An AI layer designed around a fund tax practice's return engine, and a website repaired live.
The situation
4Pines Advisors prepares partnership returns and K-1s for alternative investment vehicles. A certified tax engine computes and files the returns; the hours go into everything around it: document handling, extraction, reconciliation and K-1 review. The firm's public website had also gone live with placeholder team data, duplicate page titles and broken navigation.
What took place
Weeks inside the work
Several weeks on site mapping the recurring employee-level tasks behind return preparation, the firm's IT estate and its office network. The initiative was designed around the work as it is actually performed, not from an org chart.
The AI Initiative Proposal
A nine-page partner-facing proposal for an AI layer that leaves the tax engine as calculator and filer. Around it: document intake by client and year with missing items flagged, line-by-line reconciliation against fund accounting records, first-pass K-1 review with proposed and explained corrections, a deterministic rules engine that runs before any AI judgment, exception reports that list only what needs attention, and a reviewer's sign-off before anything is filed. Everything runs on one dedicated server the firm owns, validated first against prior-season returns so every answer is known before the system produces it.
Website repairs, live the same day
An audit of the firm's site became a three-phase plan: live fixes now, migration to the firm's own server, then a structural build. Phase 1 went in through the live WordPress admin and was verified as an anonymous visitor: SEO titles and descriptions rewritten on 10 pages, unique H1s on 5 valuation pages, 2 menus repaired, the contact form completed, an exposed admin archive closed and duplicate content consolidated behind a redirect.
What comes next
The proposal is under partner review. Approval of its two required lines starts the build the same week: server hardening, the file system, read-only pipelines from the firm's systems, then the experiments against completed returns. The site migration and structural build follow.