01
Have the model conversation.
Pricing, leverage and what to sell when the hours shrink, run as a management decision with the arithmetic on the table.
THE ARITHMETIC
The arithmetic your partner meeting keeps postponing
In every other sector, AI is an efficiency question. In yours it points straight at the revenue line, because the unit you sell is the unit AI compresses. Every partner knows this. Few say it out loud in the partner meeting.
Two default reactions are circulating, and both are wrong. The first is to ban the tools, which moves the usage to private phones and takes the client material along.
The second is to buy licences and call it a strategy, which is worse than it sounds, because of the arithmetic: under hourly billing, every hour AI saves is revenue you hand back. Under fixed fees, every hour AI saves is margin you keep.
The technology is neutral. The pricing decision is not. Faster is only cheaper if you decide it is.
The billable hour is not dying. It is becoming a choice. Which matters you price by the hour, which by outcome, and which you turn into products: that is the management conversation, and it belongs before the tool decisions, not after them.
Your corporate clients' procurement functions have started asking what AI you use and why the fee looks the way it does. It is better to have decided your answer before the question arrives.
CONFIDENTIALITY
Client confidentiality is your sovereignty analysis
The one hard constraint in this sector is genuinely yours: professional secrecy. Privilege in law, statutory confidentiality in audit, and client material under a non-disclosure agreement (NDA) everywhere else.
Under the Clarifying Lawful Overseas Use of Data Act (CLOUD Act), an American provider can be compelled to disclose data wherever the server stands. Your client did not sign up for that audience.
This is the same sovereignty analysis we run for the defence sector, in civilian clothes. The answer is not a ban. A firm that bans AI outright gets shadow use on private phones.
A firm that draws the boundary gets both safety and speed: which tools may touch which classes of material, which deployments stay inside your walls, and what runs on the open tools without a second thought.
In most sectors, unsanctioned AI use is a governance gap. In a firm bound by professional secrecy, it is a client confidentiality incident.
Here our independence stops being a virtue statement and becomes operational. We hold no partner programmes with the cloud providers and earn nothing on licences, so we can recommend the deployment your secrecy rules actually allow, including the ones that never leave your building.
THE ARCHIVE
Your archive is the asset
The moat in a professional firm was never the model. It is the archive: twenty years of matters, memos, calculations, templates and judgement calls, sitting in folders nobody can search.
AI finally makes that asset liquid. Drafts that start from your own precedents. Review that accelerates instead of queueing. Retrieval that answers like the colleague who has seen the matter before. Juniors who ramp on the firm's actual knowledge instead of folklore.
Built on your archive, inside your confidentiality boundary, owned by the firm at handover. We build, we transfer, we leave, and what remains is yours.
And everything we build here is decision support by nature: the AI drafts, the professional decides, the signature stays human. In your profession that is not a limitation of the technology. It is the product itself.
Your clients buy judgement. AI under judgement strengthens what they buy. AI instead of judgement destroys it, and we will be the first to tell you where that line runs.
THE RULES
The rules, briefly
You will read less about the EU AI Act on this page than anywhere else on this site. For most of what a professional firm runs, the regulation asks modest things: staff who use AI competently, transparency where content is AI-generated, and order among the tools you rely on.
The AI literacy requirement in Article 4 and the prohibitions have applied since 2 February 2025, the penalty rules in Article 99 since 2 August 2025, and the transparency obligations in Article 50 start to apply on 2 August 2026. No high-risk drama, no regulatory moat, and we will not manufacture one to sell you urgency.
The fundamental rights impact assessment in Article 27 does not reach you, as long as the firm neither provides public services nor deploys systems for creditworthiness or for the pricing of life and health insurance. That releases you from neither the literacy requirement nor the transparency obligations.
Use AI to screen your own candidates, on the other hand, and that is Annex III point 4, high-risk, with requirements from 2 December 2027, whatever the size of the firm. Annex I, AI built into products, applies from 2 August 2028. No harmonised standards are published yet, so compliance is shown with your own documentation.
The binding rulebook in your world is older and stricter: your professional conduct rules, your secrecy obligations, and your clients' terms. We de-dramatise the regulation and take the confidentiality seriously. Most sellers do the opposite.
WHERE TO START
Where to start
02
Draw the confidentiality boundary.
Which tools, which material, where it runs: a policy your partners can sign and your clients can read.
03
Assess before you buy.
An AI Readiness Assessment scoped to a professional firm: the archive, the matter mix, the secrecy constraints, the quick wins.
04
Build on the archive.
Implementation on your own precedents and templates, inside your boundary, owned by the firm at handover.
05
Train partners and staff.
Leadership sessions from us, breadth through AIUC, our education arm, with professional secrecy built into the curriculum.
PROOF
Our work spans the legal world and the advisory professions: Vänersborg District Court (Vänersborgs tingsrätt), audit and accounting practices, public affairs advisers and communications agencies across the Nordics. And law firms that prefer, fittingly, not to be named.
QUESTIONS
Before you let AI into client work.
AI replaces tasks in a law firm, not the profession, but it does change the leverage model, and pretending otherwise helps nobody. What compresses, what does not, and what the pyramid should look like at your matter mix is a partner decision. We bring the arithmetic, so the decision is made with numbers instead of slogans. The basis comes out of an AI Readiness Assessment.
Client material does not belong in public AI services without a confidentiality boundary someone accountable has drawn. Some deployments and some classes of material: yes. Others: never. The question is settled by professional secrecy and your clients' non-disclosure agreements rather than by the EU AI Act, and by the fact that under the CLOUD Act an American provider can be compelled to disclose data wherever the server stands. Some of your people already do it, so the fix is a fast permitted route rather than a sterner memo.
A firm facing a demand for an AI discount should answer with a straight account of how the work is done and a fee model it can defend. The efficiency is yours to price, not the client's to claim. We build both before your clients' procurement builds them for you, and write in the disclosure that the transparency obligations in Article 50 of the EU AI Act require from 2 August 2026.
Yes, AI used to source or screen candidates is high-risk under the EU AI Act, including in a firm that otherwise never comes near the high-risk rules. Recruitment and selection sit in Annex III point 4, and the requirements apply from 2 December 2027, whatever the size of the firm. It is the most common route in for a professional firm. The Article 27 impact assessment, by contrast, does not follow the high-risk list and normally does not bind a firm.
A twenty-person law firm is reached by the AI literacy requirement in Article 4 of the EU AI Act, which has applied since 2 February 2025 and sits with management, and by the transparency obligations in Article 50 from 2 August 2026. Size exempts you from neither. The work itself scales down cleanly: a confidentiality boundary, an assistant on your own archive, and one pricing decision, at fixed price and owned by the firm at handover.
Yes, a firm publishing AI-generated material on a client's behalf has to disclose the synthetic content. The marking requirement in Article 50(2) of the EU AI Act reaches synthetic content as such, from 2 August 2026, and generative systems already on the market have until 2 December 2026 under Article 111(4). The exception that matters most to a firm sits in Article 50(4): text published to inform the public on matters of public interest need not be marked where a human has reviewed the content and someone carries editorial responsibility for it. That editorial review is something a firm can actually organise.
No, Ampliro does not give legal advice to a firm. You are the lawyers. We design the machinery around your judgement: the confidentiality boundary, the build and the adoption. The assessment of privilege and professional conduct under the Swedish Bar Association's rules stays exactly where it belongs, with you. What we answer for is the named deliverables: the boundary, the implementation, and the documentation that shows compliance while no harmonised standards are published.
INSIGHTS
Recent analysis for this sector.
Decision support
Who gets the hour AI saves
71 percent of in-house counsel expect their outside firms to change commercial models. 28 percent of law firms have changed pricing in response to AI, and in that distance it gets decided whether the gain becomes a discount or a margin.
Read the analysis
Decision support
Who decides whether the data may sit with an American provider
The CLOUD Act follows the provider's control over the data, not the address of the server. What decides the question is not the American statute but which of your own rules is the strictest.
Read the analysis