Governed AI operations, run across your portfolio
The "do something with AI" mandate is portfolio-wide — but most lower- and middle-market portcos are operations-heavy and engineering-light. They have the pain, not the IT bench. We operate it for them: one portco pilot at a fixed fee, then a repeatable, governed playbook across the book.
A managed service, not another platform to staff
We operate governed AI operations for a portco — on the portco's own cloud. They don't buy software to learn, hire a platform team, or babysit it. We run it; they get the outcome and own the relationship and the record. This is the opposite of the platform tools your portcos already can't staff.
The pilot: one portco, one named service, $10,000 fixed
Pick one portco and one of three services we already operate — each moves a number and comes out of a budget line the portco already has. Time-boxed at 6–8 weeks, first before/after reading by day 30, no ongoing commitment — small enough that the portco decision does not need a committee.
Every AI platform now hands your portcos the token math — metered agent runs, usage tiers, overage surprises. We price the outcome instead: the fee is fixed, the meter is our problem, the result is the portco’s, and the receipt proves it.
Customer Response Ops
Every inbound customer inquiry answered in minutes, 24/7, bookings captured. Moves: response time and after-hours capture. Budget line: front-office labor / answering-service spend. Best fit: multi-location services, distribution, anything with a front desk.
Content & Demand Ops
The portco's content engine run weekly without hiring: SEO pages, campaigns, localization. Moves: publishing velocity into pipeline. Budget line: agency / freelancer spend.
Outbound & Research Ops
Researched, personalized outreach the portco's team just reviews and sends. Moves: qualified conversations. Budget line: SDR tools / agency line.
Scope + access
Define the one workflow and the success metric WITH the portco. Their IT grants scoped, revocable access.
Build + shadow
We stand it up on their cloud; it runs in shadow / human-in-the-loop so nothing acts unreviewed. The first before/after reading lands by day 30.
Measured production
Live on the agreed metric, human approval on anything sensitive, every action logged.
De-risked by design: runs on the portco’s infrastructure, access scoped and revocable by their IT, a human approves high-impact actions, an explicit "does-not-do" list, and a clean exit if it isn’t working. Production afterward is a per-portco managed-service engagement sized to the labor or agency spend it replaces — that’s the line it comes out of, not a software budget.
The playbook: pilot becomes portfolio
Once one portco works, the pilot is a template, not a one-off. Fan out by ops-shape — distribution, light manufacturing, multi-location services — and the second and third deployments in a shape go far faster than the first.
| Reused across portcos (we bring) | Customized per portco (we configure) |
|---|---|
| The operating architecture + governance model | The specific workflow + systems |
| The audit / approval / reporting posture | Scoped access their IT controls |
| The runbook + integration patterns | The channels in play |
And a portfolio view for you: what’s deployed where, what each portco is governing, and the audit posture across the book — the report you take to your IC and your LPs.
Safe to put in front of LPs and auditors
Four controls, identical across every portco — one governance posture across the book.
Per-tenant, per-action scoping
Each deployment gets exactly the access it needs and nothing more; nothing ambient.
Audit trail with named approvers
Drafted-by / reviewed-by / approved-at / sent-to, for every action — the paper trail an LP or carrier asks for.
Approval queue for high-impact actions
Low-risk steps run on their own; anything sensitive waits for a human.
Multi-channel, one standard
The same governance across email, SMS, web, WhatsApp, Teams, Slack, and more.
Receipts, not claims
An early, founding-portfolio-partner program — which means favorable early terms, and every claim below is checkable.
We operate customer response in production today
A live clinic's inbound channel, daily: real customer inquiries answered, appointments booked against the real schedule, human hand-off built in.
We run our own marketing this way
A multi-thousand-page site in 7 languages, weekly publishing, and the outreach behind this very page — all operated by the same system we would run for a portco.
The governance layer is built, not promised
Tamper-evident audit with external anchoring, approval gates, and autonomy that is earned per action-class from the human accept/edit/reject record.
The thinking, in long form
Three essays from our founder on the portfolio angle.
Integration Is the Moat
Why the value-creation lever is operational integration, not tooling.
Read the essay →The Job Description Is the SOW
Scoping AI operations the way you scope a hire.
Read the essay →Your Vendors Just Named Our Category
The managed-AI-operations category, as seen from the portfolio.
Read the essay →One portco. One workflow. One number by day 30.
To start we need one portco, one workflow, an internal champion, and access scoped by that portco’s IT. That’s the whole ask.