Industries · Professional services
When your product is expertise, AI changes the economics.
Law firms, tax advisors, accountants, consultancies, agencies and engineering offices sell judgement by the hour. AI agents take on the reading, drafting, searching and checking around that judgement, which raises the question of what to do with the hours saved. We help you deploy the agents under professional secrecy and answer that question deliberately.
Why professional services
No industry has more of its work in documents, and no industry has stricter rules about who may read them.
A professional-services firm is a knowledge base with people attached. Engagement files, precedents, templates, research memos, working papers and client correspondence hold most of what the firm knows, and every new engagement starts by finding and adapting what already exists. Language models are built for exactly this kind of work: reading, comparing, summarising and drafting from prior material.
At the same time, the constraints are among the strictest anywhere. Professional secrecy obligations for lawyers, tax advisors and auditors, client confidentiality, liability for advice, and quality standards that require a qualified professional to sign off. Those constraints decide where the agent may run, what it may see and who releases its output. They do not decide whether the firm can use AI at all.
The second question is commercial. If a task that took four hours takes one, a firm billing by the hour has a pricing problem and a firm on fixed fees has a margin opportunity. We treat this as part of the engagement, because a deployment that works technically and hurts the business is not a success.
- Typical starting pointInternal knowledge assistant or proposal drafting
- Human checkpointA qualified professional releases every client deliverable
- SystemsDMS, practice management, email, time and billing
- ConstraintProfessional secrecy, confidentiality, data residency
Use cases
Seven workflows that pay off in firms of ten to several hundred professionals.
Proposal and pitch drafting
From the RFP or the intake call, the agent assembles a first draft from your credentials, team profiles, prior proposals and rate structures, tailored to the client's situation. Partners edit and price; the agent handles formatting and consistency.
Research and document review
First-pass review of contracts, financial statements, technical reports or discovery material against a checklist, with findings cited to the page. Research memos drafted from your knowledge base and permitted external sources, verified by the professional who signs.
Knowledge management across engagements
A permission-aware assistant over engagement files, precedents and templates: "how did we handle a similar clause last year?" answered with citations, respecting information barriers between matters and clients.
Time and billing narratives
The agent drafts time-entry narratives from calendar, email and document activity for the professional to confirm, and prepares invoice descriptions in the client's required format. It improves billing accuracy without deciding what is billable.
Client onboarding and KYC document checks
Collecting and checking identification documents, beneficial-ownership information and engagement letters against your onboarding checklist and anti-money-laundering obligations, with the compliance officer deciding on flags.
Report and deliverable drafting
Structured deliverables such as due-diligence reports, tax memos, audit documentation and consulting reports drafted from working papers and templates, in your house style, with every figure traced to its source and a partner review before release.
Internal helpdesk
Questions about firm policies, IT, HR, templates and procedures answered from your intranet with citations, so support staff and knowledge managers are not the search engine for everyone else.
Constraints specific to professional services
Professional secrecy and confidentiality
Lawyers, tax advisors and auditors are bound by professional secrecy obligations that restrict who may process client material and under what contractual conditions. Hosting, provider contracts and access design have to satisfy them. In practice this points toward EU-hosted or on-premises models for client files and contractual no-training guarantees everywhere. This is a design constraint we take from your professional body's guidance; it is not legal advice.
Information barriers between matters
A knowledge assistant that reaches across all engagements is a conflict-of-interest problem unless permissions follow the same barriers as your DMS. Permission-aware retrieval is a hard requirement, not a feature.
Liability and sign-off
Advice carries liability, so every client deliverable is released by a qualified professional. The agent's role is to make that professional faster and better informed, and to document what it contributed.
Data residency and client contracts
Many client engagement letters restrict where data may be processed. The hosting decision has to respect the most restrictive client, or the firm needs separate configurations per client class.
Billing-model implications
Hours saved change the business. We help firms decide, engagement type by engagement type, whether to keep hourly billing with higher leverage, move to fixed fees, or productise services, before the agents make the decision for them.
How we start
Assess
Two to three weeks: engagement types and volumes, where professionals' hours go, the DMS and practice-management landscape, confidentiality obligations and client contract constraints. Output: ranked use cases, a hosting recommendation and a first view on billing implications.
Design
Permission model mirroring your information barriers, provider and hosting selection that satisfies professional secrecy, review and sign-off workflow, and the evaluation set built from anonymised past work.
Pilot
Six to eight weeks to a production assistant for one practice group or one workflow, used on live engagements by a pilot group, measured on hours per task and quality as judged by partners.
Scale
Roll out to further practice groups, add drafting and review workflows, train knowledge managers to maintain the corpus, and set up managed AI operations or an internal owner.
Frequently asked questions
Can we use AI on client files under professional secrecy?
Firms do, with the right setup: models hosted in the EU or on your own infrastructure, provider contracts with confidentiality and no-training clauses, access controls that mirror your matter permissions, and documentation of all of it. The specific requirements depend on your profession and your bar or chamber's guidance, which we work through with your compliance partner. We provide the technical and process design, not legal advice.
Will clients accept that AI was used on their matter?
Increasingly they expect it, and some ask for it in the engagement letter. Transparency helps: a clear description of what the agent does, what it does not see and who reviews its output. We help firms write that description and the corresponding clause.
How do we keep the quality of advice?
The same way as today: a qualified professional reviews and signs. What changes is that the professional reviews a cited draft rather than starting from nothing. During the pilot we have partners blind-rate agent-assisted and conventional work, so the quality question is answered with evidence.
What about our hourly billing model?
It is the most important commercial question in this industry and we treat it as part of the project. Options include higher leverage per professional, fixed-fee products for standardised work and value-based pricing for advice. The right answer differs by practice area, and the data from the pilot tells you where each applies.
Which systems do you integrate with?
Document management systems such as iManage and NetDocuments, practice-management and time-and-billing tools, Microsoft 365 and email, and, for tax and accounting firms, DATEV and similar platforms, through APIs or the Model Context Protocol where available. We verify the concrete landscape during the assessment.
How large does a firm need to be?
The workflows pay off from around ten professionals, because the knowledge base and the document volumes already exist. Larger firms benefit from the same patterns at scale and need more attention to information barriers and governance.
Related
Enterprise knowledge assistants (RAG)
Assistants that answer from your SharePoint, Confluence, DMS, ERP and ticket history: permission-aware, with citations, evaluated, hosted in the EU.
AI agents for legal & compliance
First-pass contract review, obligation tracking, regulatory monitoring and policy Q&A, with a lawyer approving everything that leaves the department.
AI governance, EU AI Act & GDPR
An AI register, risk classification under the EU AI Act, GDPR-aligned processes and a usage policy your teams will actually follow, built together with your lawyers and your data protection officer.
Sovereign AI: options for EU-hosted language models
US providers with safeguards, hyperscaler EU regions, European providers or self-hosted open-weight models: how to choose by data sensitivity and use case.
Let's find the first workflow worth automating.
A 30-minute intro call, no slides and no obligation. We listen, ask about your processes, and tell you honestly where AI agents would pay off and where they would not.