For Managing Partners · Heads of Knowledge · Legal Operations
BUILD for Legal Teams
Train your fee-earners to build their own AI tools — contract risk screeners, citation hallucination detectors, NDA comparison engines — using infrastructure your firm owns end-to-end. No vendor lock-in. No client data leaving your network. No black-box compliance risk. Built by your team, owned by your firm, defensible to the SRA.
📘 28 segments · 4 weeks👥 5–25 fee-earners per cohort⚖ SRA Code & EU AI Act aligned
Your fee-earners are already using ChatGPT, Claude, and Gemini for client work. You know it. They know you know it. Whether your firm has a policy or not, the tools are in their browsers and the work is going through them. The question isn't "should we let them use AI?" — that decision was made for you 18 months ago. The question is whether the tools they're using are yours.
⚠ The current state for most firms
Mata v. Avianca wasn't a one-off. Firms in the UK, US, Australia, and Canada have all sanctioned solicitors and barristers for filing AI-hallucinated case citations. Your professional indemnity insurer is watching.
Client confidentiality is being routinely breached every time a fee-earner pastes a draft contract into a public AI tool. Most firms have no audit trail of which clauses went through which model on which day.
Your competitors are quietly building their own internal AI tools. The firms that will win the next decade aren't the ones with the most lawyers — they're the ones with the most lawyers who can ship a working tool by Friday.
Vendor lock-in is the silent budget killer. The major legal-tech vendors charge per-seat per-month at rates that compound forever. Tools your team builds with BUILD cost roughly £20/month in compute, total, regardless of seats.
What BUILD for Legal does about it
BUILD takes any fee-earner — partner, associate, paralegal, knowledge lawyer — from "I've never written code" to a deployed AI tool running on infrastructure your firm controls. The course is the same proven 28 segments. The difference is the Legal Build Kit: pre-tuned system prompts, sector use cases, and capstone project templates that drop straight into Segments 12 and 15 to ship contract-aware, citation-aware, confidentiality-aware tools.
Section 2
What your team will actually build
Five concrete tools your fee-earners can build during the 4-week course. Each one is real, deployable, and addresses a workflow your team already does manually.
Tool 1
Contract Risk Screener
Built in Segments 11–12 · Powered by the Contract Risk system prompt below
Paste a contract clause, get back a structured analysis: the risk level (low/medium/high), the specific concern (auto-renewal, unilateral termination, indemnity scope, etc.), and the recommended redline. Used by associates to triage 50-page MSAs in 10 minutes instead of 2 hours.
Example input: "The Service Provider shall have the unilateral right to modify these Terms upon 7 days' notice, with Customer's continued use constituting acceptance."
Example output: 🔴 HIGH RISK — Unilateral modification with 7-day notice and deemed acceptance is enforceable in some jurisdictions but exposes the Customer to material change without consent. Recommended redline: require Customer's express written consent OR a minimum 30 days' notice with right to terminate without penalty.
Tool 2
Citation Hallucination Detector
Built in Segments 13–14 · Multi-model verification using the Citation Check prompt
Paste any AI-generated legal text containing case citations. The tool runs each citation through two independent models, cross-checks the output, and flags any case names or citation formats that don't match known patterns. The single most important defensive tool a firm using AI can deploy. Directly addresses the Mata v. Avianca failure mode.
Example input: "As established in Smith v. Henderson Industries [2019] EWCA Civ 4421, the duty of care extends..."
Example output: ⚠ FLAGGED — Citation "[2019] EWCA Civ 4421" exceeds the typical EWCA Civ numbering range for 2019 (final reported number was approximately 1900). No case "Smith v. Henderson Industries" appears in cross-referenced model. Recommended action: verify in Westlaw/LexisNexis before filing.
Tool 3
NDA Term Comparator
Built in Segment 14 · Multi-model orchestration via Promise.all()
Paste two NDAs side by side. The tool extracts the key terms from each (term length, scope of confidentiality, return-of-information clauses, jurisdiction, exceptions) and produces a structured side-by-side comparison. Used to triage incoming counterparty NDAs in seconds.
Example output: Both NDAs cover commercial information, but NDA-A has a 5-year term vs NDA-B's perpetual obligation. NDA-A excludes information already in public domain; NDA-B has no such carve-out. Jurisdiction: A = England & Wales, B = Delaware. Negotiation priority: the perpetual obligation in NDA-B is the higher-priority item to redline.
Tool 4
Privileged Email Draft Assistant
Built in Segments 15–16 · Sector-specific system prompt + PWA
An installable web app (PWA) on a fee-earner's phone. They dictate the bullet points of a client update; the tool drafts a privileged-marked email with appropriate caveats, the standard legal-advice-disclaimer, and a flag if the draft contains anything that should be redacted before sending. Deployed on infrastructure your firm controls.
Built-in safety: the tool refuses to draft anything that includes specific monetary advice on a position it hasn't seen evidence for, and always inserts the firm's standard "this is privileged legal advice" header.
Tool 5
Conflicts Pre-Check Browser Extension
Built in Segments 17–19 · Chrome extension that reads the current page
A Chrome extension that the fee-earner clicks while looking at a prospective client's website or LinkedIn page. The extension extracts the company names and individuals mentioned, then runs a fast cross-check against your firm's existing client list (provided as a private data source). Returns "✓ no obvious conflict" or "⚠ possible match — escalate to conflicts team." Pre-screening before the formal conflicts check.
Important: this is a pre-check, not a substitute for the formal conflicts process. The tool reinforces that in every output. But it catches the obvious cases in 5 seconds and saves the conflicts team from triaging requests that were never going to be conflicts.
Section 3
The Legal Build Kit — copy these straight into Segment 15
Five ready-to-use system prompts your fee-earners paste directly into BUILD's Segment 15 ("System Prompts — Controlling AI Behaviour") to transform the generic Text Analyser into a sector-specific legal tool. Each one is built with the 5-element framework the course teaches.
📜 Contract Risk Screener · System Prompt
For Segment 15
You are a senior commercial contracts lawyer with 15 years of experience reviewing UK and EU commercial agreements. You review contract clauses for risk to the receiving party.
EXPERTISE:
- UK contract law (English & Welsh, Scots law where applicable)
- EU directives affecting commercial agreements
- Standard market terms across SaaS, licensing, services, and supply contracts
- The most common asymmetric drafting traps in vendor-favouring contracts
CONSTRAINTS:
- You provide risk analysis, NOT legal advice. Every output must end with a reminder that this is a triage tool and a qualified lawyer must review before any contract action is taken.
- You do NOT speculate about the counterparty's intent.
- You do NOT recommend contract action that would breach professional duty of care.
- If a clause is ambiguous, you flag the ambiguity rather than guessing.
OUTPUT FORMAT (always this exact structure):
1. RISK LEVEL: 🔴 HIGH / 🟡 MEDIUM / 🟢 LOW
2. SPECIFIC CONCERN: one sentence naming the clause type and the asymmetry
3. WHY IT MATTERS: 2-3 sentences explaining the practical consequence
4. RECOMMENDED REDLINE: specific suggested wording change
5. ESCALATION NEEDED: yes/no — should this be escalated to a partner?
6. MANDATORY FOOTER: "Triage analysis only. Not legal advice. Qualified lawyer must review before action."
EXAMPLES:
Input: "Vendor may terminate this Agreement at any time without cause upon written notice."
Output:
1. RISK LEVEL: 🔴 HIGH
2. SPECIFIC CONCERN: Asymmetric termination right (vendor-only, no cause required)
3. WHY IT MATTERS: Customer has invested in onboarding/integration but has no recourse if vendor exits. This is a common SaaS vendor trap that leaves Customer stranded mid-deployment.
4. RECOMMENDED REDLINE: Make termination right mutual, OR require minimum 90 days' notice with vendor obligation to support data export and transition.
5. ESCALATION NEEDED: yes
6. Triage analysis only. Not legal advice. Qualified lawyer must review before action.
🔍 Citation Hallucination Detector · System Prompt
For Segment 15
You are a legal research librarian specialising in case citation verification across UK, EU, and US jurisdictions. You verify case citations for plausibility and flag any that appear to be hallucinated.
EXPERTISE:
- UK case citation formats: [YEAR] Court Code Number (e.g. [2023] EWCA Civ 1421)
- US case citation formats: Volume Reporter Page (Court Year)
- EU case citation formats: Case C-NUMBER/YY
- Typical numbering ranges per court per year (e.g. EWCA Civ rarely exceeds ~2000 per year)
- Common hallucination patterns in AI-generated legal text
CONSTRAINTS:
- You CANNOT confirm a citation is real. You can only flag plausibility based on format and numbering range.
- You ALWAYS recommend verification in an authoritative legal database (Westlaw, LexisNexis, BAILII) before any citation is filed or relied upon.
- If a case name is generic-sounding (Smith v. Jones, etc.) and combined with an unusual citation, you flag it with extra caution.
- You do NOT generate citations of your own.
OUTPUT FORMAT:
For each citation found:
1. CITATION: [the citation as written]
2. PLAUSIBILITY: ✓ PLAUSIBLE / ⚠ FLAGGED / 🔴 LIKELY HALLUCINATED
3. REASON: one sentence explaining the assessment
4. VERIFY IN: which database to check (Westlaw / LexisNexis / BAILII / EUR-Lex)
After all citations:
MANDATORY FOOTER: "⚠ This is automated plausibility checking only. Every citation must be verified in an authoritative legal database before filing or relying upon it. Filing hallucinated citations has resulted in professional sanctions in multiple jurisdictions (see Mata v. Avianca, 2023)."
🤝 NDA Term Comparator · System Prompt
For Segment 15
You are a commercial lawyer specialising in non-disclosure agreements. You compare two NDAs side by side and produce a structured comparison highlighting material differences.
EXPERTISE:
- Standard NDA structures (mutual, one-way, multi-party)
- Common asymmetric terms in vendor and counterparty NDAs
- Carve-outs and exception clauses
- Cross-jurisdictional NDA enforcement
CONSTRAINTS:
- You compare what is in the documents. You do NOT add terms that are not present.
- You do NOT recommend signing either NDA — you produce comparison data for a lawyer to review.
- If a key term is missing from one NDA but present in the other, you flag it explicitly as "MISSING" rather than assuming a default.
OUTPUT FORMAT:
A structured comparison table covering these terms:
- Mutuality (one-way / mutual)
- Term length (years, perpetual, until terminated)
- Definition of confidential information (broad / narrow / specific list)
- Exclusions / carve-outs (public domain, independently developed, etc.)
- Permitted disclosures (employees, advisors, regulators)
- Return / destruction of information
- Jurisdiction and governing law
- Remedies (injunctive relief, damages cap, indemnity)
- Notice period for breach
For each row: NDA-A value | NDA-B value | DIFFERENCE: brief note on which is more favourable to the receiving party.
After the table, list the TOP 3 negotiation priorities — the terms most worth pushing back on if you cannot accept the document as-is.
MANDATORY FOOTER: "Comparative analysis only. A qualified lawyer must review before signing or negotiating either document."
📧 Privileged Email Draft Assistant · System Prompt
For Segment 15
You are a drafting assistant for a UK law firm. You take bullet points dictated by a fee-earner and produce a privileged client email in the firm's standard format.
EXPERTISE:
- UK legal correspondence conventions
- Privilege markings and disclaimers
- Professional tone for client communication
- Common email types: case updates, advice memos, status reports, follow-ups
CONSTRAINTS:
- ALWAYS prepend the privilege header: "PRIVILEGED & CONFIDENTIAL — LEGAL ADVICE"
- ALWAYS append the firm's standard disclaimer at the foot of every email
- NEVER make up factual claims, monetary figures, dates, or case progression states that were not in the input bullets. If the fee-earner did not say it, you do not write it.
- NEVER recommend a specific course of action that requires regulatory or court action without flagging "CHECK BEFORE SENDING — this draft contains an actionable recommendation"
- If the input contains anything that looks like client confidential information that should be redacted before forwarding (account numbers, specific named third parties, financial figures), flag it at the top of the output.
OUTPUT FORMAT:
1. Header: PRIVILEGED & CONFIDENTIAL — LEGAL ADVICE
2. Subject line (suggested)
3. Salutation
4. Body — drafted from bullets, professional tone, paragraph structure
5. Sign-off
6. Standard firm disclaimer
7. (If applicable) REDACTION FLAG: list any items that should be reviewed before sending
MANDATORY FOOTER: "Draft only. Review every fact, figure, and recommendation before sending."
🛡 Conflicts Pre-Check · System Prompt
For Segment 15 + Chrome Extension in 17–19
You are a conflicts pre-screening assistant for a UK law firm. You receive text extracted from a webpage (a prospective client's company website, LinkedIn, news article) and identify the corporate entities, individuals, and matters mentioned that may need conflict checking.
EXPERTISE:
- Recognising company names, parent/subsidiary structures, and named individuals
- Identifying regulatory bodies, opposing counsel, and adverse parties
- Pattern-matching company name variants (Ltd vs Limited, Holdings, Group, etc.)
CONSTRAINTS:
- You are a PRE-CHECK ONLY. You do NOT replace the firm's formal conflicts process.
- You CANNOT and DO NOT access the firm's actual client list. You extract entities only — the formal conflicts team runs the check.
- You ALWAYS make this clear in every output.
- You do NOT speculate about whether a conflict exists. You list entities and flag the ones that look most material to escalate.
OUTPUT FORMAT:
1. ENTITIES IDENTIFIED:
- Companies: [list]
- Individuals: [list]
- Regulatory bodies: [list]
- Other named parties: [list]
2. ESCALATION PRIORITY: top 3 entities most worth running through the formal conflicts check first
3. ⚠ MANDATORY FOOTER: "This is a PRE-SCREEN ONLY. Every prospective engagement MUST go through the firm's formal conflicts checking process before any client work begins. This tool identifies entities; the conflicts team confirms conflicts."
Section 4
The 70/30 model — what's generic, what's legal-specific
BUILD for Legal isn't a separate course. It's the existing 28-segment BUILD course (the same one any other professional takes), plus the Legal Build Kit your fee-earners drop in at three specific points. This is intentional and matters for compliance reasons.
70% — the BUILD course core (unchanged)
The technical pipeline your team learns is identical regardless of sector: HTML / CSS / JavaScript frontends, Cloudflare Workers as a secure proxy, the Anthropic API for AI calls, GitHub for version control, Netlify for hosting. This is the standardised, defensible infrastructure layer that the firm controls end-to-end. Same code. Same architecture. Same security posture. Audit-ready.
30% — the legal customisation
The legal-specific layer is the system prompts (Segment 15), the use case examples (Segments 12 and 14), and the capstone project briefs (Segment 28). These swap in via copy-paste — your fee-earners take the prompts from Section 3 above and use them where the generic course says "your sector prompt here." The Build Kit also includes legal-tuned versions of: the Multi-Model Compare tool (Segment 13) for citation cross-checking, the System Prompt framework (Segment 15) for legal use cases, and the Final Project rubric for legal-relevant capstone projects.
Why this matters for compliance
Because the underlying technical architecture is identical to every other BUILD cohort, your IT and information security team can review and approve it once, and that approval covers every fee-earner who ever takes the course. The legal customisation is purely at the prompt and use case layer — which is where your firm's existing professional judgement lives. Compliance reviews the architecture once. Legal reviews the prompts. Both teams stay in their lane.
Section 5
Compliance & regulatory alignment
BUILD for Legal is positioned to help your firm meet (and document) compliance with multiple converging requirements.
SRA Code of Conduct
The Solicitors Regulation Authority requires firms to maintain competent supervision over technology used in client work. Tools your team builds with BUILD are fully auditable, version-controlled in your firm's GitHub, and run on infrastructure you control — exactly the supervision posture the Code expects.
Bar Standards Board
Equivalent obligations apply to barristers' chambers. The transparency of code-based tools (vs black-box vendor SaaS) makes it materially easier to demonstrate competent oversight to the BSB.
EU AI Act (Article 4)
From August 2026, organisations using AI systems must demonstrate "a sufficient level of AI literacy" among staff who operate them. BUILD produces a per-fee-earner artefact (live tool, GitHub portfolio, README documentation) that evidences exactly this.
Client confidentiality
Tools built with BUILD use the Cloudflare Worker proxy pattern: API keys never leave the server, requests are routed through infrastructure your firm owns, and you can deploy with regional pinning to keep data inside the UK or EEA. Materially better than fee-earners pasting drafts into public AI tools.
Equality Act 2010
As employers, law firms are bound by the Equality Act 2010 in their own staffing and supplier choices. Tools your fee-earners build with BUILD are designed to be auditable and bias-checkable, which gives the firm a stronger answer if a tool's outputs are ever scrutinised under the protected-characteristic provisions. Black-box vendor tools cannot offer the same audit trail.
⚖ A specific note on professional indemnity
PI insurers are starting to ask firms about their AI governance. "We use the major commercial legal-AI vendor" is increasingly a yellow flag because the firm has no visibility into the model, the prompt, or the data path. "We train our fee-earners to build their own auditable tools running on infrastructure we control" is a much stronger answer. Several BUILD-graduated firms have used the cohort artefacts in PI renewal discussions. We can introduce you to one of them on request.
Section 6
Pricing — for legal teams
Three tiers based on cohort size. All prices are the firm-wide commercial rate, not per-seat consumer pricing. Includes the full BUILD course, the Legal Build Kit, the Manager Pack, and email support across the rollout.
Pilot Cohort
£4,500 / cohort
Up to 10 fee-earners
Full 28-segment BUILD course
Legal Build Kit (5 system prompts)
Manager Pack + Capstone rubric
Email support across the 4 weeks
One IP whitelist consultation
Department Rollout
£9,500 / cohort
Up to 25 fee-earners
Everything in Pilot
Buddy pairing + cohort kickoff call
Mid-point manager check-in (60 min)
Capstone showcase facilitated by ET
Anonymised cohort Risk Report aggregation
One sector-specific prompt customisation
Firm-Wide
From £18,000
25–100+ fee-earners across multiple offices
Everything in Department Rollout
Multiple parallel cohorts
Train-the-trainer for in-house champion
Custom Legal Build Kit additions
White-label option for internal LMS
Quarterly check-ins for 12 months
All prices ex-VAT. Procurement-friendly invoicing available. Email hello@everythingthreads.com for a tailored quote or to arrange a discovery call.
Section 7
FAQ — for legal team leadership
Can our fee-earners actually do this? They're not developers.
That's exactly who BUILD is designed for. The course starts at "what is a terminal" and finishes with a deployed, working AI tool. Across hundreds of non-developer students — including lawyers, paralegals, and knowledge managers — completion rates for cohorts with manager air cover (see the Manager Pack) sit in the 80%+ range. The fee-earners who finish BUILD become the in-house champions for everyone else.
How is this different from the legal-AI vendors we already use?
Three differences. First, ownership: tools built with BUILD belong to your firm, run on infrastructure you control, and can be modified or retired without vendor permission. Second, cost: vendor seat licences compound forever; BUILD is a one-off cohort cost plus ~£20/month in compute. Third, oversight: your IT team can review the actual code, your fee-earners understand exactly what the prompts do, and your compliance team has full audit visibility — something the major vendors don't offer.
What about client confidentiality?
The Cloudflare Worker proxy pattern (taught in Segment 11) keeps API keys server-side and routes requests through infrastructure your firm controls. You can deploy with regional pinning to keep data inside the UK or EEA. Critically, BUILD teaches fee-earners to think about data flow as a first-class concern — most firms find their staff understand confidentiality risks better AFTER BUILD than before, regardless of which tools they end up using.
Who owns the tools the fee-earners build?
Your firm. The code lives in your firm's GitHub. The infrastructure is provisioned in your firm's accounts. BUILD's terms of service grant the student a perpetual, transferable license to the course materials and explicitly disclaim any vendor claim on the work product. Standard work-for-hire applies as it would for any other internal development.
What if a fee-earner builds something that gives bad legal advice?
Every system prompt in the Legal Build Kit explicitly instructs the AI that its output is triage, not advice, and requires the standard "qualified lawyer must review" disclaimer in every output. Segment 24 (Testing Your AI Tools) walks through edge case handling. Segment 27 (Security, Safety & Guardrails) covers the broader risk controls. The tools are designed to support fee-earner judgement, not replace it — and the course explicitly says so, repeatedly.
How long does the rollout take from kick-off to first cohort?
Typically 2–3 weeks from contract signature to Day 1 of the cohort. Most of that time is IT whitelisting (VS Code, Git, Node.js) and cohort selection. Once the course starts, it runs 4 weeks. Total elapsed time from "we want this" to "we have fee-earners with deployed tools" is around 7 weeks.
Should we run SHARP first or go straight to BUILD?
For most firms: SHARP first across the whole department, BUILD second for the technically curious subset. SHARP is the AI literacy and risk vocabulary layer — 4 weeks, 2–4 hrs/week, no installs. After SHARP, your fee-earners share a vocabulary for AI risk ("that's an M1, the Agreement Trap") which makes BUILD's technical content land harder. Combined SHARP + BUILD pricing is significantly better than buying them separately. Email us for the combined tier.
Do you offer non-accredited training? We have CPD obligations.
BUILD is currently non-accredited. CPD accreditation through the SRA's recognised CPD provider list is in our pipeline for late 2026. In the meantime, the course outputs (deployed tool, GitHub portfolio, README documentation) provide stronger evidence of practical learning than most accredited courses produce. Several firms have logged BUILD as CPD time under the "self-directed learning" category with their compliance team's approval. Happy to share the precedent on request.
Ready to talk?
If you're a managing partner, head of knowledge, or legal operations lead and you want to bring BUILD for Legal to your firm, the next step is a 30-minute discovery call. We'll walk through your current AI use, your compliance constraints, and which cohort tier makes sense.
EverythingThreads is contact-by-email only. We reply within 2 working days. For urgent matters during a paid rollout, mark the email subject "URGENT" and we'll prioritise.
