9, 20267 min

Building tablesalt: a CSV agent where the answer IS the UI

Building grant-pilot: a multi-turn agent that orchestrates three sub-agents over real federal data

Building fedbench: an LLM eval harness for grounded Q&A

Building fieldops-mcp: designing the tool surface an agent actually uses