What Professional Services Firms Are Building With AI

Most AI coverage in professional services focuses on what might happen. This piece is about what is happening - the specific tools firms are actually putting into production, by practice type.

None of this is aspirational. These are patterns from real engagements. The common thread across all of them appears at the end.

Law firms

Contract risk scoring. A system that ingests a contract, extracts key clauses, and evaluates them against the firm's risk criteria - flagging provisions that fall outside acceptable parameters, missing protections that should be present, and terms that require senior review. The output is a structured risk summary, not a replacement for attorney judgment. The attorney reviews the flags rather than reading every clause from scratch. A version of this appears in the law firm contract risk scoring case study.

Matter intake. A tool that classifies inbound inquiries by matter type, generates an appropriate intake questionnaire for the prospective client, and drafts the initial conflict-check request. The intake questionnaire is tailored to the matter type - a real estate matter gets different questions than an employment matter. The tool does not replace the intake conversation; it structures what gets asked before and during it.

Due diligence preprocessing. A system that takes a document set - often hundreds of files - and organizes them, identifies key provisions across the set, flags missing documents against a checklist, and produces a structured summary that tells the attorney where to focus. The attorney still reviews; the system handles the first pass that used to be junior associate time.

Consulting firms

Market intelligence briefs. A system that monitors a defined set of sources on a schedule, applies the firm's analytical framework to what it finds, and formats the output as a client-ready deliverable. The most sophisticated versions of these encode a senior partner's interpretive process - the sources they trust, the signals they weight, the structure of their analysis. The consulting firm research process case study describes how one of these gets built.

Proposal generation. A tool that pulls from the firm's library of past proposals, identifies the sections and language most relevant to a given RFP, and produces a draft customized to the client and scope. The draft requires editing - it is a starting point, not a finished document. The value is in the time saved at the first-draft stage and in the consistency of what gets included.

Client research synthesis. A system that consolidates information about a client or market from multiple sources - public filings, news, internal past work - into a structured brief before a client meeting or engagement kickoff. This used to take a junior analyst several hours. The tool does it in minutes, in a consistent format.

Accounting and advisory firms

Anomaly flagging. A system that watches transaction data for patterns worth reviewing - unusual amounts, timing anomalies, categories that appear in unexpected combinations. The system flags; the accountant investigates. The value is coverage - a system that never gets tired of checking can monitor more thoroughly than a manual review process.

Client reporting. A tool that pulls from source data - trial balances, time records, transaction logs - and formats it into the firm's report template for a specific client. The accountant reviews and adjusts before anything goes to the client. The time savings comes from the assembly step, which used to be manual and error-prone.

Regulatory monitoring. A system that tracks relevant regulatory changes - new guidance, amended rules, enforcement actions - and assesses their potential impact on specific clients. The output is a summary for each affected client, formatted for the advisor to review before a client call. Keeping up with regulatory change across a client base is one of the most time-intensive parts of advisory work; a monitoring system makes it tractable.

Across all firm types

Two categories appear regardless of practice type.

Onboarding tools. A system that delivers the firm's accumulated knowledge to new hires systematically - SOPs, precedents, style guides, answers to common questions. The alternative is relying on senior people to pass things on in conversation, which is inconsistent and expensive. A well-built onboarding tool gives a new hire access to what the firm knows from day one.

Client-facing tools. A system that gives clients structured access to their data and the firm's analysis - a portal where a client can ask questions about their matter, request a summary of recent activity, or get a structured view of their position. These range from simple to complex depending on the data sources involved and how much autonomy the tool has to answer without advisor review.

What they share structurally: a defined input, a documented process for handling it, and a specified output. The AI does not improvise; it executes. The judgment about what the process should be - what to flag, what to include, how to format the output - comes from the firm. The system makes that judgment repeatable.

This is why identifying the right first project matters as much as it does. The best candidates for a first build have a clearly defined workflow, a senior person whose judgment can be encoded in the design, and an output that is currently produced manually and inconsistently. When those three conditions are present, the build tends to work.