These are classes on building, buying, evaluating, and funding AI for legal help. Each one is built around a decision somebody actually has to make, and each one ends with that decision written down.
The Stanford Legal Design Lab runs these classes as part of the Legal Help Commons. They cover building, buying, evaluating, and funding AI for legal help, and each one is written for a particular person facing a particular decision.
We run them because this field builds better when it builds together. The aim is a more coordinated ecosystem, one that can produce public interest technology that is sustainable and that actually helps people.
Three formats
Shorter modules
Do a short module focused on one deliverable or decision. Learn and practice on this focused topic, all at your own pace.
Class tracks
You work through a set of modules at your own pace, and they build toward a written plan for a real project of yours.
In-person cohorts
You join a cohort that meets in a room together over several months, and you come as a team from one organization rather than as an individual.
We are building these courses out now, and looking for testers. Please let us know which courses you'd be interested in. And let us know if you'd be willing to be a test user.
Not open yet
Enter the catalog
We are sharing this with a small group while we build the classes out. If someone sent you a password, put it in below to see the full class list and tell us which ones you want.
Choose the description closest to your own situation and the page will show you the classes written for it.
Frontline advocates and paralegals
Using AI in your casework
You go from using these tools unsupervised to checking every answer against its source, and knowing what never goes into the box.
What these classes cover
These classes cover the everyday tasks people already use AI for, including summarizing a document, explaining a notice in plain language, and pulling a deadline out of a letter. They also cover the checks that catch a wrong answer, and the rules about what should never be typed into a tool in the first place.
What you will be able to do
You will be able to use AI on your own matters without being burned by it, and to spot a wrong answer before a client acts on it.
You go from being pitched at to making the call in writing, and holding a supplier to a standard you wrote yourself.
What these classes cover
These classes cover the build, buy, and build-together decision, vendor evaluation, contract lock-in, organizational AI policy, and how to set a quality standard a supplier has to meet. It works from your own staffing, content, and budget rather than from a general framework.
What you will be able to do
You will be able to defend a build or buy decision to a board, judge a vendor on evidence rather than a demonstration, and set a five-year direction for your organization.
You go from funding pilots that end when the grant ends to using conditions, procurement, and convening power to build something that outlasts them.
What these classes cover
These classes cover how a market for legal help technology actually functions, and which instruments move it. It works through grant conditions, requests for proposals, pooled multi-state commitments, prizes, cohorts, and the reporting requirements that decide what the field can later know.
What you will be able to do
You will be able to diagnose why a market is failing before choosing an instrument, and write grant conditions that change what gets built.
You go from a role somebody handed you informally to running studies that hold up, and publishing an instrument other people can cite.
What these classes cover
These classes cover how much evaluation a given decision requires, whether a reported number can be trusted, how to build a test set that represents the people a tool will actually meet, and how to design a study before the money is committed. It assumes deep knowledge of legal services and no quantitative background.
What you will be able to do
You will be able to commission the right kind of evaluation, tell evidence from decoration in a vendor's claim, and know when a study needs a specialist rather than your own team.
You go from a blank page to a working tool that another organization could pick up, built on patterns that already work rather than from scratch.
What these classes cover
These classes cover how to choose a problem bounded enough to build well, how to prepare content so a system can retrieve from it, how to stand up a working prototype, and how to stop a model inventing citations. It also covers what makes a tool portable enough that a second organization can adopt it.
What you will be able to do
You will be able to judge whether your idea is buildable, ship something grounded and evaluated, and avoid the failures that make a tool unusable outside the place that built it.
You go from knowing your court needs to work better to having a route you can take, whether that means building it, partnering for it, copying it from another court, or buying it.
What these classes cover
These classes cover the routes available to a court that wants better technology, including building it, partnering with another organization, replicating something a peer court already runs, and buying it. It covers what your case management system exposes, how to open court data responsibly, and how to write a purchasing requirement a supplier has to meet.
What you will be able to do
You will be able to use the authority you hold inside the building, which means changing a workflow, opening a data feed responsibly, partnering, replicating, or writing a requirement a supplier has to satisfy.
You go from a claim you make about yourself to a result a buyer can check, at a point when the standards are still being written and you can still shape them.
What these classes cover
These classes cover what evidence a public buyer will actually accept, how a conformance test differs from a demonstration, and what publishing a reference architecture does for a company selling into this field. It is about the commercial value of being checkable at a point when the standards are still being written.
What you will be able to do
You will be able to show a buyer evidence rather than a demonstration, and to help set the bar your competitors will have to clear.
Each track runs across a set of modules, taken at whatever pace you set. Every module asks you to make a decision about a real project of your own, and the modules build toward a written plan rather than a certificate. Start here if you want the full treatment of a subject.
Who this is forThis is for the people who hold money and authority in a state, meaning IOLTA and access to justice program directors, presiding judges and court administrators, and foundation trustees. It is for anyone who has funded a promising pilot and watched it disappear when the grant ended.
What it coversThe six modules cover what a functioning market requires, how to set an expansion gate a pilot must clear before it scales, how requirements shape what suppliers actually build, how committed multi-state demand changes what is worth building, and what prizes and cohorts genuinely produce. The material is built from five evaluated public programs, including the health information technology certification record and advance market commitments in global health.
Before and afterYou arrive able to fund individual projects well and unable to change the pattern they sit in. You leave able to name which market function you control and which instrument moves it.
What you leave withYou leave with a written ninety-day plan naming three actions, their dates, and the person responsible for each.
Who this is forThis is for the person in an organization who has to say, credibly and in public, whether an AI system works and whether it is safe. It assumes deep knowledge of legal services and no quantitative training, which is the combination most people carrying this responsibility actually have.
What it coversThe five modules cover how much evaluation a given decision requires, whether a reported number can be trusted, whether a test set represents the people a tool will meet, whether a study can answer the question before you pay for it, and how to run evaluation as a standing program rather than a series of one-off projects. Four laboratories build working artifacts alongside the modules, including a deterministic verifier.
Before and afterYou arrive able to sense that a quality claim is thin without being able to say why. You leave able to specify what evidence a decision requires, run the studies that do not need a specialist, and recognize the ones that do.
What you leave withYou leave with a staged evaluation plan for a real project of yours and a working deterministic verifier.
Who this is forThis is for a builder or intrapreneur with an idea and no evidence, before any money or engineering time has been committed. It is meant to be the class that comes before you build anything.
What it coversYou state what would have to be true for your idea to work, then find the cheapest way to test each condition before committing money or engineering time. The class also covers how far a demonstration sits from a product, and what evidence a buyer or a board will actually accept.
Before and afterYou arrive with conviction and no test of it. You leave with your assumptions written down, a way to check each one, and an honest view of the distance still to cover.
What you leave withYou leave with a tested assumption list and a plan for closing the gap between a demonstration and a product.
Who this is forThis is for designers and researchers working on tools that people in legal trouble will use, often while distressed and under time pressure. It is for teams who assume their problems are model problems.
What it coversYou work through the interface failures found by both published audits of legal help chatbots, none of which had anything to do with the model and all of which came from design decisions. You then apply the same review to your own screens.
Before and afterYou arrive attributing failures to the model. You leave able to name the design decisions that actually cause them, and to catch them in your own work.
What you leave withYou leave with a reviewed set of your own screens and a checklist of the failures to watch for.
Who this is forThis is for content stewards and law librarians responsible for keeping legal information correct as statutes, forms, and court rules change underneath it. Every audit so far has found that content quality determines system quality, which makes this role more consequential than it is usually treated as.
What it coversYou set ownership, refresh cadence, and provenance for your own content, and work out the real maintenance burden of keeping a knowledge base current once systems depend on it. The class covers what breaks when a form is revised or a rule changes and nobody notices.
Before and afterYou arrive maintaining content by attention and memory. You leave with named owners, refresh targets, and a visible signal when something has gone stale.
What you leave withYou leave with a maintenance plan for your own content, including owners, cadences, and staleness rules.
Who this is forThis is for executives and board members who sign contracts for systems they cannot inspect. It suits organizations moving from experiments to purchases.
What it coversYou read contract language clause by clause, work through cost models across the life of an agreement, and assemble the questions to ask before signing. The vocabulary matches what public procurement offices already use, so what you write can go straight into a solicitation.
Before and afterYou arrive negotiating on price and features. You leave negotiating on ownership, access, evidence, and exit.
What you leave withYou leave with a marked-up contract and a written list of terms you will not sign without.
Who this is forThis is for a legal aid or court leader deciding how their organization will handle intake by phone, who has to choose between building a voice system, buying one, and replicating what another organization already runs. It suits the point before you commit to a vendor or to an engineering plan.
What it coversYou work the build, buy, and replicate options against a working reference architecture rather than against a pitch. You then set the conformance tests a voice system has to pass before it takes a real call, and work out how to evaluate a system that talks and listens rather than one that types.
Before and afterYou arrive able to tell that voice intake is worth trying and unable to say what a good one would actually do. You leave with a written specification and a test that a vendor or your own team has to pass.
What you leave withYou leave with a route decision, a conformance checklist for voice intake, and an evaluation plan for whichever system you choose.
Who this is forThis is for an executive director or operations lead who wants staff to be able to experiment with AI on real work, inside something the organization controls. It suits an organization whose staff are already using consumer AI tools that nobody can see.
What it coversYou work through what it takes to stand up a shared internal platform, following the LibreChat setup that Maryland Legal Aid built and runs. The class covers hosting choices, what to connect the platform to, how to handle client confidentiality, and how to get staff using it rather than ignoring it.
Before and afterYou arrive with staff experimenting in tools you cannot see and cannot govern. You leave with a plan for a platform you host, with the access rules and the rollout written down.
What you leave withYou leave with a deployment plan for an internal AI platform, covering hosting, access rules, and how you will roll it out to staff.
We do not have a track for this group yet. Tell us what you would need to know.
The catalog
Classes by what you are trying to do
These are shorter modules, and each one ends with a single finished document. None of them depends on any of the others, so you can take one on its own or take several in whatever order suits you.
Shorter modules
Design and build
These are for people making something, whether that means preparing content so a system can use it, standing up a working prototype, or making a tool safe enough to put in front of the public.
Who this is forThis is for someone at the very start of a build, either inside an organization or as their own venture, who has an idea and has not yet tested whether it is the right shape. Take it before you write any code.
What it coversYou work out whether your task is bounded enough to build well, meaning whether it has a right answer a lawyer could confirm. You then study how comparable public interest technology projects formed, funded themselves, and survived, and decide how you will work with what already exists rather than beside it.
Before and afterYou arrive with an idea and an instinct. You leave with a tested judgement about whether the idea is buildable and a written commitment about how you will build it.
What you leave withYou leave with a completed workbook you can take to your team and to any partner organization.
Who this is forThis is for a builder with a working tool that answers questions on legal content, who cannot currently guarantee that a warning appears when it needs to. It matters most if any of your content touches deadlines, service of process, or immigration consequences.
What it coversYou sort your content by risk tier, write display rules for the high-stakes material, and set links that force a warning to travel with every chunk drawn from a risky page. You then test the rules against scenarios that ought to trigger a warning.
Before and afterYou arrive relying on the model to remember the warning, which it will not always do. You leave with deterministic rules that produce the warning every time regardless of what the model does.
What you leave withYou leave with a working set of display rules and must-show-together links covering your high-stakes topics.
Who this is forThis is for a content lead or developer at an organization with a library of legal help articles, facing the question of how to make that content usable by an AI system. It suits the point where you have decided to build and do not know where to start with your own material.
What it coversYou take a single article of your own and put it through the full process, which means tagging it with jurisdiction, legal issue, content type, and authority tier, then splitting it the way its content type requires, then attaching the warnings that must travel with any answer drawn from it.
Before and afterYou arrive with a library of pages written for human readers. You leave with one article genuinely prepared for machine use and a repeatable method for the rest.
What you leave withYou leave with one finished article and a checklist you can run on the next hundred.
Who this is forThis is for someone who has to make decisions about a system of this kind, either building one or judging someone else's, and has never assembled one. You do not need to be a developer, and you do need to be willing to be frustrated for an afternoon.
What it coversYou load a small set of content, prepare it for retrieval, and put a retrieval layer in front of a model so that it answers from your material rather than from its training. You wire in the patterns that separate a working system from a demonstration, including searching small passages while handing the model the whole section, retrieving broadly then reranking, and filtering by jurisdiction before searching at all.
Before and afterYou arrive treating these systems as something close to magic. You leave having built a rough one yourself, which is what gives you the standing to judge anyone else's.
What you leave withYou leave with a working prototype and a concrete understanding of what these systems are made of.
Who this is forThis is for a builder whose tool cites legal authority to members of the public or to a court. A single fabricated citation is the fastest way to lose a judge, and this class exists because it keeps happening.
What it coversYou change the system so that quotations come from retrieved text and links come from your own stored metadata rather than being generated by the model. You then add a verifier that checks every citation before the answer ships.
Before and afterYou arrive with a system that can generate a plausible citation to a case that does not exist. You leave with one that can only cite material you actually hold.
What you leave withYou leave with a constrained citation setup and a working verifier that refuses anything it cannot confirm.
Who this is forThis is for someone routing incoming requests at volume, whose organization has its own category list that no off-the-shelf classifier knows about. It suits an intake or triage project that is drowning in manual sorting.
What it coversYou run ensemble classification, meaning several models vote and their votes are weighted, against your own categories rather than a generic taxonomy. You then map your categories onto the shared legal issue taxonomy so that your results mean something outside your own organization.
Before and afterYou arrive sorting requests by hand or with a classifier that does not know your categories. You leave with one trained on the categories you actually use.
What you leave withYou leave with a working classifier mapped to the shared legal issue taxonomy.
Who this is forThis is for anyone about to put real case documents through an AI system, who has not yet worked out how to remove the personal information first. Take it before the documents go anywhere, not afterwards.
What it coversYou apply four risk tiers and a five-part assessment to your own documents, and see which categories of personal information the automated masking catches and which it misses.
Before and afterYou arrive intending to be careful. You leave with a tested process and a documented view of what it does not catch.
What you leave withYou leave with a masking setup you can run across a real collection of documents.
Nothing in this group is written for you yet. Tell us what you would want here.
Shorter modules
Evaluate
These are for finding out whether a tool actually works, and for judging the claims other people make about theirs.
Who this is forThis is for anyone who sits across the table from vendors and has to assess quality claims without a statistics background. It is the shortest thing in the catalog and the fastest way to find out whether the evaluation material is for you.
What it coversYou read eighteen quality claims about real legal AI systems, phrased the way vendors phrase them, and identify the specific evidentiary defect in each. The defects repeat, and by the end you recognize them.
Before and afterYou arrive able to tell that a claim feels thin without being able to say why. You leave able to name the defect out loud in a meeting.
What you leave withYou leave able to identify the four recurring ways a quality claim falls apart.
Who this is forThis is for someone responsible for a real project who has been asked how it will be evaluated and does not yet have an answer. It works whether the project is a chatbot, a document reviewer, a drafting tool, a guided interview, or a procedural coach.
What it coversYou answer twelve questions about your project and receive a staged plan covering which studies to run, in what order, with sample size arithmetic and the staffing each stage requires. The plan also states plainly what it will not be able to establish.
Before and afterYou arrive with an intention to evaluate and no design. You leave with a staged plan you can put in front of a funder or a board.
What you leave withYou leave with a written evaluation plan sized to your project, budget, and stage.
Who this is forThis is for an advocate or paralegal already using AI at work, who has not established a habit for checking its output. It is the first class to take if you are using these tools and nobody has trained you.
What it coversYou run three everyday tasks on a real document, meaning you summarize it, explain it in plain language, and pull the deadline out of it. You then check every line against its source, using a method that asks whether each fact traces back to the document and each rule traces to a real legal authority.
Before and afterYou arrive trusting an answer because it sounds confident. You leave knowing that a fluent answer and a correct answer look identical from the outside, with a habit for telling them apart.
What you leave withYou leave with a working source-checking routine and a clear view of where the tool will fail you.
Who this is forThis is for someone who has built or commissioned a tool and has been asked whether it works. It is for the point where the honest answer is that nobody has checked properly.
What it coversYou write a rubric shaped to the specific task, assemble a set of genuinely difficult questions including edge cases, and run one honest evaluation that reports where the tool falls over. You also learn why a model asked to grade another model passes nearly everything, and where a deterministic check earns its place instead.
Before and afterYou arrive able to say only that the tool seems to work. You leave able to answer the question with a number and a list of specific failures.
What you leave withYou leave with a task-shaped rubric, a set of hard test questions, and the results of one honest run.
Nothing in this group is written for you yet. Tell us what you would want here.
Shorter modules
Plan and strategize
These are for setting direction, whether for one organization or for a whole state, and for writing the policies and standards that direction depends on.
Who this is forThis is for a program officer, IOLTA director, or trustee who is receiving more AI proposals than they can assess, and who has no written position on what is worth backing. It suits the point in a funding cycle before the next round opens.
What it coversYou weigh durable shared infrastructure against one-off tools, and work out where a dollar builds something that outlasts the grant. The session draws on the field's assessment of what national leadership is missing and on survey data about building versus buying.
Before and afterYou arrive assessing each proposal on its own merits with no stated position. You leave with a written thesis that tells applicants in advance what you will and will not fund.
What you leave withYou leave with a one-page funding thesis naming where your next dollar goes and why.
Who this is forThis is for someone who convenes several organizations, whether as a funder, a state coordinator, or a network lead, and who can see the same content being prepared separately in each place. It is for the point before those projects have gone too far in different directions.
What it coversYou agree the smallest set of fields every organization would tag, covering jurisdiction, legal issue, language, last updated, licence, and provenance. You also set the validation rule that no deadline or procedure publishes without a supporting legal authority attached.
Before and afterYou arrive with a group of organizations each preparing content their own way. You leave with a written agreement small enough that they will actually sign it.
What you leave withYou leave with a minimal data contract for your network and a plan for the first organizations to adopt it.
Who this is forThis is for anyone about to convene a group of organizations around a technology project, who has seen cohorts produce good conversation and no shipped software. It suits the planning stage, before invitations go out.
What it coversYou choose between a cohort that generates frameworks and prepared data and one that moves a working tool from one organization to another on a faster cadence. You then set the roles, the deliverables, and the shipping rhythm that follow from that choice.
Before and afterYou arrive with an intention to convene people and no structure for it. You leave with a cohort design specific enough to open applications on.
What you leave withYou leave with a written cohort plan covering shape, roles, deliverables, and cadence.
Who this is forThis is for anyone handling client matters who has never been told what is and is not allowed to go into an AI tool. It is also for the supervisor who suspects their staff are already pasting things they should not.
What it coversYou draft your own list of what never goes into the box, learn the two or three signals of a fabricated legal rule, and decide where your own line sits for matters too sensitive to run through a tool at all.
Before and afterYou arrive without a rule, which in practice means the rule is whatever each person decides in the moment. You leave with a written line you can hold yourself and your colleagues to.
What you leave withYou leave with a written list of what stays out of the tool, specific to the matters you actually handle.
Who this is forThis is for a director whose staff are already using AI tools with no written guidance, which is most directors. It suits the point where you have realised you do not actually know what people are pasting into these tools.
What it coversYou draft the allowed uses, the prohibited uses, the categories of data that may and may not enter a tool, who approves the higher-risk uses, and the human review requirement for anything that reaches a client. The material is built from risk tiers already tested on 150 real scenarios.
Before and afterYou arrive with an unwritten rule that varies by person. You leave with a written policy that a board can adopt and staff can follow.
What you leave withYou leave with a draft AI use policy sized to your organization.
Nothing in this group is written for you yet. Tell us what you would want here.
Shorter modules
Decide to buy or build
These are for the decision about where a tool comes from, and for the contract review that follows it.
Who this is forThis is for an executive director or program lead with a proposal in front of them, who has to say whether the organization will build its own tool, buy one, or join other organizations building together. It is written for the moment before a board or a funder asks you to justify the choice.
What it coversYou work the three options against your own staffing, content readiness, timeline, and appetite for risk. You also see how 74 other justice professionals answered the same question, including that 70 percent named privacy as their main concern about vendors and 66 percent named staffing as the barrier to building their own.
Before and afterYou arrive with the question either unresolved or answered on instinct. You leave having made the decision in writing, with reasoning that survives a hard question from a board member.
What you leave withYou leave with a one-page written decision that records the option you chose and the tradeoffs behind it.
Who this is forThis is for a director, funder, or court administrator with an AI plan underway or a contract about to be signed, who has not yet worked out how difficult it would be to leave. The best time to take it is before you sign, while changes are still cheap.
What it coversYou run your current plan through four questions covering who owns the content, who owns the data, whether your knowledge would be trapped in a proprietary index, and what switching suppliers would actually cost you. You then mark each place where you are exposed.
Before and afterYou arrive assuming your plan is reversible. You leave with a specific list of the places it is not, and an idea of what each one would cost to fix now rather than later.
What you leave withYou leave with a marked-up plan showing every point of exposure and what it would take to close it.
Who this is forThis is for whoever will sit through the demonstration and then be asked to approve the purchase. It suits the weeks between a shortlist and a signature.
What it coversYou turn the evaluation method on a product rather than a build, and work out which evidence to demand before agreeing to anything. The session covers the difference between a demonstration and a test, and why a system can pass a thousand casual checks and still fail the one query that matters.
Before and afterYou arrive assessing suppliers on the quality of their presentation. You leave assessing them on evidence, with the specific questions a sales team hopes you will not ask.
What you leave withYou leave with a vendor scorecard and a written list of the evidence you will require before signing.
Nothing in this group is written for you yet. Tell us what you would want here.
We do not have a shorter module for this group yet. Tell us what you would want.
A group over several months
In-person cohorts
These run over several months and include time in a room together. You join as a team from one organization rather than as an individual, and you work on your own jurisdiction's constraints throughout.
Who this is forThis is for the people who decide what gets built, funded, and trusted in housing legal help, including legal aid leaders, self-help centre directors, court innovation staff, funders, and the technologists and researchers working alongside them. No technical background is required, and you bring your own jurisdiction's caseloads, staffing, and budget.
What it coversThe cohort runs an on-ramp, four modules, and a capstone, covering how to set a build agenda, how to write a quality standard using evidence tiers, how to commission work without crowning a supplier, and how to set an evaluation bar a tool must clear before it reaches a tenant. A working application anchors the cohort by walking one real eviction case stage by stage.
Before and afterYou arrive with instincts about what should be built and no shared language for arguing about it. You leave having set the agenda, the standard, the specification, and the quality bar for your own jurisdiction.
What you leave withYou leave with a build agenda, a written standard, a build specification, an evaluation bar, and a plan sized to your own organization.
Designing the Future of Debt Collection Legal Help
Who this is forThis is the same cohort as the housing class, run for the people who decide what gets built in debt collection work. It suits legal aid leaders, court staff, and funders dealing with collection calls, default judgments, and garnishment.
What it coversThe structure is identical to the housing cohort, and the domain content covers being called by a collector, being sued on a debt, and what happens when a default judgment turns into garnishment. The same four artifacts are produced across the modules.
Before and afterYou arrive with a debt practice and no shared standard for the tools being proposed to you. You leave with an agenda, a standard, a specification, and a quality bar for your own jurisdiction.
What you leave withYou leave with a build agenda, a written standard, a build specification, an evaluation bar, and a plan sized to your own organization.
Who this is forThis is the same cohort run for people working on reentry, meaning clearing, sealing, or reducing an old record so that it stops blocking work and housing. It suits legal aid leaders, court staff, and funders working on record clearing at any scale.
What it coversThe structure is identical to the housing cohort, and the domain content covers eligibility for record clearing, assembling court and criminal records, and the collateral consequences that persist afterwards. The same four artifacts are produced across the modules.
Before and afterYou arrive with a record clearing practice and no shared standard for the tools being proposed to you. You leave with an agenda, a standard, a specification, and a quality bar for your own jurisdiction.
What you leave withYou leave with a build agenda, a written standard, a build specification, an evaluation bar, and a plan sized to your own organization.
Who this is forThis is for a court team of two people from the same court, meaning one judicial officer and one information technology director. Individuals are not admitted, because a judge without an IT director leaves with intentions and an IT director without a judge leaves with a document nobody will sign.
What it coversThe cohort covers what your case management system actually exposes, what your existing contract already obliges your supplier to provide, how the market is structured and where each player makes money, how to turn a stated problem into purchasable requirement language, and how to specify an acceptance test a supplier must pass before signing. A full module covers what to do when the answer is no and a modern interface is not coming this budget cycle.
Before and afterYou arrive unable to say what your own system exposes or when your contract ends. You leave with requirement language your purchasing office will accept and a test written before any supplier sees it.
What you leave withYou leave with a capability sheet, a marked-up contract review, requirement language, an acceptance test list, and a fallback plan with a date attached.
We do not run a cohort for this group yet. Tell us who you would want in the room.
Register your interest
Tell us which of these you want
We are building these out now, and what people tell us they need decides the order. Let us know who you are and what decision you are facing, and we will come back to you as the class closest to your situation is ready.