Best Practices: Legal and Contract Drafting — Rakenne vs Ironclad, Harvey, Litera, HotDocs

Legal and contract drafting sits at the intersection of structure (clauses, playbooks, standards) and generation (first drafts, amendments, review). This article outlines best practices for choosing a drafting approach and compares Rakenne to the main legal/contract alternatives: Ironclad, Harvey, Litera, Contract Express (Thomson Reuters), and HotDocs.

Best practices in this space

1. Anchor on authority — Drafts should pull from approved clause libraries, playbooks, or standards. The tool should make it easy to load and enforce those references, not rely on the model’s memory.
2. Separate assembly from negotiation — Document assembly (question-driven or workflow-driven first draft) is different from redlining, approval workflows, and CLM. Choose a tool that matches where you need strength.
3. Make workflows auditable — For regulated or high-risk documents, the path from “scope → references → draft → check” should be explicit and repeatable.
4. Use checks where possible — Logic gates (e.g. “no solutions before root cause”), completeness checks, and required sections reduce risk of incomplete or non-compliant drafts.
5. Decide who authors the “spec” — Lawyers or domain experts should own the workflow and references; the tool should make that authoring straightforward (e.g. markdown, playbooks, or visual config).

Alternatives in legal / contract drafting

Rakenne vs alternatives: features, strengths, weaknesses

Rakenne

Features: Document-elaboration workflows defined in markdown skills; users work with an LLM agent in the browser; one agent per project; skill library (e.g. contract-elaboration, SOW); references and AGENTS.md for context; export to DOCX, PDF; optional extension tools (validation, logic gates).

Strengths

• Workflow as spec — Skills define steps (e.g. scope → load playbook/reference → draft → validate); repeatable and transparent; experts author in markdown.
• References on demand — Clause libraries, standards, or playbooks live in the skill/workspace; the agent loads them so drafts stay aligned to authority.
• Validation and gates — Extension tools can enforce “no solutions before root cause,” required sections, or completeness; the agent corrects until checks pass.
• Single agent, one project — One conversation and one workspace per matter or deal; no scattered “which system produced this clause.”
• No lock-in on logic — Workflows and references are files in the workspace; portable and versionable.

Weaknesses

• Not a full CLM — No built-in negotiation, approval workflows, repository, or lifecycle tracking; Rakenne is drafting and elaboration, not end-to-end contract lifecycle.
• No Word-native editing — Users draft in the browser and export; redlining and “edit in Word” happen outside Rakenne.
• No out-of-the-box clause bank UI — Clause libraries are reference files (e.g. markdown); there’s no dedicated “clause picker” or visual playbook builder.
• Legal-specific features — No built-in comparison (blackline), party/entity management, or matter-centric repository.

Ironclad

Features: Contract lifecycle (create, negotiate, sign, store, renew); workflow automation; AI for drafting and review; integrations (Salesforce, etc.); repository and analytics.

Strengths: Full CLM; strong for companies that need negotiate → sign → store in one place; good for high volume and process.

Weaknesses: Heavier and process-centric; drafting is one phase of the lifecycle; less “workflow as code” for experts who want to own the drafting spec in markdown.

Harvey

Features: Legal AI for drafting, research, and review; model-driven; API; law-firm oriented.

Strengths: Strong AI capabilities; good for research and complex drafting; flexible use cases.

Weaknesses: Less structured “workflow + references + validation”; more conversational and model-centric; no first-class document-assembly or clause-library workflow.

Litera

Features: Word add-in; templates; clause libraries; compare (blackline); compliance and drafting tools; AI drafting.

Strengths: Lawyers stay in Word; familiar comparison and template workflow; deep Word integration.

Weaknesses: Office-centric; workflow and “spec” live in Litera’s config, not in portable markdown; less emphasis on agent + validation tools.

Contract Express (Thomson Reuters) / HotDocs

Features: Question-driven document assembly; clause libraries; templates; produce first draft from answers.

Strengths: Mature assembly model; predictable output from answers; strong for standardized documents and repeatable questionnaires.

Weaknesses: No conversational LLM agent; logic is in the assembly engine (HotDocs/Contract Express), not in an open workflow format; less flexible for “chat to refine” and validation loops.

When to choose which

• Choose Rakenne when: You want workflow-defined, validated contract (or legal) drafting with a single agent per project, and your experts are comfortable authoring workflows and references in markdown. You’re okay with export for Word/redlining and will use a separate CLM or deal desk if needed.
• Choose Ironclad when: You need full CLM (create → negotiate → sign → store) and want drafting as one phase inside that lifecycle.
• Choose Harvey when: You need flexible legal AI (research, drafting, review) and don’t require a fixed workflow or validation tools.
• Choose Litera when: Word-native drafting, comparison, and clause libraries are mandatory and your team won’t adopt a browser-first tool.
• Choose Contract Express or HotDocs when: You want classic question-based assembly with minimal “conversation,” and your document types are well-defined by questionnaires and clause sets.

Best practice: align the tool to where you need control—workflow and references (Rakenne), lifecycle and process (Ironclad), Word and comparison (Litera), or question-based assembly (Contract Express, HotDocs).