AI infrastructure consulting

AI in production.

Not in a deck.

ProjectOps designs, builds, and operates AI infrastructure for businesses that need to ship. Custom agents. MCP integrations. Deployment, observability, and the ops layer behind it — in your repo, on your stack, by month’s end.

Book an audit See selected work
The Audit Brief

A scoped plan,
before any contract.

Every engagement starts with a 3-day audit. Out comes a written brief: the workflows worth automating, the cost, the timeline. You decide if we keep going.

Book the audit See what we ship
Audit BriefPO·0509 Scoped
AC

Acme Co · sales + ops

27 employees · 4 candidate workflows · 2 pilot-ready

Workflows4
Pilot ready2
Est. saved$14k/mo
Day 3→cto@acmeco.com
May 9 · written brief delivered
Decision by Friday
Step one · the audit

We map your AI surface area.

Three days, on-site or async. We sit with your team, read the code, watch the workflows. Out comes a written list of what an agent could own — and what it shouldn’t.

Audit deliverables
3 days
Current AI stack Mapped
Workflow opportunities Listed
MCP & integration gaps Found
Build vs buy guidance Per workflow
Risk & compliance Reviewed
Step two · the pilot

We ship one thing in 30 days.

A fixed-fee pilot. One agent or workflow, taken to production. You see the code, the costs, the decisions, the deploy URL — not a deck.

Pilot · sales-triage agent
30-day
Audit completeday 3 · brief delivered
Pilot scopedday 5 · SOW signed
Build startedday 8 · repo live in your GitHub
Integration + QAday 22 · staging deployed
Deploy + handoffday 30 · production live
Step three · the bill

Fixed fees. No surprises.

Every phase is fixed-fee, written up front. Audit, pilot, optional retainer. No T&M. No scope creep. No "just one more sprint."

EngagementQ1 plan · acme-co
PO-0509
Audit · 3 days, fixed$4,500
Pilot build · 4 weeks$18,000
Operate · monthly retainer$6,000
Q1 total$28,500
What an in-house AI team would have been $90k/qtr
The terms

Boring contracts. On purpose.

Each engagement runs scoped to one customer. Outbound is gated. Secrets stay in the vault. The deliverables are yours. Boring posture, on purpose.

Engagement posture
Standard
Per-engagement isolationEnforced
Outbound approvalGated
Secret storageVaulted
Cross-customer shareNone
Handoff doc + runbookOn exit
The engagement

Three phases. No retainers’ you can’t end.

Audit, pilot, operate. Each is fixed-fee. Each ships something real. Each one ends if you want it to.

01

Audit

Map your stack.

Three days. On-site or async. We sit with your team, read the code, watch the workflows. Out comes a written brief with two or three candidate pilots, each scoped and costed.

02

Pilot

Ship one thing.

Thirty days, fixed fee. We build one agent or workflow in production. Repo lives in your GitHub. Decisions documented. Deploy URL by day 30. Your team can read every line.

03

Operate

Run it. Or hand it off.

Optional monthly retainer. We run the system, iterate, and hand off when you have an internal owner. No vendor lock-in. Thirty-day exit. Your team owns the code regardless.

Engagements

Three ways to start.

Fixed fees up front. Larger or specialized engagements priced separately.

Audit
$5,000

Just the assessment. We map your stack, your workflows, and your AI surface area — then write up what’s worth automating, what isn’t, and why.

  • Stack & workflow assessment
  • Candidate workflows, scoped & costed
  • Build vs buy guidance per workflow
  • Risk & compliance review
  • Written brief, no obligation
Book the audit
Specific Need
$10,000

You already know what you want built. Skip the audit, start with the work. One targeted deliverable, taken to production.

  • One agent or workflow, scoped
  • Direct to build — no audit phase
  • Decisions log + runbook
  • Production deploy
  • Handoff to your team
Start the build
Most common Audit + Project
$15,000

The audit, then the project. Most engagements start here. Extends to ongoing if the work warrants it.

  • Everything in the Audit
  • One project taken to production
  • Decisions log + runbook
  • Optional ongoing retainer
  • Ongoing: $10,000 / month
Start an engagement

Larger or specialized engagements priced separately. Other costs — cloud infrastructure, model usage, third-party services — pass through at cost.

The difference

Decks. Or systems.

Same problem space. A meaningfully different way to actually get AI into your business.

The traditional path
  • Hire an AI engineer in SF: 3 months to first hire, $250k+ comp.
  • Big-firm consulting: 6-month engagements, slide decks, no production code.
  • DIY with ChatGPT: stuck at proof-of-concept, no infra, no observability.
  • One-size SaaS: doesn’t model your business, can’t leave.
  • Wait until Q3 when you have time. You won’t.
The ProjectOps path·
  • Three-day written audit, fixed fee, scoped in writing.
  • Thirty-day pilot, in production by day 30, on your stack.
  • Code lives in your repo from commit one. No vendor lock-in.
  • Fixed fees, written terms, 30-day exit on every phase.
  • You decide what to keep. We hand off when you’re ready.
Selected work

Built. Shipped. In production.

Two engagements where the deliverable was a working system — not a recommendation, not a roadmap, not a deck.

Finance · Wealth Engagement · 2026

Relationship intelligence for a financial-advisory practice.

Designed and shipped an AI capture-and-recall layer for a national wealth-management workflow. Every client meeting transcribed and indexed. Every promise tracked. Follow-up emails drafted before they slip. The advisors stopped opening their CRM — the system tells them what to do next.

“We came in expecting a deck. We left with software that’s been in production for six months.”

Founding partner · financial-advisory practice
Legal · Litigation Engagement · 2026

Sovereign legal AI, designed in the courtroom.

Built a deposition co-pilot and firm-locked LLM with a working litigation team. Timestamped audio audit layer. Citations grounded in the firm’s own corpus, not the open web. Every iteration tested against real depositions, real motions, real workflow — not whiteboard sessions.

“Built inside the firm. Tested against real depositions. Doesn’t hallucinate, doesn’t grandstand. It feels less like software and more like a paralegal who never forgets.”

Practicing litigator · 18 years
Common questions

The boring stuff,
answered up front.

Procurement, legal, and security questions, answered before they hold up the engagement.

Who owns what we build?
You do. Everything we deliver is yours — the code, the decisions log, the infrastructure config, the runbook. There’s no separate license, no hosted SaaS layer, no “ProjectOps platform” you’re renting access to.
What if we want to bring it in-house?
Your team takes over. We deliver a handoff doc, the runbook, and the decisions log — your engineers own it from there. No exit fee, no clawback, no licensing wrinkle when we step out.
How do you handle security and data isolation?
Each engagement is scoped to one customer. No data crosses portfolios. Secrets stay in your vault — we use deploy hooks and env-injection, never plaintext keys in code. Outbound calls (email, posting, billing) are gated through approval queues you control. We sign whatever DPA, MSA, or NDA your legal team requires.
Can we start with just the audit?
Yes — it’s the recommended path. The audit is fixed-fee, no strings. If the brief doesn’t make a compelling case for a project, you walk away with a written assessment of your AI surface area. About a third of audits stop there; the rest proceed to a project.
What size company is this for?
Most engagements run with companies between $5M and $200M in revenue — a working ops or engineering team but no dedicated AI hire. Below that, the per-engagement fees usually outpace ROI; above that, you typically have an internal AI team that we’d augment rather than replace.
What models and stack do you build on?
Whatever fits the workload and your existing infrastructure. We’re not tied to a single provider. Most pilots route across Anthropic and OpenAI; orchestration sits on Vercel or Railway depending on the deploy story; data lives in whatever your stack already uses. The audit decides the stack — not us, in advance.

An audit.
And a system, by month’s end.

Tell us about your business and the workflow you’d most like an agent to own. We’ll send back a scoped audit brief inside one business day.

Book an audit

3-day audit · Fixed fee · will@projectopsai.com