TechnologyLawyers Author, AI Reconciles: Why ilovelawyer Refuses to Pick a Winner
Standfirst: When two lawyers on the same case disagree, most software would merge their views into one answer. FORHU built ilovelawyer to do the opposite. Picture two lawyers on the same estate…
By MAGAZINE TECHY·November 6, 2026·2 min
Standfirst: When two lawyers on the same case disagree, most software would merge their views into one answer. FORHU built ilovelawyer to do the opposite.
Picture two lawyers on the same estate claim. One, J. Patel, builds the theory around maintenance, resting on the bank transfers. The other, S. Okafor, argues unmet needs, resting on the GP evidence. Both are reasonable, and they cannot both lead.
A general-purpose AI asked to help would probably write a blended theory. ilovelawyer, the AI case workspace FORHU builds for lawyers in the UK and the Philippines, is designed not to.
Side by side, never merged
In ilovelawyer, Case Theories keeps each lawyer's theory as they wrote it, side by side. The AI does not rewrite them into one. It produces a diff instead: what the theories share, where they disagree, and the evidence that would decide between them.
In the fictional demo case, the diff points to the weight of a 2021 letter of wishes. That turns a disagreement between colleagues into a list of evidence to get.
The same idea, aimed at the other side
The design choice runs through the product's analysis tools. The Red Team pane applies it to the opponent. It lists opposing counsel's likely arguments, ranked by threat, each with a suggested response. In the same demo case, its top threat is the claim that "the payments were gifts, not maintenance". The response it suggests is to prove the purpose of the payments and the claimant's needs.
Case Reconstruction does something similar for the narrative. It writes the story three ways (general, for the court, and from the other side), then lists what it does not cover.
Why not just give the answer?
Because in litigation, the answer is the lawyer's job. FORHU presents every output here as a provisional assessment that a lawyer can edit, re-rate or dispute. The company makes no claims about outcomes. The software organizes the disagreement; the people who are accountable resolve it.
That is a less dramatic pitch than "AI that wins cases". It may be the more durable one.
