ERISA Benefits Tech
ERISA SKILLs Suite · for power users

Twenty-six ERISA practice areas, installed in your own Claude account.

The Suite is the power-user version of the Desk Reference research assistant: the same ERISA knowledge and the same way of working through a question, installed as AI SKILLs in the Claude account your benefits team already has. There it works alongside your plan documents, your Projects and Claude’s memory, which stay in place from one conversation to the next, and it produces analysis and first drafts as well as answers. It is the skill library we built for our own ERISA practice, tuned for plan sponsors and benefits professionals. It comes with a getting-started SKILL that teaches your team to set it up and use it well, and a perpetual license.

Buy the SKILLs Suite → Compare it with a subscription →

Why the Suite

Four ways general AI gets ERISA wrong, and how the Suite closes them.

Ask a leading model a textbook question and it will often answer well. Ask it the question your team actually has, the one that turns on your plan's language, your service-crediting method or your committee's past practice, and it will answer just as confidently while answering a different plan's question and leaving out issues that matter.

Gap one

It does not know how to think about ERISA's myriad interlocking provisions

General AI answers from broad knowledge of ERISA, but it can miss the related issues a question touches, such as the testing failure that follows a deferral or match correction, and it is as confident about an issue it should have left aside as about the one that matters in your case.

What closes it: instructions on how to work through each ERISA question, and on what else to consider when one issue is raised, the way ERISA professionals approach the work.

402(g) 401(k) SPD ADP EPCRS 415(c) Your question general AI reaches these the SKILLs also reach these
Gap two

It doesn't know your plan

General AI knows a great deal about 401(k) plans but has not read your adoption agreement, so when eligibility, vesting or a distribution turns on a design choice the statute permits several ways, it supplies the most common answer rather than the one your plan chose.

What closes it: your own documents. The getting-started SKILL walks you through setting up a workspace holding your plan documents, SPDs and forms, and confirming that the AI is actually reading them.

THE PROVISION Eligibility service Elapsed time 1,000 hours — the common answer Monthly equivalency — yours ? general AI guesses here your adoption agreement, in your own Project
Gap three

It's stale in a field that moves

With SECURE 2.0 provisions phasing in, agency positions shifting and litigation reshaping fiduciary practice, an answer that was right eighteen months ago may be wrong today, and a general model's knowledge stops at its cutoff date.

What closes it: dated releases and currency checks built into the SKILLs. The SKILLs tell the AI to check current figures for the items that change most often, such as contribution limits, ACA affordability percentages and excise tax amounts, and each release states the date its law is current to and says so when currency matters.

2023202420252026 model knowledge cutoff known known new contribution limits agency position shifts SECURE 2.0 phase-in SKILL: “CHECK CURRENT FIGURES” what a general model answers from memory
Gap four

It doesn't know when to stop

A general model does not distinguish an administrative question it can handle from a judgment call that needs a lawyer, and it answers both in the same tone, at the same length and with the same confidence.

What closes it: criteria for recognizing a judgment call, and a notice that appears whenever one arises, telling you plainly not to act on the answer until ERISA counsel has reviewed it.

A question arrives turns on judgment? NO Answer proceeds YES ERISA counsel reviews first ⚠ JUDGMENT CALL — DO NOT ACT ON THIS. Fires automatically. Cannot be turned off.
Straight answer first

What you are buying, and what you are not.

This is a set of AI SKILLs designed to tell Claude how to think about the ERISA issues that plan sponsors — and the third parties that advise them — face every day. It is not custom designed for particular employers, but it includes features that help you to give Claude your specific ERISA facts so that its output is more precise.

What you get

  • Twenty-six substantive ERISA SKILLs, each covering one practice area, with reference summaries, checklists, and worked numerical examples.
  • A master SKILL that routes a question to the right practice area and holds the house analytical conventions — including what else to consider when one ERISA issue is raised, and step-by-step instructions for working through each question the way ERISA professionals do.
  • A getting-started SKILL that walks your team through setting up a Project, loading your own plan documents, and prompting well.
  • Currency checks built in. The SKILLs tell the AI to check for updated figures on the items that change most — contribution limits, ACA affordability numbers, excise tax amounts — and each release states the law as of a given date and says so when currency matters.
  • A perpetual license. The release you buy does not expire and does not stop working.
  • Judgment-call flagging. The SKILLs recognize when a question turns on legal judgment rather than a plan term, and say so before you act.

What you do not get

  • We do not load your documents. You do that yourself — the getting-started SKILL teaches you how, and it takes an afternoon, not a project.
  • No attorney reviews the output. Not before you see it, not after, unless you engage competent ERISA counsel to review it and advise you as to your specific situation. Every answer is unreviewed by definition.
  • No customization. Your house positions and interpretations are not encoded. Nothing is built to your specification.
  • We host none of the Suite. The SKILLs run inside your own Claude account, under your own agreement with Anthropic.
  • The Desk Reference is not included. The Desk Reference and its research assistant come together, as a separate subscription.
  • Installation is yours. Roughly fifteen minutes, no developer tools, no command line — but it is your fifteen minutes.
Coverage

Twenty-six practice areas, plus setup and routing.

Organized the way benefits work is actually done — by the question in front of you, not by statute. You will not use all of them. A sponsor with a 401(k) and a health plan may only ever use ten or twelve SKILLs. But they are there when you need them, even when you don’t know you need them.

01 · THE QUESTION YOU ASK 02 · THE SKILLS IT ACTIVATES 03 · YOUR OWN DOCUMENTS 04 · WHAT COMES BACK “An employee asked HR to raise their deferral from 3% to 5% six months ago. Nobody processed it. What do we need to do now?” Corrections — EPCRS Plan documents & elections Qualified plan testing Distributions & 402(g) THE OTHER TWENTY-TWO STAY OUT OF THE WAY plan document · adoption agreement · SPD · election form YOU LOAD THESE. WE NEVER SEE THEM. The correction method, with the testing knock-on raised, against your own election procedure. ⚠ JUDGMENT CALL A DRAFT FOR YOUR COUNSEL
Five SKILLs

Qualified Retirement Plan Compliance

  • Testing & eligibility
  • Plan documents, amendments & terminations
  • Distributions, loans, rollovers & RMDs
  • Pension & cash balance plans
  • 403(b) plans
Four SKILLs

Health & Welfare Benefits Compliance

  • COBRA, HIPAA & welfare plans
  • ACA employer mandate & reporting
  • Parity, transparency & the CAA
  • Fringe benefits & payroll taxation
Three SKILLs

Fiduciary Governance & Plan Administration

  • Fiduciary duties & prohibited transactions
  • Benefits cybersecurity & data incidents
  • Reporting, disclosure & claims
Three SKILLs

Corrections — When Something Goes Wrong

  • Plan corrections
  • Government audits & investigations
  • Litigation risk & preemption
Two SKILLs

Executive & Equity Compensation

  • Nonqualified deferred compensation & 409A
  • Equity & incentive compensation
Four SKILLs

Corporate Transactions & Workforce Changes

  • Employee benefits in M&A
  • Severance, WARN & reductions in force
  • Controlled groups & affiliated service groups
  • Multiemployer plans & withdrawal liability
Three SKILLs

Specialized Employers & Plan Structures

  • Government, church & tribal plans
  • ESOPs & employee ownership
  • Pooled, multiple-employer & small-employer plans
Two SKILLs

Life Events & Employee Leave

  • Leave, military service & wellness programs
  • QDROs & dividing retirement benefits
The getting-started skill

The answer to “I’m one person — I can’t set this up.”

A one-person benefits function can set this up. Setup is a product feature rather than your problem. One of the twenty-eight SKILLs does nothing but teach you to use the other twenty-seven. But if you prefer to get right to using the Tools, consider a subscription to the Research Assistant instead.

01

Install the files about 15 minutes

A printed sheet ships with the package. Upload the SKILLs to Claude and switch them on. No developer tools and nothing to configure. You might not use all twenty-eight, but add all of them so that they are there and will be used by the AI as questions come up.

02

Ask the skill to set you up guided

Tell it to teach you how to use it. It assumes you are starting cold — no Project, no documents, no prior conversation — and walks you through creating a Project, naming it, and writing the instructions that tell it how to treat your plans.

03

Load your own plan documents your grounding layer

It names the specific documents to start with: plan document and adoption agreement, current SPD and SMMs, trust agreement, administrative forms, service agreements, the last two Form 5500 filings. This is the step that makes answers about your plan rather than a plan in general — and you keep control of what goes in.

04

Confirm the documents are actually being read verification

A step people might inadvertently skip, so the skill insists on it. It gives you a test question and tells you how to read the answer: a good one names the file and the section it came from. A general description means your documents are not being reached, and it tells you what to check.

05

Learn to ask well six habits

Six named prompting habits, each with a worked before-and-after example — leading with context, pointing at the documents instead of letting the tool assume, asking for an issue-spotting pass before a finished memo, working iteratively, correcting bluntly, and treating every output as a draft for your counsel. Including the one question that catches more errors than any other: what did you assume that I did not tell you?

What you need before you start

A Claude account on any plan, with code execution enabled. Projects, for the plan-document workflow — though you can ask useful questions in a plain chat on day one and set up a Project later. An organization-wide install path is available on Team and Enterprise Claude plans at the Enterprise and Professional tiers. The Suite is built for Claude and is not currently packaged for other platforms.

Every single answer

It tells you not to trust it.

This is the whole posture of the product, and it is not a disclaimer buried in a license. It is stamped on the bottom of every response the Suite produces, it cannot be turned off, and it does not get shorter if you ask nicely.

On every response

⚠ AI-generated output. Not legal advice.
This response has not been reviewed by an attorney and cannot be relied on.
Confirm it with competent ERISA counsel before acting.
ERISA SKILLs Suite · release and license ID

And when a question turns on judgment

⚠ Judgment call — do not act on this.
This question turns on legal judgment, not just plan terms or a figure that can be looked up. Unreviewed output is especially unsafe to rely on here. Have competent ERISA counsel review this analysis before acting, communicating a position, or committing the plan.
License and pricing

Three tiers. Buy once, update when you want to.

Each tier is a perpetual license to the release you buy. It does not expire, lapse, or stop working, and it is not a subscription. ERISA guidance keeps moving, so updated releases are sold separately at half the then-current license price for your tier — a rate that never expires, however long you wait — and each release states the date its law is current to, so you can decide when to update. An older copy still works; it is simply answering from older law, which is why it tells you its release date when currency matters.

Tier one

Small Employer

One employer, one licensed user. Built to support a small employer’s benefits function.

$3,500

$1,750 per update release

Tier two

Enterprise

One employer, unlimited users inside that employer. For a multi-person benefits department, and where in-house counsel will use it alongside the benefits team.

$7,500

$3,750 per update release

Tier three

Professional

One organization that provides benefits guidance to others — a TPA, advisory firm, or broker — for use in serving its own customers. It does not permit handing the files themselves to those customers.

$12,500

$6,250 per update release

Your account, your data, your settings

You install the Suite into your own Claude account, under your own agreement with the platform. We do not host it, and we do not receive, store, or have any access to the documents, questions, or output you put into or get out of that account. We also do not set or control any setting in your account, including whether your material may be used to train or improve a model, how long it is retained, whether it is logged, and whether it is subject to human review. Those are yours to elect, and your agreement with the platform governs them. We will gladly tell you which settings we think are worth considering — that is guidance, not a warranty of anyone else’s conduct. Read your platform’s terms, and check your own organization’s data-handling and vendor policies, before you load plan documents or participant information.

Federal law is the subject. State examples are Arizona.

The Suite addresses ERISA and other federal law. Where it discusses state law, its examples are generally Arizona-based and illustrative only — included to show the kind of state law that bears on a question, not to state the law that governs you. It will not reason from an Arizona rule by analogy to another state’s. If your question turns on the law of a state, it is outside what the Suite can tell you and belongs with counsel licensed there. This is a term of the license, not a footnote.

What the license does not allow

Redistributing, reselling, sublicensing, publishing, or posting the files. Use beyond the scope of your tier, or sharing with another organization. Removing or suppressing the AI-output labels, the identification block, or the response footer. Building a derivative product for distribution from the content. Using the content to train or fine-tune any AI model. Each copy is individually identified: the license ID appears in the footer of responses your copy produces, and the copy carries further identifying characteristics that are not visible on inspection.

Where this came from

Built by people who practice ERISA law.

This is not a general AI consultancy's read of the benefits field. The content was written by practicing ERISA attorneys for use in their own work, and it exists because they needed it. The reference summaries are original syntheses of primary authority. The worked examples — coverage testing, ADP correction, cross-testing rate groups, top-heavy, 415 aggregation — are the calculations a practitioner actually runs.

What that means, and what it does not: it describes who wrote the content. It does not make the product legal advice, and it does not make anyone your lawyer.

ERISA Benefits Tech, LLC is a technology company, owned by the professional entities of the two partners of an Arizona law firm that practices ERISA and employee benefits law. It is not itself a law firm: it does not practice law, and it does not represent you. Buying or using the Suite creates no attorney-client relationship with anyone. Output is unreviewed, is not legal advice, and must be reviewed by competent ERISA counsel of your own choosing before you act on it.

If you outgrow it

Beyond the Suite: a custom build.

A few months of real use of the Suite will tell you more about where AI helps your function, and where it does not, than any proposal can. Some teams then want more than a packaged library provides.

A custom build

Where the Suite ends, a custom build begins: SKILLs grounded in your own plan documents by us rather than by you, your house positions and interpretations encoded into how each SKILL reasons, and — for employers — escalation criteria configured to route judgment calls to counsel, with ERISA legal review provided by ERISA Benefits Law Group, PLLC under a separate engagement letter.

How you would get there

It starts with a fixed-fee readiness assessment: an inventory of your plans, documents, workflows, and current AI posture, and a written roadmap you keep whatever you decide next. Every custom build is scoped individually. Pricing for the assessment and for a build is quoted to scope — start a conversation and we will walk you through it.

The SKILLs Suite is on sale.

Choose your tier at checkout, where you also accept the Purchase and License Agreement. You install it into your own Claude account, and the license is perpetual for the Release you buy. The training is free whether or not you buy anything.

Buy the SKILLs Suite → Consider a subscription to the Research Assistant and Desk Reference → Log in to access the free content →