Writing
Notes from the desks we automate. What worked, what didn't, and what we'd do again.
Practical notes for estimators, order desks, finance teams and engineering leads, written by the founders and engineers who build the agents.
Latest
Start with the workflow, not the model
Your first agent project succeeds or fails on the workflow you pick. How to choose one by volume, clear rules, systems access, a measurable baseline and contained risk.
4 min read
Earlier posts
5 min read
Merged pull requests is a vanity metric
Coding agents push merged-PR counts up whether or not anything improved. The six numbers to report instead, how to baseline them in week one, and the one result that counts.
5 min read
We'd rather it said “no match”
How to design a BOQ-to-catalogue matcher an estimator will sign. Attributes with criticality, rules before the model, a written rubric, abstention, alternates, a review queue and a trail.
4 min read
What makes an AI agent production-ready
A demo proves the model can do the task. Production needs scoped tools, guardrails in code, approvals, an audit log, evals from real cases, monitoring and a cost ceiling.
4 min read
Evidence per pull request
The cover note that makes an agent's PR reviewable. What changed, tests run fresh at this commit, what it didn't touch, review result, risks, what's not done, and what humans still read.
5 min read
The verify tax
Review time decides whether AI estimating and quoting pays. How to design so every line links to its source and only exceptions get reviewed, and the three numbers to measure.
4 min read
When agents shouldn't open the pull request
Some changes belong with a senior owner. Auth, payments, crypto, irreversible migrations, infra with blast radius and untested code, and how to encode that in repo instructions and branch protection.
5 min read
Customer part number to our SKU
The cross-reference problem behind every RFQ and PO. Lookup tables, attribute matching, revisions, UoM, MOQ, contract price, the exception card, and writing to the ERP without duplicates.
4 min read
It ran green for three weeks and did nothing
The production checklist every agent workflow should ship with, drawn from what buyers find after freelance and agency handovers. Outcome checks, idempotency, cost caps, owned credentials, a runbook.
6 min read
Why invoices actually stop
The real reasons three-way-match invoices land on hold, which ones an agent can settle with a proposal, which must stay with a person, and the GST e-invoice checks Indian AP teams add.
5 min read
Token budgets are a delivery problem
Why AI coding spend spikes, why per-seat caps hit the wrong engineers, how per-task model routing and cost per merged PR work, and budget policies that don't punish the people getting value.
4 min read
What changed in addendum 3?
Diffing tender revisions, corrigenda and addenda before bid day. Classify each changed line, re-match only what moved, and keep the deviation register in step with the BOQ.
Send us ten real documents. We'll show you what an agent does with them.
A one-hour call with a founder and an engineer. Bring a BOQ, a stack of POs, last month's invoices or a backlog, and leave knowing what would work, what wouldn't, and what it would take.
170+ projects · 65 companies · founders on every engagement