Most AI products on offer today are subscriptions sitting on top of a model someone else owns.
We build the version your firm actually wanted, scoped to how you already work.
And we hand it over. The code lives with the firm. No seat fee. No vendor lock. No quiet quarterly price hike.
Maule & Co. designs and builds AI tools for tax, accounting, finance, and legal practices. Each engagement starts with the way the firm already works. We sit with the partners and the associates, map what they do by hand, and write software that does only the parts that should not still be done by hand.
What we ship is yours. The repository, the model access, the prompts, the evaluations. Nothing is held hostage on our infrastructure. When we leave, a senior person at the firm knows how it works and can change it without calling us.
Every engagement runs against a database the firm owns and can query. It holds the cases, the prior engagements, the filings, the memos, the intake notes, and the institutional knowledge that walks out the door whenever a senior person retires. The data lives in one place, structured, in the firm's environment, searchable in plain language.
When a partner asks whether the firm has seen a fact pattern like this before, the answer is the answer the senior associate would give after thirty years at the desk. When a new hire drafts something the firm has drafted a hundred times, the prior drafts are in front of them, not in a folder nobody can find. The practice-area work that follows runs against this layer.
Large firms have been able to build this kind of internal knowledge layer for decades. Small firms have not. We build it for one firm at a time, in the course of a single engagement, and the firm keeps it.
Tax, accounting, finance, and legal. Equal weight, named in plain language, with examples of what we'd actually build.
Return prep, memos, research, and correspondence in the firm's voice.
Return prep that pulls from intake documents without re-keying. Position memos drafted against the IRC and the firm's prior memos. Research that returns the regulation, the case, and the cite, with the firm's prior view attached. Client correspondence written in the partner's voice, not a model's.
A close that runs the same way every month.
Month-end close that runs the same way every month and flags only the items that look unlike last month. Reconciliation review at the partner level. Audit support packets assembled from the working papers the firm already keeps. Bookkeeping QC that catches what a careful human would catch.
Memo, model, and monitor against the firm's own precedent.
Investment memos drafted from the model and the data room. Comp sets assembled and kept current as a deal progresses. Documentation review against the firm's own precedent. Post-close monitoring that pulls the actual numbers and reports them against the original case.
Discovery, abstraction, and drafting tied to the firm's prior work.
Discovery review with citations back to the document and the page. Contract abstraction into the firm's own schema, not a vendor's. Deposition summaries written the way the senior associate would write them. Research that returns the case, the holding, and how the firm has read it before.
A sample of work to date. Identifying details generalized.
A boutique advisory firm needed to run pre-acquisition tax and financial diligence on small private targets without staffing up. We built a workflow that takes the target's books and tax filings, walks them for related-party items and recurring exceptions, and produces a diligence memo in the firm's house format with citations back to source.
A small CPA practice wanted to move individual returns from client intake to a complete 1040 draft in a single pass. We built file-level intake that reads W-2s, 1099s, K-1s, brokerage statements, and the prior-year return, posts to the firm's tax software, and surfaces only the items that need a preparer's judgment.
An employment law boutique wanted a single intake and matter system. We built one that ingests client emails, court filings, and prior firm work, parses the underlying facts and dates into a structured record, and pre-populates the firm's standard letters, complaints, and discovery requests for the partner to mark up.
A NYC nonprofit family law practice handling Article 10 cases needed faster file review under deadline. The tool reads the petition, ACS records, and prior court filings on a case, summarizes the procedural posture, and returns family court rulings on similar fact patterns from the last several years, with cites back to the orders.
A small wealth advisory wanted partner-level monitoring of every client portfolio against each client's mandate. We built a dashboard that pulls live custodian data, runs the firm's allocation rules against actual holdings, and surfaces drift, concentration, and tax-lot opportunities each morning before the partners' first call.
Our clients are small. The kind of firm where one lunch and one follow-up call are enough to commit. Larger firms can buy AI software and never use it. Smaller firms have to put it to work, because nobody else is going to put it to work for them. That is the work we want.
Week one is on site at the firm. We watch the work as it actually happens, not as the org chart says it should.
Week two we describe that workflow back to the firm in writing, and the firm tells us where we got it wrong.
From there we build, install in the firm's environment, and stay close enough to fix what does not survive contact with real files.
When the engagement ends, the firm owns the code. Routine changes are made by the firm's own people, who already know the system and the work. When the underlying technology has advanced enough that the system should be rebuilt rather than patched, we come back for that engagement.