This page explains how we build customized AI agents, firm by firm, case by case. Not a license fee.
Introducing Fresh Eyes of Alveia for legal work.
We provide an alliance.
It started with our founder Alfred Vanderpol. Certified in AI Fluency by Anthropic, he applied his AI agents to his own legal cases before offering it to anyone else.
It worked, and now it's available to the people who need it.
The AI now being sold to law firms and practices under monthly licenses mainly searches jurisprudence, checks norms, standardizes documents. It knows the general practice of the law, it doesn't know you. Alveia builds dedicated AI agents around your case file, your needs, and your specificities. It runs the analysis, checks the documents, and delivers a strategic brief for your review — whether the matter goes to court or not.
Two ways to work with us.
You bring the file, Alveia builds the agent in-house, and we deliver the strategic brief, contracts or the courthouse document, ready for your lawyer of record to review, sign and deliver.
Alveia builds the agent to your specification and installs it in your own AI account — Claude, ChatGPT, Grok, etc., whichever you prefer. It's yours from that point on. Training and updates are available as you see fit.
If you're not tech savvy, the first mode delivers the result. If you are, the second puts the instrument in your hands.
In either mode, the client remains yours. The fee remains yours. The name on the document is yours. What the alliance changes is the depth and speed of the work behind it.
Since 2025, virtual employees built by Alveia have been drafting litigation strategy and reviewing contracts on live matters, reviewed by lawyers of record before anything is filed or signed.
From analysis of the order of direct award of property, writs of execution for delivery of possession, motion to void proceedings, appellate ruling on ejectment, to writing court filings for the assessment of default interest, disclosure of assets and attachment, service by affixing notice at the address, authorization to act on non-business days and outside court hours, and the appointment of a different court bailiff, our AI agents have been there to help.
The most recent matter shows how the alliance works in practice. A firm had won a civil case at first instance, then lost on appeal. Preparing for the supreme court, the team came to us: they sent the underlying contract, the trial judgment and the appellate ruling. We build a specialized agent for the case only and it returned a full assessment of the odds of prevailing — the procedural route and whether the ordinary remedies had been exhausted; a principal claim of constitutional violation, on the ground that the appellate court mischaracterised the underlying contract and read it against its own text, in breach of the statutory rules of contractual interpretation; a demonstration of why the error determined the outcome of the judgment; and the relief to be sought. Alongside it came the opposing party's strongest arguments, each graded and answered, and the questions of fact still to be verified in the record before anything is filed.
The analysis and strategic brief resolved the question of how to attack the appellate ruling. The firm then asked for the agent to draft the documents for filing — and after careful review, the owner of the firm filed it under his own signature.
While the AI agents work, the lawyers are where they're most needed.
What used to take weeks now takes days. Leaving plenty of time to refine the documents.
The agent works at a depth a person can't sustain. It holds all the files at once — every document, every cross-reference — and reads all of it with the same attention on every document, on every page.
The firm can show a real AI capability. A software subscription is invisible to the client. An alliance is something the firm can put its name to.
There's no monthly license and no per-seat fee sitting on your books between cases. You pay for the matter in front of you, not for a login nobody's using in the meantime.
Every brief is tested against the strongest arguments the other side could make — before it is filed, not after.
At Alveia, when we build an AI agent, we're not writing a simple assistant for repetitive tasks — we're writing a functional virtual employee with a complete job description. Think of it exactly like hiring someone: you define the profile you're looking for, the tasks required, the environment for the job, what you need produced, how you want it done, what cases they're operating inside. And if you show them examples of excellent work, they perform much better. An agent works the same way, except it does it on the first minute of the first hour of the first day.
An agent is built by what's called a prompt — and a prompt is nothing more than a text structured by sections that define the architecture of the virtual employee. At Alveia, we've developed and refined our own blueprint for this: a standard six-section structure we build every agent against, so that quality never depends on who wrote it or what day it was written.
It establishes who the agent is: its expertise, the methods it works from, the credentials it carries.
Defines the one thing it produces, for whom, and to what standard.
Holds the rules and constraints — and the reasoning behind them, because an agent that understands why a rule exists will apply it correctly to situations we never anticipated.
Gives it the business it's working inside: the client, where it sits in the workflow, what its output feeds, and what a weak output actually costs downstream.
The highest-leverage section of all — real samples of excellent work, because these models imitate far more reliably than they follow written instruction.
Carries the final constraints: output format, language rule, scope guard, fallback behaviour.
An Alveia AI Agent is a specialist that performs to the same standard on the hundredth job as on the first — no off days, no drift, no retraining.
Once it exists it's a valuable asset — which is why we build a team of narrow specialists rather than relying on one generalist that's adequate at everything and excellent at nothing.
Contact us.