From Case Intake Notes to Demand Themes: What Should Be Preserved, Verified, or Discarded

An intake note can contain the first useful clue in a personal-injury file—and the first unsupported assumption. By the time a demand is drafted, a sentence originally captured from a hurried phone call may have been copied into a case summary, repeated in an internal memo, and presented as if it came from a police report or medical record.

For plaintiff PI firms using AI-assisted drafting, the central question is not whether intake notes should enter the workflow. It is how to preserve their investigative value without letting unverified language harden into a carrier-facing fact.

Intake notes are leads, not a verified factual record

Intake teams work under practical constraints. They are screening liability, limitations concerns, treatment status, coverage, conflicts, and fit while trying to capture a caller’s account accurately. The resulting notes are operationally valuable, but they are usually a mixture of direct statements, paraphrase, staff shorthand, preliminary impressions, and unanswered questions.

That mixture creates risk when information moves downstream without its source label. Consider an intake entry stating that the adverse driver “ran the red light while texting.” The client may have personally observed the signal but inferred the phone use. A witness may have mentioned the phone. Or the intake specialist may have condensed several statements into one sentence. Those are materially different foundations for a demand narrative.

The same problem appears in damages. “Unable to work for six weeks” might mean the client missed every scheduled shift, performed modified duty, stopped working before receiving a disability note, or simply estimated the period during the first call. Until payroll records, employer confirmation, and medical restrictions are compared, the entry identifies a damages issue; it does not establish the amount or duration of wage loss.

California Evidence Code section 1220 may make a party’s own statement relevant as an admission, while Evidence Code section 1271 sets separate requirements for admitting a business record. Neither provision turns an internal intake paraphrase into a substitute for the underlying statement or record. Even in prelitigation negotiations, where formal admissibility is not the only concern, source quality affects credibility. An adjuster can exploit an inconsistency between a polished demand and the police report, recorded statement, treatment record, or later testimony.

Preserve context before turning notes into themes

A reliable workflow separates three layers: what was reported, what has been verified, and what the firm concludes. AI can help organize those layers, but only if the source material and instructions preserve the distinction.

Preserve contemporaneous observations and attribution

Keep the original intake entry, including who provided the information and when. A useful source label might distinguish “client reported during initial intake,” “family member reported,” “staff observation,” and “attorney assessment.” Do not silently rewrite a reported statement as an objective event.

Preserve uncertainty as well. Phrases such as “client believes,” “approximately,” “treatment provider not yet confirmed,” and “possible prior injury” are not drafting defects at the intake stage. They are flags for investigation. Removing them too early can create false precision.

Also preserve facts that may later explain apparent inconsistencies. Language barriers, medication, a chaotic collision scene, delayed symptoms, or incomplete access to records can affect the initial account. The point is not to excuse every discrepancy; it is to retain enough context for an attorney to assess it rather than discover it after the demand is challenged.

Verify each proposition against the best available source

Before a statement becomes a demand fact, connect it to the strongest available source. Liability propositions may be checked against the traffic-collision report, scene photographs, video, witness statements, vehicle damage, and applicable traffic rules. Treatment propositions belong against provider records and billing, not a staff summary. Wage-loss claims require employment and medical support. Coverage statements should be tied to declarations, correspondence, or verified policy information.

This is where a source matrix earns its keep. For each material proposition, record:

• The proposed fact: the sentence or proposition the demand may use.
• The originating source: intake call, client document, police report, medical record, bill, photograph, or attorney note.
• The verification source: the best evidence that confirms, narrows, or contradicts it.
• The current status: verified, partially verified, disputed, or unresolved.
• The drafting treatment: state directly, attribute, qualify, omit, or investigate further.

This process does not require a litigation-grade evidentiary memo for every routine demand. It requires enough discipline to prevent repetition from being mistaken for corroboration. Three internal documents repeating the same intake statement are still one source.

Separate factual proof from advocacy

Demand themes are not raw facts. They are attorney-selected explanations of why the evidence matters: a rear-end impact produced a consistent treatment course; a gap in care has a documented explanation; objective findings support continuing limitations; or a carrier’s liability position ignores corroborating evidence.

The theme should emerge after the factual layer is assembled. If the workflow starts with the desired narrative and asks an AI system to find support, confirmation bias can shape what gets included and what gets ignored. A safer sequence is facts first, conflicts second, theme third.

That does not make the demand timid. It makes advocacy more defensible. A forceful paragraph grounded in identified records is stronger than an emphatic paragraph built from a blended case summary whose sources cannot be traced.

What should be discarded—or quarantined—from the drafting set

Not every intake detail deserves to follow the file into demand preparation. Firms should remove or quarantine material that adds noise, creates avoidable privacy exposure, or invites unsupported assertions.

• Duplicate administrative entries: repeated phone numbers, scheduling chatter, and copied status notes can bury the facts that need review.
• Unsupported staff conclusions: labels such as “clear liability,” “policy-limits case,” or “client is fully compliant” should not be treated as evidence.
• Unnecessary sensitive data: use only the information reasonably needed for the task. Medical and identity data should not be uploaded merely because it exists in the intake system.
• Superseded versions: old summaries should remain available for audit purposes but should not compete with the current verified source set.
• Speculation presented as fact: assumptions about speed, distraction, causation, prognosis, or future care require support or express qualification.

“Discard” should not mean destroying material subject to retention duties or erasing an audit trail. It means excluding low-quality material from the active drafting corpus while retaining it under the firm’s records policy where appropriate.

A five-step intake-to-demand control process

1. Freeze the original intake record. Preserve the initial entry and metadata so later edits do not obscure what was first reported.
2. Tag source and confidence. Mark whether each material statement is client-reported, third-party-reported, staff-observed, document-supported, or attorney-assessed.
3. Build an exception list. Identify conflicts, missing records, treatment gaps, prior conditions, unclear wage-loss periods, and unresolved coverage questions before drafting.
4. Approve the factual set. Have the responsible attorney or designated reviewer decide which propositions can be stated, attributed, qualified, or omitted.
5. Draft from approved facts, then review against sources. The final QA pass should compare the carrier-facing narrative with the record—not merely with the internal summary used to generate it.

For a complementary framework on choosing issues before prose, see why AI demand drafting should start with issues, not templates.

How Legal Power AI fits

Legal Power AI supports PI demand-letter workflows by helping firms organize case materials and produce a reviewable draft, but the attorney remains responsible for source selection, factual accuracy, legal judgment, and the final document. When medical records or other protected information are involved, firms should apply their privacy, privilege, work-product, access-control, and vendor-governance requirements before using any AI-assisted workflow.

The strongest theme starts with source discipline

Intake notes should not be ignored, and they should not be promoted automatically into demand facts. Their value is highest when the firm preserves attribution, verifies material propositions against better sources, surfaces conflicts, and lets the attorney build advocacy from an approved factual set. AI can accelerate organization and drafting, but it cannot decide that an assumption became true because it appeared in several derivative summaries. Source discipline keeps the demand accurate, the theory coherent, and the attorney in control.