Every allegation met. Every theme answered once. Every gap visible.
analysing opposing evidence and structuring affidavits in reply.
Reply evidence that strays into new territory risks being struck out; reply evidence that misses an allegation concedes it by silence. The margin for error is a paragraph wide.
When responsive evidence lands (often several affidavits at once, weeks before hearing) the reply task is deceptively brutal:
- Every adverse paragraph must be met. Silence can be read as acceptance.
- Every reply paragraph must be proper reply, responsive to their evidence, not a new case, or it invites strike-out.
- Repeated themes must be answered once, well. Opposing deponents typically repeat the same handful of narratives; answering each occurrence separately produces six inconsistent rebuttals instead of one strong one.
- Admissions must be harvested. The other side’s affidavits routinely contain concessions that help your case, found today mostly by luck.
Done by hand, this means one drafter holding hundreds of paragraphs from multiple deponents in their head simultaneously, under the worst deadline pressure in the litigation calendar. Fatigue guarantees something gets missed, and what gets missed doesn’t announce itself until submissions.
Decompose · Build · Validate · Supervise
Every paragraph of every opposing affidavit becomes a unit of analysis with a defined question set: Is it adverse, neutral, or helpful? What does it actually assert? What contemporaneous material touches it? Is it proper subject-matter for reply, or bait to overreach?
Three connected structures:
- A per-deponent posture map, every paragraph classified, with the reasoning attached. This regularly surfaces the counterintuitive: “opposing” witnesses whose evidence substantially helps, and precisely which concessions to harvest.
- A cross-cutting themes register. The system detects the narratives that recur across deponents and pairs each with a single evidence-anchored rebuttal, answered once, deployed consistently.
- An issues matrix, allegation by allegation: what they say, what the record says, where the reply answers it, and its status. The rows marked unaddressed are the whole point.
Every draft paragraph is audited against two rules before it reaches the legal team: it must anchor to the source record or the deponent’s own words, and it must be reply, not new case. Fail either test, and it’s flagged, not filed.
The system produces working drafts and the matrix; solicitor and counsel make every call. Each paragraph carries its rationale and source anchors, so counsel can interrogate any line back to its evidence.
| Before | After |
|---|---|
| Hundreds of paragraphs held in one drafter’s head | Every paragraph mapped, classified, and traceable |
| Repeated themes answered inconsistently | Each theme answered once, deployed everywhere |
| Gaps invisible until opposing submissions expose them | Gaps surfaced as explicit matrix rows marked “unaddressed” |
| Admissions found by luck | Admissions harvested systematically from every deponent |
The system doesn’t write the evidence, the deponent’s words stay the deponent’s, and counsel settles every paragraph. What engineering guarantees is coverage (nothing unanswered), discipline (nothing that overreaches), and memory (nothing inconsistent between documents drafted weeks apart), the three things human drafting under deadline reliably loses first.