The 4-Hour Pre-Suit Demand Package (AI-Assisted Workflow)
Verify it yourself — free, no login
See how AI medical-record review links every fact to the exact Bates page that proves it — click any citation and jump straight to the record.
See the 60-second demo →A complete pre-suit demand package — medical summary, liability narrative, itemized damages, exhibit index, cover letter — has historically been a 30–60 hour project spread across an associate, an LNC, and a partner. For most plaintiff firms, that turnaround time is what keeps demands from going out until months after records arrive, and what keeps many demand letters from going out at all on smaller cases that cannot justify the labor.
With AI assistance, the same package is a four-hour project for one attorney, with internal review extending that by another hour or two. This is a walkthrough of the actual hours, what gets done in each, and — just as important — what does not get delegated to the AI at any stage.
Preconditions
This workflow assumes you already have:
- The complete medical records you intend to cite — not a partial production
- Any available billing ledgers, itemized CMS-1500s or UB-04s, and insurance EOBs
- The client's wage documentation (pay stubs, W-2s, or tax returns as appropriate)
- A signed engagement letter and HIPAA authorization, because you are about to move fast
- Client intake notes capturing the narrative in the client's own words
If you do not have the records, you are not ready for this workflow. Drafting around missing records is how demand letters get made into cross-examination exhibits later.
Hour 1: Records intake and review
Load the records into the Records Analyzer. On a typical 500–3,000 page medical record this runs in minutes. What you get back is a structured index: provider-by-provider summaries, a chronological list of significant events, a medication timeline, an imaging-and-procedure list, and a first pass at a problem list.
What you do in this hour, personally:
- Read the summary and flag anything that contradicts the client intake narrative. Contradictions between the records and the client story are the single most important thing you can identify before drafting.
- Mark which providers are central to causation and which are peripheral.
- Note any gaps — missing office notes, referrals with no corresponding specialist visit, imaging reports without the underlying studies.
- Identify the two or three "moments" in the record that will carry the liability theory. A demand letter has a story. You decide the story in this hour.
What the AI does not do here: it does not identify which providers breached the standard of care. That is a clinical judgment that stays with you and, eventually, your expert. The AI indexes the record. You interpret it.
Hour 2: Timeline and liability narrative
Feed the flagged events from hour one into the Timeline Builder. What comes back is a visual and text timeline keyed to source pages, with every event citing the specific record it came from. This is the backbone of the demand letter.
What you do in this hour, personally:
- Edit the timeline to tell the liability story. Not every event belongs. A demand-letter timeline highlights the decisions and omissions that support your theory; the full timeline stays in your file for discovery.
- Draft the liability narrative in your own voice using the timeline as source material. The AI can produce a first-pass narrative, but plaintiff firms are consistently better served having the attorney write the story using the AI's structured events as paragraph-level building blocks.
- Confirm every cited page in your draft against the source. The AI gives you citations; you verify them. One miscited page in a demand letter trains defense counsel to discount the whole document.
What the AI does not do here: it does not make the call on whether the standard of care was breached, and it does not characterize a provider's motives. "The nurse ignored the bedside alarm" is an inference that requires judgment. "The bedside alarm fired at 03:14 per the telemetry log; the next documented nursing assessment was at 06:20" is a fact statement. The AI drafts the second kind. You decide whether the first kind is supportable.
Hour 3: Damages model
Run the Damages Calculator against the records, billing ledgers, and wage documentation. What you get is an itemized economic damages block (past medicals, future medicals, past wage loss, lost earning capacity with present-value calculation), a framing for non-economic damages tied to the specific medical picture, and a life care framework for any catastrophic or ongoing-care component.
What you do in this hour, personally:
- Sanity-check every line item against its source. Billing ledgers are messy; the AI will faithfully reproduce whatever is in them, including errors and duplicated charges. You catch those.
- Decide which damages categories you will plead in the demand and which you will reserve. Not every category belongs in every demand — sometimes you want to leave headroom for post-demand development.
- Set the non-economic damages framing. The AI can frame. You decide the framing. This is advocacy, not calculation.
- If this is a catastrophic case, flag the life-care-plan needs for an eventual retained life care planner. The AI produces a defensible framework; the planner refines it.
What the AI does not do here: it does not decide settlement authority. The demand number is a strategic choice that depends on jurisdiction, venue, defense counsel, insurance posture, and your read of the case. The calculator gives you a defensible working number. You decide the demand.
Hour 4: Demand draft and exhibit assembly
Use the Demand Letter tool to produce a first-draft cover letter and body, seeded with the timeline, liability narrative, and damages model from the prior hours. The output will be structurally complete: factual background, liability discussion, damages itemization, demand figure, exhibit list, response deadline, preservation language.
What you do in this hour, personally:
- Rewrite the opening paragraph. AI-generated openings read as AI-generated, and defense adjusters notice. Your opening signals that a real attorney is behind the document.
- Tighten the liability discussion. The AI tends toward over-hedging ("may have" and "potentially") where the record supports flat statements. Cut the hedges where the evidence warrants it, keep them where it does not.
- Confirm the demand figure, the response deadline, and the exhibit index. These are the three things defense counsel will cross-reference immediately and the three things that must be unambiguous.
- Assemble the exhibit package. Every medical record excerpt, wage document, photograph, and cited article goes in a labeled exhibit bundle with a one-page index.
What the AI does not do here: it does not sign the letter. It does not decide whether to extend a response deadline, accept a counteroffer, or move to suit. Those are all attorney calls, and they are calls made with context the AI does not have — statute of limitations, your pipeline, the adjuster's history with your firm, your client's life circumstances.
The internal-review hour (or two)
No demand letter should go out without a second set of eyes. Build an hour of partner or senior-associate review into the schedule. At this stage the reviewer's job is not to verify every citation — you did that at each stage — but to check the three things that determine whether the demand lands:
- Does the liability theory survive a cold reading? If the reviewer has to ask "wait, how did that cause the injury?" the theory needs another paragraph or another exhibit.
- Is the demand number defensible to a reserves committee? The adjuster who gets this letter has to justify the reserve internally. Your job is to give them the facts and numbers they need to do it. Your reviewer's job is to make sure you did.
- What will defense counsel cross-examine on? Every demand letter becomes a deposition exhibit eventually. Anything you would not want read back to a jury should not be in the letter.
What this workflow is not
This is not a trial package. It is not an expert report. It is not substitute work product for a retained economist, a retained LNC, or a retained clinical expert. What it is, is a fast, disciplined pre-suit artifact that (a) gets the defense the facts they need to open negotiation, (b) documents your theory of the case in a way that survives later scrutiny, and (c) frees up case-team time for the cases that are not yet at this stage.
A good demand letter is a sales document for the plaintiff's theory of liability and damages. It's not an expert opinion. Confusing the two either over-engineers the demand or under-prepares the trial.
What AI should never do in this workflow
- Set settlement authority. The number is a litigation-strategy decision. Authority comes from your client, informed by your analysis. The calculator informs the analysis; it does not set the authority.
- Judge doctor credibility. Whether a provider was careless, indifferent, or worse is a judgment the AI cannot make from records alone. That judgment requires the provider's deposition and your clinical expert.
- Make a final liability call. A demand can be strategic about what it asserts; a verdict cannot. Keep the AI out of the "who is at fault" conclusion and reserve that for after expert review.
- Draft the client communication. Every demand package sent should be accompanied by a plain-language letter to the client explaining what was sent and what happens next. Write that yourself.
Where the real savings come from
The four-hour workflow does not exist because the AI is "doing" the legal work. It exists because AI eliminates the two time-sinks that used to consume the demand process: manual extraction of a chronological record from thousands of chart pages, and manual itemization of damages from disorganized billing ledgers. Those two extractions alone historically took 20–40 hours on a serious case. Compressing them into minutes is the shift. The attorney work — theory, framing, judgment, advocacy — still takes the same hours it always did. It just stops being crowded out by the extraction work.
That shift is also why the workflow rewards experienced attorneys more than junior ones. The AI is very good at the extraction-and-indexing half of the job. The advocacy half — the part the AI cannot do — is where your time and judgment go. A four-hour demand package is four hours of attorney advocacy time, not four hours of AI-watching. Plaintiff firms that treat it otherwise produce worse demand letters, not better ones.
Related tools
Run the full workflow with the Records Analyzer for intake and medical indexing, the Timeline Builder for the liability chronology, and the Demand Letter tool to assemble the package itself.