AI infrastructure & unit economics · independent consulting
Nobody owns a cost they can’t see.
I’m Tim Urista. By day I’m a Senior Cloud Engineer at Apple, working on private cloud for internal AI workloads. Independently, I help teams attribute their AI spend, cut the waste, and ship more with less by delegating real engineering work to agents.
Cash-pay · advisory from $250/hr · fixed-price sprints & audits
Receipts, not slides
Line items from my own stack
-
TendForm
commits co-authored by an agent
149 / 176
-
TrendVesting
commits in production since Feb 2024
1,984
-
TigerMill
first 8 image-model calls
≈ $1.00
-
Roku
AWS spend attributed per quarter
$10–12M
Unit that matters
$ per accepted output
Cloud spend attributed
$10–12M/qtr
Cost-attribution pipeline at Roku; co-led the ECS→Kubernetes migration.
Fleet under attribution
~700K servers
Financial attribution where quiet misattribution warps budgets by millions.
Production AI
2.5 years
TrendVesting has run since Feb 2024, on an invoice I pay myself.
Agent co-authored
85%
Of TendForm’s commits: a HIPAA-capable SaaS shipped solo in 11 weeks.
What I do
Spend less on AI. Get more out of it.
Three kinds of work, one discipline: measure the unit that matters, then change the system until the number moves.
01 · Attribute
Find out what your AI actually costs
Per request, per customer, per accepted output. Map spend to owners so someone can act on it, then rank the waste by dollars.
Learn more →02 · Delegate
Hand real engineering work to agents, safely
Gates, attribution, allowlists, and verified-live checks. The same setup I use to ship my own products with a fraction of the hours.
Learn more →03 · Rebuild lean
Replace the expensive thing with a smaller one
Websites and internal tools rebuilt AI-native: less code, lower run cost, and a deploy pipeline you own.
Learn more →New series · Built with Agents
What the commit history actually says
Field notes from shipping real products with AI coding agents: what the commit history says, what it cost, and what I would delegate differently next time.
- Part 1
Sep 13, 2026 · 12 min · TendForm
TendForm: 176 commits, 85% co-authored by an agent
I built a HIPAA-capable form builder alone, on my own time, with Claude on 149 of 176 commits. What the agent did, what broke, and where the money goes.
- Part 2
Sep 13, 2026 · 12 min · TrendVesting
TrendVesting: 2.5 years of production AI, before and after agents
1,984 commits in three eras. Agents sped up the code, but the bigger change was that measuring got cheap, and the numbers overturned what I believed.
- Part 3
Sep 13, 2026 · 12 min · TigerMill
TigerMill: an AI content factory in 33 hours
Two agents built a gated webtoon factory in a weekend. The cheapest lever was not generating, and the metric that matters is the one it doesn't record yet.
Case studies
Costs, at every scale
From quarterly cloud bills in the millions to the dollar a side project spends on image models.
My own practice · 2025–2026
88%
of TrendVesting commits co-authored by an agent, Jul–Sep 2026
How I delegate engineering work to AI agents without losing the plot
Three products shipped mostly by agents, and the operating model that kept quality and cost under control: one lead per repo, gates with attribution, no spend in the toolbelt, and receipts in every commit.
Read case →TendForm · Jun–Sep 2026
85%
of 176 commits co-authored by an agent
A HIPAA-capable SaaS, shipped solo, priced against its own cost stack
A form builder with multi-form packets, a markdown DSL, a 27-tool MCP server, and a HIPAA tier, built in 11 weeks with an agent co-authoring 85% of commits. The cost decisions mattered as much as the code.
Read case →TrendVesting · Feb 2024–present
+54.7 pts
stated confidence above the realized win rate
When cheap measurement overturned 2.5 years of beliefs about an AI product
A production AI signals platform I've run since February 2024. Agents made rigorous measurement cheap enough to do, and the numbers said the confidence scores meant nothing. That's the most valuable finding of the whole project.
Read case →TigerMill · Sep 2026
≈ $1
for the first 8 measured image-model calls
An AI content factory in 33 hours, and the metric that actually matters
A webtoon production pipeline with five gated stages, built by two agents over a weekend. Image generation is cheap per call; what it costs per accepted panel is the number nobody tracks.
Read case →Try it
AI cost teardown
Describe a workload in a sentence. Get a monthly estimate, a unit cost, and the levers, the way I’d sketch it on a first call.
Estimates from a language model. Useful for orientation, not a quote.
AI Cost Teardown
Describe an AI or cloud workload in plain English. You'll get a fully-loaded monthly estimate, a per-unit cost, and the highest-leverage ways to cut it. Live, powered by Claude.
Rates
Published prices. Cash-pay.
Invoiced directly, paid by ACH or card. No platforms or agencies in the middle.
Advisory hours
$250/hr
2-hour minimum
Executive Cost Briefing
$2,500
90 minutes
Unit Economics Sprint
from $5,000
1 day
AI Cost & Unit Economics Audit
$12k–$25k
1–2 weeks
ML Cost & Reliability Evals
$15k–$30k
2 weeks
AI-native rebuild
Fixed quote
After a Sprint