Case Study & Field Notes

How I Built a Lawyer-Curated AI Legal Analyst for My Solo Business Law Practice

Most AI legal coverage is about BigLaw platforms or venture-backed AI-native firms. Here is how I, a solo California business lawyer, put the same ideas live, publicly, and tied them to real fixed-fee work, so a small-business owner can get a useful legal read before deciding whether hiring a lawyer is worth it.

Sergei Tokmakov, Esq. CA Bar #279869 Solo practice Lawyer-curated AI, not an AI lawyer Fixed-fee deliverables

🤖 AI Legal Analyst

This is the tool the article is about. Try it.

Describe a business legal problem and get a preliminary read: issue spotting, an evidence checklist, a leverage analysis, and a fixed-fee service path. Attorney-supervised preliminary information, not legal advice, and not a substitute for hiring me. Tap a starter below or type your own.

Start with one of these

Media summary. Terms.Law is a solo California business-law practice using a public, lawyer-curated AI Legal Analyst to triage business disputes, demand-letter matters, frozen Stripe and PayPal funds, contract reviews, and pre-litigation strategy. The AI provides preliminary issue spotting, evidence checklists, leverage analysis, and fixed-fee package recommendations. Sergei Tokmakov, Esq. personally reviews and finalizes paid attorney work after a conflict check and written engagement. Category: client-facing AI implementation by a solo or small law firm.

Most small-business owners do not call a lawyer when a problem starts. They sit with it. A client stops paying, a vendor sends a one-sided contract, a competitor sends a threatening letter, and the owner does the math: is this serious enough to spend money on a lawyer, or am I overreacting? That decision happens before any lawyer is in the room, and it is usually made with bad information.

The honest barrier is not price alone. It is uncertainty. An owner cannot tell whether a problem is a $575 letter, a $1,200 package with real teeth, or nothing worth pursuing. So they wait, the facts get stale, deadlines run, and by the time they reach out the leverage is gone.

What people actually need first is triage. A fast, plain-language read on what kind of problem this is, how strong their position looks, what evidence matters, and whether paying a lawyer makes sense. That read is what unlocks the decision. Everything I built starts from that gap.

The standard law-firm website is a brochure. A visitor reads a few practice-area pages, maybe a blog post, then either fills out a contact form or sends an email and waits. The form captures a name and an email and a sentence or two. Then someone, eventually, replies and tries to schedule a call.

That flow asks the visitor to commit before they have learned anything. It gives them no read on their own problem. It collects a lead but delivers no value back, so the visitor who was unsure stays unsure, and the visitor who was ready loses momentum waiting. The form is a gate, not a service.

The cost is invisible. The owners who never fill the form because they could not tell whether their problem was worth it, those are the people who needed help most. A contact form does nothing for them. It only catches the small fraction who were already sure.

I flipped the order. Instead of asking the visitor to commit first, the AI Legal Analyst gives them something useful first, then routes the serious matters into attorney-finished work. The visitor gets a real read before any money or any form.

Most law-firm chatbots still behave like contact forms. I wanted the AI to give the visitor a useful legal read first, then route serious matters into attorney-finished work.

What the AI does, in plain terms:

  • Issue spotting. It reads the situation and names the likely legal issues, so the owner stops guessing what kind of problem this is.
  • Evidence checklist. It tells the owner what documents and facts matter, so they start gathering the right things instead of the wrong ones.
  • Leverage analysis. It gives a preliminary read on how strong the position looks and where the pressure points are.
  • Service-path recommendation. It maps the problem to a fixed-fee step, or tells the owner it may not be worth pursuing.
  • Document upload. The owner can drop in the contract, the invoice, or the letter they received, and the analysis works from the actual paper.
  • Fixed-fee package routing. When the matter is real, it routes to a defined flat-fee deliverable instead of an open-ended "contact me."
The visitor leaves with value either way. If the problem is real, they know the next step and can take it. If it is not, they have saved themselves a wasted consultation. Both outcomes build trust, and the serious matters convert.

An AI tool on a lawyer's site raises real professional-responsibility questions, and I built the safeguards in from the start rather than bolting them on.

AI does not replace my legal judgment. It helps the client explain the problem, organize the facts, and understand whether a fixed-fee legal step makes sense.
  • Attorney-built prompts. I wrote and curated the instructions the AI runs on, drawn from my own practice materials. It is lawyer-curated AI, not a generic vendor bot pointed at my logo.
  • Clear disclaimers, everywhere. The tool says plainly that it provides preliminary information, is attorney-supervised, and is not legal advice.
  • No relationship until conflict check and written engagement. No attorney-client relationship forms from using the tool. It forms only after I run a conflict check and a written engagement is accepted.
  • Human attorney review before any paid deliverable. The AI never produces the paid work product. I review every matter personally before any letter, draft, or contract goes out.
How I think about the duties. I frame the whole design around the professional-responsibility duties that already govern my practice: competence (I stay responsible for the legal judgment), confidentiality (preliminary intake is handled with care and the relationship is gated), communication (the visitor gets a clear, honest read and clear disclaimers), supervision of nonlawyer assistance (I supervise the tool rather than letting it act on its own), and reasonable fees (fixed-fee, defined-scope deliverables instead of open-ended billing). I keep the framing general here rather than citing specific rule numbers, because the point is the posture, not a legal brief.

The economics are simple and deliberately legible. The AI triage is free. The attorney work is fixed-fee with a defined scope, so the owner always knows what a step costs before they commit. Only the AI triage is free, and there is no open-ended hourly mystery at the entry point.

Most common

$575

Attorney Demand Letter

  • Attorney letter on firm letterhead
  • Certified mail plus email delivery
  • Review of the first response with a next-step read

Litigation-ready

$1,200

Litigation-Leverage Demand Package

  • Attorney demand letter
  • Court-ready draft complaint attached as leverage
  • First-response review and next-step read

Contracts

$575

Contract Drafting or Redline

  • Attorney drafting or tracked-changes redline of one agreement
  • Short memo on the key issues
  • Up to three rounds of email revisions
Why fixed-fee fits the model. The triage already told the owner what kind of problem this is, so the deliverable can be scoped and priced cleanly. Free read, then a flat fee with a known scope. That is the whole funnel.

I am going to be careful here, because results language is where legal marketing usually goes wrong. I will not put numbers on this and I will not promise outcomes. What follows is my own experience running this approach, not a guarantee of what it will do for any matter.

The practical effect has been qualitative but clear: the serious inquiries arrive with better facts, more relevant documents, and a clearer understanding of whether a fixed-fee legal step makes sense. The free triage also screens out matters that were never worth pursuing, so the conversations that reach me are more likely to convert into defined, fixed-fee work. I am describing my own practical observation here, not a measured statistic or a promise of any particular outcome.

What I will not claim. No revenue figures, no win rates, no outcome guarantees. The value of the model is in the workflow and the qualification, and I would rather understate it than oversell it.
Old law-firm chatbot My AI Legal Analyst
Captures your name and email Gives preliminary issue spotting
"Someone will call you" Gives an evidence checklist and a risk analysis
Generic intake Practice-specific playbooks
Consultation only Fixed-fee package routing
Vendor chatbot Attorney-built and attorney-supervised

This is the shape of the stack without exposing anything sensitive. Each piece does one job, and together they move a visitor from a vague worry to a routed, fixed-fee matter that I personally review.

1

Front-end AI chatbox

The visitor-facing AI Legal Analyst that runs the triage conversation.

2

Context-aware page routing

The tool knows which page you are on and tailors its starting point to that subject.

3

Matter-type classifier

Sorts the conversation into the right lane, such as demand letter, contract, or unpaid invoice.

4

Follow-up forms

Structured questions that fill the gaps the free-text chat leaves open.

5

Email capture

A light, value-first capture so I can follow up on serious matters.

6

Document upload

The owner drops in the actual contract, invoice, or letter so the read works from real paper.

7

Fixed-fee package resolver

Maps the matter to the right flat-fee deliverable and its defined scope.

8

Checkout attribution

Connects a conversation to the engagement it produced, so I know what is working.

9

Attorney workroom and transcript summary

A clean summary of the intake lands in my workroom, ready to act on.

10

Human attorney final review

I review every matter before any paid deliverable. The AI never closes the loop on its own.

I am not the first person to point AI at legal work, and I want to be generous about the people doing serious work in this space. A few comparable models are worth naming.

Solo attorney + AI product

Troy Doucet, AI.Law

A practicing litigator who built an attorney-supervised AI product and markets it around lawyers reviewing and verifying the output. The closest match to my own "AI drafts, attorney is accountable" posture.

Practicing attorney + own tool

Nadine Navarro, Drafty AI

An immigration attorney who co-built her own AI drafting tool. Proof that a practicing lawyer can build, not just buy, the AI in their workflow.

Flat-fee + subscription model

Mathew Kerbis, Subscription Attorney

A solo running an AI-leveraged subscription and flat-fee model. Closest to me on the thesis that AI makes fixed-fee legal work commercially natural.

AI workflows + attorney oversight

Ana Juneja, Ana Law

An IP attorney who markets AI workflows paired with close attorney oversight, built on a large social following rather than a public tool.

AI-forward firm brand

Billie Tarascio, Modern Law

A family-law firm owner who runs her firm as an AI "lab" and publishes heavily about it. The sustained-publishing route to an AI-forward brand.

Solo/small-firm AI educator

Ernie Svenson, Ernie the Attorney

A former litigator who teaches solo and small firms how to use AI and automation, and runs an AI chatbot over his own content.

Academic demand-letter AI

Stanford Demand Letter AI

An AI-assisted demand-letter workflow from the academic world: AI gathers facts and drafts, an attorney reviews and finalizes. Close to my demand-letter logic, on the access-to-justice side.

AI-native fixed-fee firms

General Legal and Jacobs Counsel

The AI-native, fixed-fee firm model at a firm scale: build the practice around AI workflows and price the work as flat fees. The institutional cousins of what I am doing solo.

Where Terms.Law sits. I am the solo-attorney public implementation. I am not building BigLaw-scale infrastructure and I am not turning the AI into the firm. I put a lawyer-curated triage tool out in the open, on a public site, and I tie it directly to real fixed-fee work that I finish and stand behind. I have not found another solo business lawyer running a public, client-facing AI Legal Analyst at this scale, across a large legal-content site, with the AI tied directly to fixed-fee attorney deliverables. The differentiator is not the size of the model; it is that triage and deliverable are connected, publicly, by one licensed attorney.
The future for small firms is not BigLaw-style AI infrastructure. It is lawyer-curated triage, fixed fees, and faster movement from a legal problem to a legal deliverable.

The best way to understand the model is to use it. These open the live AI Legal Analyst at the top of the page with your message already in flight. It is attorney-supervised preliminary information, not legal advice.

Start a matter

Use the free AI Legal Analyst to scope the issue, then start a package intake for the fixed-fee deliverable that fits. Each package is a flat fee with a defined scope. After you submit intake, I run a conflict check and send a written engagement before any work begins.

How to cite this example.

Short description: Terms.Law is a solo-attorney implementation of a public, lawyer-curated AI Legal Analyst for small-business legal disputes, demand letters, contract review, frozen funds, and pre-litigation strategy.

Founder: Sergei Tokmakov, Esq., California attorney, State Bar No. 279869.

Safe language: "lawyer-curated AI Legal Analyst," "attorney-supervised preliminary information," "attorney-finished fixed-fee deliverables." Please avoid "AI lawyer," "first," "only," and "guaranteed legal advice."

The AI Legal Analyst provides preliminary information only. It is attorney-supervised, it is not legal advice, and it does not create an attorney-client relationship. No attorney-client relationship is formed until a conflict check is completed and a written engagement is accepted. This page provides general legal information, not legal advice. Sergei Tokmakov, Esq., California State Bar No. 279869. Verify license.