Educational drafting challenge
Beyond Precedent
Beyond Precedent is Beyond Horizons’ educational drafting challenge: two law students from any country draft an agreement or legal document using ten prompts and free public AI tools, then revise with a short edit log. The page centres the academic research that grounds the work and how our team — which started independently in 2023 — is designing legal services for a post-AI age, including how talent learns when generative AI sits in the first draft.
Beyond Precedent is an educational drafting challenge run by Beyond Horizons. It is not legal advice, not a client engagement, and not a guarantee of internship, employment, or any other opportunity with Beyond Horizons or any related entity. Student drafts and edit logs are teaching materials only and must not be used with real counterparties.
Why we started this
Beyond Horizons’ team began independently in 2023.
Beyond Precedent was initially envisioned as vibe drafting — drafting with AI the way “vibe coding” uses AI for code.
As a practice we took on a further challenge: design legal services for a post-AI age, including how to develop legal talent when models produce the first draft.
How it works
- Invite two law students (any nation) into a supervised session.
- Issue a fixed set of ten drafting prompts (fictional / teaching fact patterns only).
- Each student drafts with any free public AI tool and names the tool (and model if known).
- Each submission includes a clean supervisor-ready draft plus a short edit log (what the AI got wrong, invented, or the student rewrote).
- A small panel — firm counsel and a guest in-house counsel — scores for structure, hallucination control, clause fitness, jurisdiction humility, and process honesty.
- Selected artefacts may enter the public archive only after consent and anonymisation rules are cleared — never as client work.
Research premise
Firms and schools are testing how generative AI changes junior drafting speed, quality, and later judgment. Beyond Horizons runs Beyond Precedent to explore talent development and service design in that setting. The studies do not endorse Beyond Horizons.
University of Minnesota exam study (Choi & Schwarcz), as reported by Reuters, 29 August 2023
- Students sat exams without AI, then with GPT-4 after prompting training.
- As reported, multiple-choice scores rose sharply with GPT-4 (about +29 points overall; about +45 for lower-performing students).
- As reported, essay performance did not improve on average; top performers scored about 20 points lower with GPT-4.
- The authors discussed a possible equalizing effect across skill levels, and cautioned that heavy AI framing can blunt independent issue-spotting.
Bednar, Cleveland, Erbsen & Schwarcz — Artificial Intelligence and Human Legal Reasoning
- As reported, a randomised controlled trial of roughly 100 upper-level Minnesota law students across synthesis, comprehension, application and revision tasks.
- In that study, AI on early synthesis produced stronger, faster memos.
- Early AI use did not impair later comprehension; the AI-exposed group outperformed on a later application task without AI.
- When everyone used AI to revise, weaker memos improved and stronger memos often regressed.
- The authors’ takeaway: AI does not inevitably erode or promote independent legal reasoning — effects depend on when and how it is used.
Beyond Precedent therefore trains the skill the research keeps circling — judgment about when to trust, rewrite, or reject a model draft — not “faster paste.”
Figures and findings are summarised from published research as reported in the cited sources. They describe those studies’ settings and do not predict outcomes for any particular student, lawyer, or matter.
Session archive
A post-editing work product is what a student produces after the AI-assisted first pass: a revised draft plus a short edit log recording what the model got wrong, invented, or needed rewriting. Archive items, when cleared, are illustrative only — not legal advice and not client deliverables.
Why a law firm runs this
- Designing legal services for a post-AI age.
- Talent development when generative AI sits in the first draft.
- Supervising AI drafts as a professional skill for juniors.
- Separating “first draft from a model” from “counsel-ready paper.”
Frequently asked questions
What is Beyond Precedent?
Beyond Precedent is Beyond Horizons’ educational drafting challenge. It was initially envisioned in 2023 as vibe drafting — drafting with AI the way vibe coding uses AI for code. In each session, two law students from any country draft an agreement or legal document using ten prompts on free AI tools, then revise with a short edit log. The firm uses the programme to explore how talent develops and how legal services should be designed when generative AI is in the first draft.
Who can take part?
Law students from any nation may be invited into a session. Participation is by selection for each cohort; this page describes the concept — it is not an open application portal unless a live apply form is added later.
What do the ten prompts and free AI tools mean in practice?
Each pair works from a fixed set of ten drafting prompts and may use publicly available free AI writing tools. The point is not a vendor bake-off. The point is to watch how students prompt, what the model returns, and what human editing must still do.
What are “post-editing work products”?
After the AI-assisted first pass, students revise the draft and record a short edit log. When cleared for public view, the session archive may show selected revised excerpts or artefacts — illustrative only, not client deliverables and not legal advice.
How does research on AI and legal writing relate to this programme?
Public research often finds that generative AI can change drafting speed and can help some writers more than others, while quality and later reasoning still depend on human judgment. Beyond Precedent is grounded in that conversation, including widely reported AI-and-legal-writing studies and related academic work such as SSRN papers on AI and human legal reasoning. The firm uses the challenge to explore talent development and post-AI service design — it does not claim the studies endorse Beyond Horizons.
Is this a hiring pipeline or a job offer?
No. Beyond Precedent explores how future lawyers learn to supervise AI drafts. It is not a promise of employment, internship placement, pupillage, or a retainer.
Does Beyond Horizons use these session drafts for client matters?
No. Session materials are for teaching, demonstration, and talent-exploration only. Client work stays under separate engagement and professional obligations.
How do I follow sessions or ask about the next cohort?
Use the consultation or contact path on beyondhorizons.sg, or watch Insights for related writing.
Ask about the next cohort
If you advise juniors, run a legal team, or would like to guest-judge a session, ask about the next cohort. Schedule a consultation, contact us at HL@beyondhorizons.sg, or chat on WhatsApp.
Beyond Precedent is an educational drafting challenge run by Beyond Horizons. It is not legal advice, not a client engagement, and not a guarantee of internship, employment, or any other opportunity with Beyond Horizons or any related entity. Student drafts and edit logs are teaching materials only and must not be used with real counterparties.