Skip to main content
How Can We Help?Contact

Professional Services

Make your firm’s knowledge work harder, and every engagement earn what it should.

We join what your time and billing, documents, CRM and people systems each know. Partners see utilization, realization and margin while an engagement can still change; precedents and past work are found instead of remembered; and AI is built inside the confidentiality, consent and review your profession requires, and inside each client’s terms.

Talk through your firm’s data See the contract-review case

Law firms · Accounting and tax firms · Consulting firms · Architecture and engineering firms

One engagement at the centre, its ring filling stage by stage: a pursuit won from the pipeline, a team of four staffed, the margin tracked against plan, and the client’s terms approved for AI use1234
  1. 1WinPursuit won, priced from the closest past wins
  2. 2DeliverTeam of four staffed on skills and availability
  3. 3MeasureMargin at 31% against a 35% plan, with three weeks to act
  4. 4ProtectAI use cleared for this client under its engagement terms
Illustrative: one engagement, won, delivered, measured and protected.

Where firms lose time, margin and institutional knowledge

Four patterns we see in law, accounting, consulting, and architecture and engineering firms alike. None is a shortage of talent or data, and none is solved by another license.

1.1Felt by associates, managers and bid teams

Precedents that live in people’s inboxes

Every first draft, proposal and tax position starts with an email asking who has done this before, and the answer depends on who replies.

See § 4: what changes, firm by firm

1.2Felt by practice leaders and finance

Margins seen only after the work is done

Time, rates, write-downs and staffing sit in different systems, so an engagement’s margin shows up once it closes.

See § 3: seven firm metrics, defined

1.3Felt by general counsel, risk and IT

AI used one person at a time

People try AI tools on their own, with no approved list, no position on client consent and no record of which client’s information went where.

See § 6: checking an AI use case

1.4Felt by managing partners and business development

One client, known five different ways

Each practice and office keeps its own view of the client, so the whole relationship is seen by nobody.

See § 4: one view of each client

A precedent: contract review that starts with the exceptions

One engagement in legal and procurement work, set out as a brief. The figure is the client’s own; the client is described, not named.

The client

The legal and procurement teams of a global financial information and ratings provider

Contract compare-and-comply across MSAs, SOWs, POs and invoices

The facts

Master agreements, statements of work, purchase orders and invoices sat in many versions across the business, and every document needed a reviewer to find the terms that departed from standard.

What we built

One version-aware library of the agreements; terms extracted and checked against the approved clause library; purchase orders and invoices matched to the terms that govern them; and review reports that lead with the exceptions, with attorney review kept in place.

The result

15–25%of documents are nonstandard: now the only ones that need a reviewer

As reported by the client.

Read the full case study

Seven firm metrics, and the definitions worth settling

When finance, practice leaders and partners report different numbers, the dashboard is rarely the cause. The definition is, and each one moves a decision. Pick a metric to read it the way an agreement would define it. Firms differ: profit per partner fits firms run as partnerships, and net multiplier fits architecture and engineering.

Utilization

“Utilization” means, depending on who reports it:

  1. Against 2,080 hours or available hours: billable hours divided by a standard 2,080-hour year, or by the hours each person was actually available after holidays and time off;
  2. Billable or chargeable: only the hours billed to clients, or every client hour, including written-off and fixed-fee time; or
  3. Who is in the base: partners, part-time staff and people on leave counted or left out.

What it decidesHiring, staffing and how much capacity the firm really has.

Illustrative example. A consultant records 1,600 billable hours in a year with 1,800 hours available after holidays and time off: utilization is 77% against 2,080 hours and 89% against available hours.

What the firm has to agreeAgainst which hours, counting which time, and for whom?

Sources: time entry, the people system, resourcing, finance.

Realization

“Realization” means, depending on who reports it:

  1. Billing realization: what was billed, against the value of the time worked;
  2. Collection realization: what was collected, against what was billed; or
  3. Standard or agreed rates: time valued at the firm’s standard rates, or at the rates each client agreed, which hides negotiated discounts.

What it decidesPricing, collections, and how partners are judged.

Illustrative example. Time worth 100 at standard rates is billed at 88, and 83 is collected: billing realization is 88%, collection realization 94%, and overall realization 83%.

What the firm has to agreeBilled or collected, and against standard or agreed rates?

Sources: time and billing, receivables, the general ledger.

Lockup

“Lockup” means, depending on who reports it:

  1. Unbilled time and receivables: days of unbilled time plus days of unpaid invoices, or only one of the two;
  2. Gross or net: before or after the amounts already expected to be written off; or
  3. Which daily revenue: divided by the last three months’ daily average, or the year’s.

What it decidesCash, borrowing, and when to chase a client for payment.

Illustrative example. Unbilled time worth 45 days of revenue and unpaid invoices worth 60 days: lockup is 105 days, or 60 if only receivables are counted.

What the firm has to agreeUnbilled time and receivables together, gross or net, against which period’s revenue?

Sources: time and billing, receivables, finance.

Engagement margin

“Engagement margin” means, depending on who reports it:

  1. Direct or fully loaded: after the cost of the people on it, or after a share of overhead as well;
  2. Partner time: partner hours costed at a salary equivalent, or left out because partners share the profit; or
  3. Whose write-offs: charged to the engagement that caused them, or absorbed by the practice.

What it decidesWhether a fixed fee is repriced, restaffed or declined next time.

Illustrative example. Fees of 100; the team’s time costs 55 and its share of overhead is 25: the margin is 45% on direct cost and 20% fully loaded.

What the firm has to agreeWhich costs, which partner time, and whose write-offs?

Sources: time and billing, payroll, the general ledger, practice management.

Profit per partner

“Profit per partner” means, depending on who reports it:

  1. Equity or all partners: divided by equity partners only, or by every partner;
  2. Before or after partner pay: profit before any partner is paid, or after salaried partners are paid; or
  3. Heads or full-time equivalents: partners counted as heads, or by the time each one works.

What it decidesPartner pay, admission to the partnership, and the firm’s standing against peers.

Illustrative example. Profit of 60 million in a firm with 20 equity and 10 salaried partners: 3.0 million per equity partner, or 2.0 million across all 30.

What the firm has to agreeWhich partners, and profit before or after whose pay?

Sources: the general ledger, partner accounts, the people system.

Win rate

“Win rate” means, depending on who reports it:

  1. By count or by value: proposals won as a share of proposals sent, or the value won as a share of the value proposed;
  2. Which pursuits count: formal proposals only, or every qualified opportunity, renewals included; or
  3. When a pursuit is lost: at the client’s decision, or after a set time without one.

What it decidesWhich pursuits get partner time.

Illustrative example. Forty proposals worth 200 in total; ten won, worth 70: a win rate of 25% by count and 35% by value.

What the firm has to agreeBy count or value, counting which pursuits, and lost when?

Sources: the CRM, proposal records, finance.

Net multiplier

“Net multiplier” means, depending on who reports it:

  1. Net or gross revenue: fees after consultants and reimbursable expenses are passed through, or before;
  2. Direct labor at what cost: salaries only, or with benefits added; or
  3. Project or firm: per project, or for the firm as a whole, which hides the projects that lose money.

What it decidesFees on the next proposal, and which projects the firm takes on.

Illustrative example. Net revenue of 3.0 million on direct labor of 1.0 million is a net multiplier of 3.0; count 0.6 million of consultant fees passed through and it reads 3.6.

What the firm has to agreeNet of pass-throughs, against which labor cost, by project or firm-wide?

Sources: project accounting, payroll, the general ledger.

Which of these do your partners dispute at every review? Bring it to a 30-minute call, and we will tell you what it would take to define it once and report it the same way everywhere. If several are in dispute at once, the engagement profitability diagnostic is the usual first step.

Talk through one firm metric

What changes, firm by firm

Each piece of work is written the way a firm marks up a draft: how it works today, struck through, and how it works after.

4.1

Law firms

  • Precedents found, not remembered

    A first draft starts from was: whatever the last associate remembers now: the firm’s closest precedents, each one cited

    Matters, agreements and memos the firm already has, searchable by issue and clause, with drafts that show which precedent each passage came from and a lawyer reviewing every change.

    Case study: Contract compare-and-comply

  • Bills that hold up, and realization you can see

    Write-downs are found was: after the client cuts the invoice now: before it goes out, against each client’s billing guidelines

    Time entries drafted from the day’s documents and calendar for timekeepers to confirm, every invoice checked against the client’s outside counsel guidelines, and realization and write-offs shown by partner and client, so fewer bills are cut or sent back.

    First step: AI & Agent Pilot

4.2

Accounting and tax firms

  • Busy season seen weeks ahead

    Review bottlenecks are found was: in the last week before the deadline now: while there is still time to move work and people

    Returns in progress, client documents outstanding, reviewer capacity and likely extensions joined by office and partner, so work and reviewers are moved before the deadline, and quality is not traded for speed.

    First step: Data Platform Accelerator

  • Analyze the full ledger, and investigate what stands out

    The audit team examines was: a sample, and extrapolates now: the whole ledger it was given, and investigates what stands out

    Analysis of the full population of entries the client provides, with the reliability of the data tested first, every flagged item investigated and documented, and the audit procedures and conclusions left to the engagement team.

    First step: AI & Agent Pilot

4.3

Consulting firms

  • Proposals that start from past wins

    A proposal starts from was: a blank page and a reply-all now: the closest past wins and the people who delivered them

    Past proposals, statements of work and deliverables cleared for reuse, searchable by sector and problem, with first drafts that show what they borrow and a partner approving every one.

    First step: AI & Agent Pilot

  • The right people, and the margin, before the team is confirmed

    Staffing is decided was: by whoever answers first now: on skills, availability, the pipeline and the margin it leaves

    Skills, availability, past engagements and the pursuits likely to close joined, so resourcing sees who fits, the engagement’s margin is forecast before the team is confirmed, and utilization is planned rather than discovered at month end.

    First step: Data Platform Accelerator

4.4

Architecture and engineering firms

  • Project margin while it can still change

    Fee overruns are found was: at the final invoice now: while the project is half done

    Hours, fees, change orders and percent complete joined by project and phase, with remaining labor estimated and a forecast to complete, so project managers see fee burn early enough to act and can trust the percent complete they report.

    First step: Data Platform Accelerator

  • Standards answered from the firm’s own library

    Engineers search was: shared drives and old projects now: the firm’s standards, details and past specifications, cited

    An assistant over the firm’s own standards and specifications; the engineer of record checks every answer and stays in responsible charge.

    First step: AI & Agent Pilot

4.5Felt by partners, practice leaders and HR

Adoption: where partners set the pace

New tools become the way the firm works was: after a launch email and a training session now: when partners expect them in review, and the hours they save count for people, not against them

In a professional firm, a tool is used when the people who judge the work expect it. That depends on partner expectations, incentives, established workflows, training and professional judgment far more than on the tool itself, so the work covers all five.

  • What partners expect

    Partners ask for the tool’s output in reviews and pitches, so using it is how work gets accepted, not an extra step.

  • What is counted

    Hours saved are not held against people measured on billable hours; the time goes to work clients will pay for.

  • How work is reviewed

    The tool sits inside the existing review step, and the judgment stays with the professional who signs.

  • What people learn

    Training by role and practice, on the firm’s own kind of work, rather than a general course.

First step: Adoption Review. Related practice: AI & Data Literacy.

4.6

Across every kind of firm

  • One view of each client

    Relationships, engagements, billings and pursuits joined across practices and offices, so partners see the whole client and the cross-practice work it could buy next.

    First step: Data Platform Accelerator

  • An AI policy with a register behind it

    Approved tools, the position on client consent and every AI use on one register, each with an owner, so a client’s question about AI gets an answer the same day.

    First step: AI Governance Quickstart

Your firm does something else, or several of these at once? Bring the one that costs the most partner time to a 30-minute call.

Talk through your firm’s priority

The professional obligations that come with the data

Three tests apply to any AI use, and passing one does not pass the others. The obligations below are the first test, grouped by whom they bind; they are obligations to evaluate, not a complete checklist. Each one links to its source.

Three tests

  1. What your professional rules permit

    The bar, the IRS, the PCAOB, the engineering board: the rules that bind the person who signs the work.

  2. What the client’s agreement permits

    Each client’s engagement letter and guidelines, client by client, which can say more than the rules do.

  3. What your firm’s own policy allows

    Your AI and data policy, tool by tool: which tools are approved, for what, and on whose information.

Lawyers

  • ABA Formal Opinion 512 (opens in a new tab)Issued Jul 29, 2024

    Lawyers using generative AI keep their duties of competence, confidentiality, supervision and candor. Whether a client’s informed consent is needed depends on the tool, the information and the risk of disclosure, and a firm-approved tool does not settle that on its own. Output is checked before it reaches a client or a court, and hourly bills reflect time actually spent. State bars and some courts add their own rules.

Tax practitioners

  • IRS Office of Professional Responsibility Alert 2026-19 (opens in a new tab)Issued Jun 24, 2026

    Under Circular 230: verify every fact, citation and calculation an AI tool produces, use only secure, firm-approved tools for client data, keep written firm procedures for training, data handling and vetting tools, and let time savings show in fees.

  • Internal Revenue Code section 7216 (opens in a new tab)In force

    Tax return information is used or disclosed outside the regulation’s exceptions only with the taxpayer’s written consent. Whether sending it to an AI tool counts as a disclosure is not settled; the answer depends on the activity, the vendor arrangement and the exceptions. A specific signed consent, with a way to serve clients who decline, is the more cautious way to manage that risk, not a requirement in every case.

Firms that hold clients’ financial information

  • FTC Safeguards Rule (opens in a new tab)Breach notice since May 13, 2024

    Tax preparation firms and other covered financial businesses keep a written information security program, and report to the FTC, within 30 days, a breach of unencrypted information affecting at least 500 consumers.

Auditors of public companies

  • PCAOB amendments on technology-assisted analysis (opens in a new tab)Fiscal years from Dec 15, 2025

    For PCAOB-registered firms auditing public companies, not private-company audits or tax work: when software analyzes a client’s data, the auditor evaluates whether that data is reliable and investigates the items the analysis flags. Upcoming: the new quality control standard, QC 1000, takes effect December 15, 2026.

Engineers

  • State engineering boards: North Carolina’s guidelines on AI (opens in a new tab)Adopted Jul 16, 2025

    Licensure, sealing and responsible charge are set state by state; North Carolina’s guidelines are one example. There, the licensee keeps responsible charge of AI-assisted work and full responsibility for it: controls the inputs, validates and can reproduce the outputs, keeps records of both, and seals only work in their responsible charge. The national council of engineering boards set the same expectation in a 2025 position statement, and other state boards are issuing their own.

Every firm that hires

  • New York City Local Law 144 (opens in a new tab)Enforced since Jul 5, 2023

    An automated tool that substantially assists or replaces discretionary hiring or promotion decisions in the city needs a bias audit within the past year, a public summary of it, and notice to candidates ten business days before it is used. It does not reach every tool that handles employee data.

  • Illinois Human Rights Act, as amended by HB 3773 (opens in a new tab)Since Jan 1, 2026

    Employers tell applicants and employees when AI is used in decisions about hiring, promotion, discipline or discharge, and may not use AI that has a discriminatory effect, including through ZIP codes used as a proxy. Ordinary HR administration is not an employment decision.

A summary for orientation, not legal or professional advice. Each obligation links to its source. Last checked October 2026.

Schedule A

The firm’s systems we join

Time and billing
Practice management, LEDES, UTBMS codesTime entries, rates, invoices, write-downs and collections, by client and engagement.
Documents and precedents
Document management, email, matter workspacesAgreements, memos, workpapers and deliverables, with their versions and permissions.
Clients and pursuits
CRM, experience records, proposalsRelationships, pursuits, past proposals and the engagements each client has bought.
Intake and conflicts
Intake, conflicts checks, engagement lettersNew matters, the parties to them, conflicts checks and the terms each client agreed.
People and skills
People system, resourcing, CVsSkills, experience, availability and who worked on what.
Finance and projects
General ledger, receivables, project accountingRevenue, cost, unbilled time, receivables and project budgets.
Design records
BIM (IFC), drawings, specificationsModels, drawings and specifications, for architecture and engineering firms.

Check an AI use case against your professional duties

Pick a use case to see what it produces, the client information involved, the obligations that may apply, what has to be in place before it touches client work, and who would usually approve it.

Re

Drafting and reviewing contracts

For
Law firms

  1. What it produces

    What goes into a client’s agreement.

  2. Client information

    • Client confidential informationYes
    • Goes to a client, a court or a regulatorYes
    • Could be relied on as professional adviceDepends
    • May need the client’s consentDepends
    • Affects decisions about peopleUsually not

  3. Obligations that may apply

    • ABA Formal Opinion 512: client information goes into a tool that learns from it only with informed consent, and a lawyer reviews what it produces.
    • The client’s own terms: some engagement letters and outside counsel guidelines limit AI on their matters.

  4. In place before it touches client work

    • A firm-approved tool that does not learn from client documents.
    • Every clause compared with the firm’s own precedent and playbook.
    • A lawyer reviews every change before the client sees it.

  5. Suggested approval owner

    Practice-level approval

First step: AI & Agent Pilot. Case study: Contract compare-and-comply

Re

Legal research and court filings

For
Law firms

  1. What it produces

    What the firm tells a court.

  2. Client information

    • Client confidential informationDepends
    • Goes to a client, a court or a regulatorYes
    • Could be relied on as professional adviceYes
    • May need the client’s consentDepends
    • Affects decisions about peopleUsually not

  3. Obligations that may apply

    • ABA Formal Opinion 512 and the duty of candor: analysis and citations checked before anything is filed.
    • Court rules: some courts require AI use in a filing to be disclosed or certified.

  4. In place before it touches client work

    • Every citation checked against its source by a lawyer.
    • Drafts marked as drafts until a lawyer has reviewed them.
    • Each court’s own rules on AI checked before filing.

  5. Suggested approval owner

    Firm-level approval

First step: AI Governance Quickstart

Re

Time entries and invoices

For
Law firms

  1. What it produces

    What clients are billed.

  2. Client information

    • Client confidential informationYes
    • Goes to a client, a court or a regulatorYes
    • Could be relied on as professional adviceUsually not
    • May need the client’s consentUsually not
    • Affects decisions about peopleUsually not

  3. Obligations that may apply

    • ABA Formal Opinion 512 on fees: bill the time actually spent, and do not charge clients for learning a tool.
    • Client billing guidelines: the task codes, staffing and rates each client has agreed.

  4. In place before it touches client work

    • Timekeepers confirm every suggested entry.
    • Nothing billed that was not worked.
    • Every invoice checked against the client’s guidelines before it goes out.

  5. Suggested approval owner

    Practice-level approval

First step: AI & Agent Pilot

Re

Preparing returns and answering tax questions

For
Accounting and tax firms

  1. What it produces

    What goes on a return, and what clients are advised.

  2. Client information

    • Client confidential informationYes
    • Goes to a client, a court or a regulatorYes
    • Could be relied on as professional adviceYes
    • May need the client’s consentYes
    • Affects decisions about peopleUsually not

  3. Obligations that may apply

    • Section 7216: the taxpayer’s written consent before return information is used or disclosed outside the exceptions.
    • IRS Alert 2026-19 under Circular 230: verify facts, citations and calculations, use secure firm-approved tools, and keep written procedures.
    • FTC Safeguards Rule: a written information security program, and breach notice to the FTC.

  4. In place before it touches client work

    • A signed consent where it is needed, and a way to serve clients who decline.
    • Only the firm’s approved tool, inside its controls.
    • The preparer and the reviewer sign off every AI-assisted position.

  5. Suggested approval owner

    Firm-level approval

First step: AI Governance Quickstart

Re

Testing journal entries in an audit

For
Accounting and tax firms

  1. What it produces

    Which items the audit team investigates.

  2. Client information

    • Client confidential informationYes
    • Goes to a client, a court or a regulatorDepends
    • Could be relied on as professional adviceYes
    • May need the client’s consentUsually not
    • Affects decisions about peopleUsually not

  3. Obligations that may apply

    • PCAOB amendments, for public company audits of fiscal years from December 15, 2025: the data’s reliability evaluated, and flagged items investigated.
    • QC 1000, the new quality control standard, from December 15, 2026.

  4. In place before it touches client work

    • The reliability of the client’s data tested before the analysis.
    • Every flagged item investigated and documented.
    • The engagement team, not the model, reaches the conclusion.

  5. Suggested approval owner

    Firm-level approval

First step: AI & Agent Pilot

Re

Design checks against codes and standards

For
Architecture and engineering firms

  1. What it produces

    What goes into drawings and specifications a licensee seals.

  2. Client information

    • Client confidential informationDepends
    • Goes to a client, a court or a regulatorYes
    • Could be relied on as professional adviceYes
    • May need the client’s consentUsually not
    • Affects decisions about peopleUsually not

  3. Obligations that may apply

    • Responsible charge: the licensee controls the inputs, validates the outputs, and seals only work in their responsible charge (state board guidance, such as North Carolina’s in 2025).
    • Codes and standards: the editions adopted where the project is built.

  4. In place before it touches client work

    • The engineer of record reviews, and can reproduce, every check.
    • Inputs, outputs and validation kept with the project record.
    • Checks run only against the code editions that apply to the project.

  5. Suggested approval owner

    Firm-level approval

First step: AI & Agent Pilot

Re

A knowledge assistant over the firm’s own work

For
Every firm

  1. What it produces

    What people reuse and tell clients.

  2. Client information

    • Client confidential informationYes
    • Goes to a client, a court or a regulatorDepends
    • Could be relied on as professional adviceDepends
    • May need the client’s consentDepends
    • Affects decisions about peopleUsually not

  3. Obligations that may apply

    • Confidentiality: each client’s information is used only as that client’s terms allow.
    • Ethical walls and information barriers: people walled off from a client must not reach its files through the assistant.

  4. In place before it touches client work

    • The assistant follows every document’s existing permissions and walls.
    • Every answer cites the documents it came from.
    • Content owners keep the library current.

  5. Suggested approval owner

    Practice-level approval

First step: AI & Agent Pilot

Re

Proposals and marketing content

For
Every firm

  1. What it produces

    What prospective clients read about the firm.

  2. Client information

    • Client confidential informationDepends
    • Goes to a client, a court or a regulatorYes
    • Could be relied on as professional adviceUsually not
    • May need the client’s consentDepends
    • Affects decisions about peopleUsually not

  3. Obligations that may apply

    • Professional advertising rules, such as the lawyers’ rule against false or misleading statements about their services.
    • Confidentiality terms: which clients can be named, and which past work can be reused.

  4. In place before it touches client work

    • Only material cleared for reuse in the library.
    • Client names left out unless the client has agreed.
    • A partner approves every proposal before it goes out.

  5. Suggested approval owner

    Practice-level approval

First step: AI & Agent Pilot

Re

Screening candidates

For
Every firm

  1. What it produces

    Who gets an interview.

  2. Client information

    • Client confidential informationUsually not
    • Goes to a client, a court or a regulatorUsually not
    • Could be relied on as professional adviceDepends
    • May need the client’s consentUsually not
    • Affects decisions about peopleYes

  3. Obligations that may apply

    • These apply because the tool helps decide who is hired; ordinary HR administration is not covered.
    • New York City Local Law 144: a bias audit within the past year, a public summary, and notice to candidates.
    • Illinois, since January 2026: notice when AI is used in employment decisions, and no discriminatory effect.
    • Federal anti-discrimination law applies to AI-assisted hiring decisions as it does to any other.

  4. In place before it touches client work

    • A bias audit before use, repeated every year.
    • Candidates told before the tool is used.
    • A recruiter makes every decision.

  5. Suggested approval owner

    Firm-level approval

First step: AI Governance Quickstart

Re

A firm policy and how-to assistant

For
Every firm

  1. What it produces

    How people follow the firm’s policies and procedures.

  2. Client information

    • Client confidential informationUsually not
    • Goes to a client, a court or a regulatorUsually not
    • Could be relied on as professional adviceUsually not
    • May need the client’s consentUsually not
    • Affects decisions about peopleUsually not

  3. Obligations that may apply

    • Little beyond the firm’s own policies, provided client information is kept out of it by design.

  4. In place before it touches client work

    • Only the firm’s current policies, each with a named owner.
    • Every answer names the policy it came from and when that policy was last approved.
    • Questions about a client matter referred to the right partner.

  5. Suggested approval owner

    Tool-owner approval

First step: AI & Agent Pilot

A first read, not advice. What applies depends on your profession, your firm, your clients’ terms and how the tool is used; it is not a legal, ethical or professional determination. The obligations in full: § 5.

Suggested approval owners. Several may be needed: the firm approves the tool, the practice approves the workflow, and each client’s terms decide which matters it may touch. Your own governance decides who signs.

  • Firm-level approval: general counsel or the risk partner, information security, the client-consent question and quality review.
  • Practice-level approval: the practice leader, information security, testing on the firm’s own work, and a named owner.
  • Tool-owner approval: an owner for the content, access controls and a periodic check.

A use case not on the list, or one already in use without these safeguards? Thirty minutes is usually enough to decide what it needs before it goes further. If AI is already in use across the firm, start with the AI-use and client-terms review.

Talk through one AI use case

Three options for a first engagement

Set out as a proposal would: start from the question you have. The first two are our assessment and governance offers, scoped to that question, and none of the three commits you to the next.

  1. Option A

    Engagement profitability diagnostic

    For a managing partner asking why engagements underperform

    Two weeks, run as the AI & Data Assessment Sprint

    Utilization, realization and margin traced to the time, billing and staffing systems behind them; the gaps on one list with a view of the cost; and the next steps in order.

    Ask about a two-week sprint

  2. Option B

    AI-use and client-terms review

    For general counsel or a risk partner worried about AI already in use

    One real use case, run as the AI Governance Quickstart

    One real use case taken through your professional rules, the client’s terms and your firm’s policy to a decision, with the register and approvals the rest of the firm’s AI use can follow.

    Ask about an AI-use review

  3. Option C

    Data Maturity Assessment

    For anyone who would rather start on their own

    Free, about six minutes

    Fifteen questions on lineage, quality, ownership, access and platform. A partner can finish it between meetings; the result is instant and needs no email.

    Score your data foundation

After either review: a first build in one practice, two to four weeks once access and approvals are in place, measured against a number the practice already tracks, then a decision to scale, change or stop.

AI & Agent PilotData Platform Accelerator

Questions from managing partners, general counsel and IT

Will client documents leave our environment?

No. We build inside your cloud and your document controls, and what we build keeps the permissions and ethical walls you already have. Nothing learns from client documents unless you decide it should, with the consent that takes.

Some clients’ terms restrict AI on their work. Can we still use it?

Yes, where their terms allow, and not where they do not. We start by sorting clients by what their engagement letters and guidelines say, so tools are switched on client by client, and any consent you need is asked for once and recorded.

Our document, practice management and tax software already have AI features. Why build anything?

Use them where they fit; they are often the quickest route. We help you decide which to switch on and bring them under your AI policy, and build only where a question crosses systems no single vendor sees, such as margin across time, billing and staffing, or one view of a client across practices.

Can partners see what an engagement earns while it is running?

Yes. Time, rates, write-downs, staffing and billing are joined by engagement, with the definitions in § 3 agreed first, so partners see utilization, realization and margin while there is still time to restaff, reprice or talk to the client. It usually starts with one practice and the engagements its partners worry about most.

Who is responsible for AI-assisted work?

The professional who signs it, as before. The ABA, the IRS Office of Professional Responsibility and engineering boards say the same thing: AI assists, and the lawyer, the tax practitioner or the engineer of record reviews, decides and signs. We design the review into the tool rather than around it.

Do you work only with law firms?

No. We work with law firms, accounting and tax firms, consulting firms, and architecture and engineering firms. The work differs: for law firms it is usually precedents and billing; for accounting and tax firms, research, returns and audit analytics; for consulting firms, proposals and staffing; for architecture and engineering firms, project margin and standards.

Can we start with one practice?

Yes, and it is usually the right place to begin. A managing partner usually starts with the engagement profitability diagnostic, a two-week Assessment Sprint scoped to that question; a general counsel with the AI-use and client-terms review, run as the AI Governance Quickstart. The free Data Maturity Assessment takes about six minutes. A first build in one practice takes two to four weeks once access and approvals are in place, and ends with a decision to scale, change or stop.

Bring the answer that lives in one partner’s inbox.

The margin you see too late, the precedent nobody can find, or the AI tool your clients have not yet agreed to. Thirty minutes with the person who would lead the work, on the first thing to change.

Book 30 minutes on that answer

Margins seen too late? Thirty minutes on the margin, the precedent or the AI policy.

Book time on your firm’s data