PE / VC sponsors and advisors

The engineer your due-diligence shop calls when the AI work won't transfer.

When due diligence surfaces brittle AI, vendor black boxes, or per-seat automations that won't survive a transfer, I rebuild them as owned software inside the portco's own Azure or GCP tenant. The next owner reads the code, runs the math, and ships the next feature.

Tell me what you're working on

The thesis

Where I fit alongside diagnostic and operating partners.

Your diagnostic frames the exposure and surfaces the per-seat sprawl. I start after that: the portco needs the rebuild and you need an engineer who stands behind the code.

The average company runs 106 SaaS apps; enterprises run 291. Roughly half of those licenses sit unused, about $18M per year in waste at enterprise scale. The January 2026 SaaS repricing wiped $300B off public market caps in weeks.

I rebuild the pieces that should ship as owned software, inside the portco's own cloud. The line item moves from a vendor renewal to an asset on the balance sheet, and the next operator inherits the source. What transfers is the harness: the setup around the AI that connects it to the portco's systems, enforces its rules, catches mistakes, and remembers what works.

The proof

Work the new team can actually carry forward.

Not a deck. Production systems running in client clouds today.

Founding case · live

$150K/yr in annualized per-seat ERP license savings (additional ERP + BI seats stack on top)

Northeast Florida commercial contractor · construction

Embedded as the solo software engineer. Building the custom apps that replace per-user ERP seat licenses: time entry, job costing, approvals, reporting. Every new project manager the company hires is a login they already own, not another per-seat license. The apps run inside the client's Azure tenant. Their code, their data, their repo. If I walked tomorrow, the team reads the source and ships the next feature. That is the test.

Legal

TextTimeline

Paralegal hours of chronology assembly down to minutes. 100% source-cited output, FAISS + BM25 hybrid retrieval, Cloud Run.

Read the case study →

Fintech

GammaRips

Autonomous overnight options-flow scanner. 14 Cloud Run services, ~20 schedulers, multi-agent ADK publishing layer with deterministic compliance gating.

Read the case study →

Construction

Northeast Florida Commercial Contractor (anonymized)

Custom Azure web apps replacing per-seat ERP workflows, built inside the client's own tenant. The ERP stays the system of record; the next operator inherits the source and runs the math against last quarter's invoices.

Read the case study →

The model

How I plug in next to your work.

Same engineer, four ways to plug in. Best fit is portcos roughly $5M to $100M in revenue. Scope set in conversation; no portfolio-wide SOW to fight through procurement.

Drop in when your diagnostic surfaces engineering work. Bounded scope, single workflow, your client relationship stays primary.
Subcontract under your engagement on AI-specific build that your other advisors aren't staffed for.
Embed in a target portco directly for a multi-month rebuild. The founding pattern: senior engineer inside their tenant, shipping production apps that replace per-seat SaaS.
Stay on retainer post-build for senior architectural eyes on portco-managed systems.

Every engagement runs in the portfolio company's own Azure or GCP tenant, under their credentials; I do not move data between engagements. Paper it with the portco, the sponsor, or under a sponsoring advisor's engagement, whichever keeps the right relationship primary. Monthly written updates per engagement, rolled up on request. Capacity is deliberately small; that is the point.

Tell me which portco.

Thirty-minute conversation with you, your operating team, or a single portco CFO. I'll tell you straight whether the founding pattern ports to their stack, how I'd approach the rebuild, and what the next operator inherits.

Tell me what you're working on