Prompt Pilotby Write.law
All prompts

Strategy

Decide what this matter should cost and who does it

Phases the matter with a decision point at the end of each, assigns every task to the cheapest competent level, exposes the assumptions driving the number, and names the two or three decisions that control it.

About 20 minintermediateIn-house, Solo / small firm, Litigation

Your prompt6,040 characters

Still to fill in: The matter, Forum and schedule, Who you actually have

RoleYou are a lawyer who has watched a matter budget become a forecast nobody could act on, because it had a total and no decision points. You phase work so a client can stop between phases, you assign every task to the cheapest level that can competently do it, and you put the assumptions in front of the number rather than behind it. You will say out loud which tasks should not be done by a lawyer at all.What I needBuild the phased budget and staffing plan for the matter below in Forum and schedule, using what I actually have.InputsThe matter: The matter Forum and schedule: Forum and schedule Who I have: Who you actually have What the client is buying: What the client is buying Money, and what they have already been told: Money and any number already givenHow to work this1. Break The matter into phases, each ending at a real decision point: a ruling, a disclosure, a deadline, or a number becoming knowable. Say what decision the client makes at the end of each phase and what information they will have that they do not have now. A phase that ends because time passed is not a phase. 2. Within each phase, list the tasks and assign each to the cheapest level in Who you actually have that can competently do it: partner judgment, associate drafting, paralegal or project assistant, a vendor, software, or the client's own team. Name explicitly the tasks that should not be done by a lawyer at all, because that list is where most of the savings live and nobody writes it down. 3. Put the assumptions in front of the number, not behind it. For each phase, name the three or four assumptions that drive its cost: how many depositions, whether a dispositive motion happens, whether the other side is cooperative, whether documents sit in one system or nine, whether an expert is needed. Then give a range for the phase tied to those assumptions, and say which single assumption moves the range most. 4. Name the two or three decisions that control the total more than everything else combined, and put them in front of the client as choices rather than burying them in an assumption list. These are usually about scope and posture rather than about efficiency. 5. Say what you do not know, and rank what would tighten the estimate. An honest phase marked "cannot be estimated until the scheduling order issues" is more useful than a confident number. 6. Do not produce hourly rates, blended rates, or a grand total unless Money and any number already given or Who you actually have supplies them. Work in hours, in ranges, and in phases. A total invented from assumed rates is the number that gets quoted back to you. 7. Give the stop-and-reassess triggers: the specific events that mean the plan is wrong and the budget has to be rebuilt rather than adjusted.Ask me firstYou have the matter, the forum, the team, and the constraints. Ask me these three, which none of that reveals, then stop and wait: 1. Is the client buying an outcome or buying predictability? A plan that caps the number and a plan that maximises the chance of winning are rarely the same plan, and I need to know which one to build. 2. Who inside the client sees this number, and does it have to survive a procurement or finance review? That changes the format and it changes how much of the range I can show. 3. What is the relationship worth beyond this matter? A plan that under-resources the first phase to hit a number is a different risk when this is the only matter than when it is the first of many. Do not begin until I answer. If I tell you to proceed anyway, state each assumption at the top of your output and mark it [ASSUMPTION - verify].Output formatA phase table: Phase | What happens | The decision at the end | Hours range by level | The assumption that moves it most. Then the do-not-use-a-lawyer list, by task and by who should do it instead. Then the assumption register, grouped by phase, each with its effect on the range. Then the two or three controlling decisions, written as choices for the client with the consequence of each. Then what cannot be estimated yet and what would fix that. Then the stop-and-reassess triggers. Close with one line naming the two of my answers that most changed this plan, and what the phasing would have looked like without them. If an answer changed nothing, say so, because it means I should not have been asked.Never do this- If the plan would fit any matter of this type anywhere, it is too generic. The phases have to come from what this forum's schedule actually does and the staffing from the people I said I have. - No hedging filler. "Costs will vary depending on circumstances" is the premise of the exercise, not a finding. Do not tell me to consult an attorney; I am the attorney building this. - Never invent an hourly rate, a blended rate, a total, or a market benchmark for what this kind of matter costs. If a number appears that did not come from my inputs, it is fabricated and it will be quoted back to me by a finance team. - Never state what a rule or a local schedule requires in Forum and schedule. Where the calendar drives a phase, mark it [CONFIRM - local rule or standing order] and phase around the uncertainty. - Where you cannot estimate a phase, say you do not know and mark it unestimable. Do not smooth over the gap with a range wide enough to be meaningless. - Do not pad. If the honest answer is two phases and a decision point, say that. Length is not value.Before you answer- Does every phase end at a decision the client actually makes? - Did I produce any rate, total, or benchmark that was not in the inputs? - Is there a task on the do-not-use-a-lawyer list, or did I staff everything to lawyers by default? - Are the assumptions stated before the ranges rather than after? - Would this plan be useless for a different matter in a different forum? It should be.

The run walks turn one, the pressure test, the follow-up, and a check on what came back. The Cockpit adds driver's-seat tunes. Your values come with either one.

2

Pressure-test it

Makes the AI switch hats and attack its own answer.

Procurement sees this number without the memo attached to it. Read the budget as the finance analyst whose job is to find the soft phase: which range is wide enough to mean nobody actually estimated it, which phase has hours but no deliverable a non-lawyer could recognise, and where does the plan assume the client's own team does work the client has not agreed to do? Then tell me which phase I should rebuild before this goes anywhere near a review.
3

Go deeper

Pushes the work further once the basics are right.

Budgets get approved and then nobody looks again until the overrun. Write the monthly one-pager that keeps this honest: hours against the phase range, which assumptions in the register have since been resolved and which way, the decision point coming up next and what the client will need to decide it, and a single line stating whether the plan still holds or is being rebuilt.
4Check what came backPaste the answer here and work a checklist against this prompt's own rules.

Before you run it

What to gather first

  • What the matter is, its posture, and what has already happened
  • The forum and the schedule, since a court calendar drives most of the cost
  • Who you actually have: lawyers, paralegals, vendors, software, client staff
  • What the client is trying to achieve and what they can spend
  • Any number the client has already been given

Watch for

  • The model has no knowledge of what legal work costs and will invent rates, totals, and benchmarks. Any number it produces that you did not supply is fiction, and finance teams quote fiction back.
  • A budget with no decision points is a forecast. If the client cannot stop between phases, you have given them a number rather than a plan.
  • Under-resourcing the first phase to hit a target is the most common way a budget is beaten on paper and lost in practice. The early work is where the number is actually determined.
  • Court calendars drive most litigation cost and vary by forum and judge. Confirm the schedule before committing to phase boundaries.
  • Do not paste client-confidential matter facts or fee arrangements unless your firm's or company's AI policy permits it.

What comes back

A phase table (what happens, the decision at the end, hours range by level, the assumption that moves it most), a do-not-use-a-lawyer task list with the alternative owner, an assumption register grouped by phase with each one's effect on the range, the two or three controlling decisions written as client choices, what cannot be estimated yet and what would fix it, the stop-and-reassess triggers, and the answer-consumption line.

See an example of what you’ll get
*(After I answered: the client is buying predictability, the number goes to finance and has to survive a review, and this is the first of what should be several matters.)* Phases | Phase | What happens | The decision at the end | Hours by level | The assumption that moves it most | |---|---|---|---|---| | 1. Response and early assessment | Responsive pleading, early case assessment, payroll data pull | Fight, mediate early, or pick off the named plaintiffs | Partner 15 to 25; Associate 60 to 90; Paralegal 20; Client analyst 30 | Whether the payroll data is in one system. If it is in three, associate hours roughly double | | 2. Pre-certification discovery | Written discovery, named-plaintiff depositions, data analysis | Whether to oppose certification on the merits or to move to a class-wide resolution | Partner 40 to 70; Associate 200 to 340; Paralegal 80; Vendor review; Client analyst 40 | Number of named plaintiffs deposed, and whether the class definition is amended | | 3. Certification | Opposition, expert declaration, hearing | The only decision that matters: settle at the post-ruling number, or litigate a certified class | Partner 80 to 120; Associate 250 to 400; Expert | Whether an expert is needed on commonality. That single call is the difference between the low and high end of the whole phase | | 4. Post-certification | Not estimated | Not applicable | Not estimated | See below | Do not use a lawyer for these. | Task | Who should do it | |---|---| | Payroll and timekeeping data extraction | The client's HR analyst. You have one; use them. This is the largest single saving available | | First-pass document review | The review vendor, with an associate writing the protocol and sampling the output | | Deposition scheduling, exhibit assembly, transcript indexing | Paralegal | | Class-member contact data deduplication | Software, then the paralegal, then an associate spot-check | | Building the damages model's data layer | The client's analyst under an associate's specification. A lawyer should read the output, not build the spreadsheet | Assumption register - Phase 1. Payroll in one system (if three, associate range moves from 60 to 90 up to roughly 120 to 170). No early motion practice. Client analyst is genuinely available; if they are not, those 30 hours land on your paralegal at best and your associate at worst. - Phase 2. Two named plaintiffs deposed, not five. Other side is cooperative on ESI scope. No motion to compel. - Phase 3. One expert declaration. This is the assumption that drives the widest range in the plan. - Across all phases. The scheduling order lands roughly where this district usually puts it [CONFIRM - local rule and the judge's standing order; the entire phase structure sits on this and I cannot verify it]. The decisions that control the number. Three, and they matter more than every efficiency in this plan combined. 1. Whether you contest certification on the merits or work toward a class-wide resolution before the motion. Phase 3 is the largest phase and this decision either funds it or removes it. 2. Whether you retain an expert for certification. Nothing else in the plan has this range. 3. Whether the client's analyst does the data work. It is roughly 70 hours of work at the cheapest level in the plan or at one of the most expensive, and it is a client decision, not yours. What cannot be estimated yet. Phase 4. Post-certification cost depends on a ruling that does not exist, and a number for it here would be invented. Say that to finance in those words; a plan that marks one phase as unestimable is more credible in a review than one that estimates everything. The two things that would make phase 4 estimable are the certification ruling and the class size as actually certified. Stop-and-reassess triggers. A second suit filed anywhere on the same practice. An amendment adding a new claim type or a new state. The class definition expanding beyond the payroll population you can actually pull. A dispositive motion by either side. Any of these means rebuild, not adjust. Assumptions this plan rests on. That the 900 putative class members are all in one state [verify: a multistate class changes phase 2 entirely]. That the client's analyst has authority to spend the hours [verify: this is the assumption most likely to quietly fail]. What your answers changed. That the client is buying predictability is why phase 4 is marked unestimable rather than given a wide range, and why the three controlling decisions are pulled out as client choices instead of sitting in the assumption register. A client buying an outcome gets the opposite treatment: the register stays internal and the plan optimises phase 3. That this goes to a finance review is why the output is in hours by level rather than in a total, and it is the reason there is no dollar figure anywhere above. Your answer that this is the first of several matters changed one thing only, which is the note against under-resourcing phase 1, so it was the cheapest of the three questions.
Why this prompt is built the way it is
## Framework 1. **Phase the matter with a decision point at the end of each.** A budget without them is a forecast; with them, the client can stop. 2. **Assign every task to the cheapest competent level,** and name the tasks a lawyer should not be doing at all. 3. **State the assumptions separately from the number,** and give a range per phase tied to them rather than a point estimate. 4. **Name the two or three decisions that control the number** more than everything else combined, and put them in front of the client. 5. **Say what you do not know** and what would tighten the estimate. 6. **Produce no rates and no total** unless the inputs supply them. 7. **Give the stop-and-reassess triggers.**