Drafting With Technology Doesn’t Have to Be Painful: A Practical Framework from Catherine Bamford

What if improving legal drafting wasn’t about buying the newest AI tool, but about getting clear on what drafting actually involves and matching the right technology to the right job?

That was a big theme in Radiant’s recent webinar with Catherine Bamford (BamLegal), a long-time legal tech and legal engineering expert who has spent the last 15+ years helping law firms and in-house teams improve how they draft, negotiate, and deliver documents.

Catherine shared a simple, useful framework for cutting through the noise in today’s crowded legal tech market and making real progress, especially in high-volume contracting work.

Here are some key takeaways from the conversation:

1. Drafting Is Not One Task (It’s Three)

Before choosing tools, Catherine recommends breaking “drafting” into three distinct activities:

• Creating the first draft (getting something workable on the page)
  
• Bespoking (tailoring the document to the specific deal or matter)
  
• Negotiating (responding to markups, redlines, or third-party templates)
  

This matters because different tools shine in different parts of the process. If a tool is great at first drafts but weak in negotiation, or vice versa, it will frustrate users and sit unused.

2. Match the Tool to the Document Type (Not the Hype)

Catherine also shared a useful way to think about document types, because drafting needs vary widely:

• Transactional (contracts, leases, loan agreements)
• Litigation (court forms, witness statements, briefs to counsel)
• Recording (board minutes and structured records of what happened)
• Reporting (due diligence reports, corporate or banking reports)
• Advisory (client memos and advice)

The point: don’t treat “AI for drafting” as one category. A tool that works well for board minutes may be totally wrong for a complex SPA.

‍

3. Document Automation vs GenAI: Deterministic vs Probabilistic

One of Catherine’s clearest distinctions was this:

• Document automation is deterministic: “If this, then that.” Same input leads to the same output every time.
  
• GenAI drafting assistants are probabilistic: they generate what’s most likely next based on patterns, so results vary each time.
  

That means:

• If you need repeatable, safe, consistent outputs, document automation is often the better fit.
  
• If you need help turning notes into a structured narrative, GenAI can be very effective.
  

Catherine’s practical example:

• Board minutes + meeting notes + a simple template = good GenAI use case
  
• A long, complex lease suite or SPA with conditions and cross-references = document automation is usually safer and better
  
  

4. The Line Is Blurring (And That’s a Good Thing)

Historically, document automation tools and GenAI tools were separate categories. Catherine explained that vendors are now blending both approaches:

• Doc automation platforms are using AI to help design questionnaires, suggest questions, and speed up “build” work.
  
• GenAI tools are trying to add more structure and predictable outputs, using guardrails and more deterministic prompting.
  

The outcome is that teams can increasingly combine AI’s speed with automation’s reliability, especially when drafting needs to scale.

‍

5. Templates Are the Biggest Blocker (Not Technology)

A key message from Catherine: even the best tool won’t help if your starting documents are a mess.

Common blockers include:

• No agreed precedent or template
• Too many options and clauses that nobody understands
• Old “patchwork” documents that grew over years of risk layering

Catherine also shared a helpful principle: if lawyers keep selecting Option A 90% of the time, consider removing B and C. Modern tools can give data that supports continuous improvement and document simplification over time.

‍

6. Don’t Ask “What Should We Automate?” Ask “What Do We Do Every Day?”

Catherine’s advice for getting traction was simple: start with daily work, not documents.

A better discovery approach:

1. Ask the team: What matters are you working on most often?
2. For each matter: What documents or tasks are involved?
3. Then: What is your current starting point? (template, past example, nothing, third-party paper)

This gives a realistic view of readiness and helps identify patterns. It also stops teams from investing in tools before they understand what they truly need.

‍

7. In-House Teams Don’t Always Need Big Budgets to Move Forward

A question came up about teams with limited tech budget. Catherine’s answer was practical:

• If you already have Microsoft, there is more you can do than people realise (Power Automate, Forms, workflows), especially with internal tech support.
  
• Some legal-specific tools for drafting and review are now available at much lower price points than expected, sometimes under $100 per user per year depending on use case.
  

Her bigger point: focus on the one task you most want to remove or reduce, then build the business case around time saved and risk reduction.

‍

8. InfoSec Must Be In the Room Early

Catherine stressed that teams should involve IT and InfoSec from the start.

Otherwise, you can spend months on demos and selection only to hit a long delay right before piloting. Involving the right experts early prevents “red tape surprises” late in the process.

9. Pilots Fail When Nobody Runs Them

Catherine also flagged a common pattern: pilots that “die quietly” because no one manages them.

If you run a pilot, treat it like a project:

• Set success criteria
• Decide what you’re testing
• Assign owners
• Track actual usage
• Define what “go/no-go” looks like

Otherwise, teams often reach the end of a trial period without enough evidence to make a confident decision.

‍

Final Thought: Start Small, but Start Properly

The most consistent message from Catherine was that progress comes from clarity and focus, not more tools.

Start with:

• What your team does every day
• Which part of drafting is slowing you down
• Whether you need repeatability (automation) or flexibility (GenAI)
• Whether your templates and precedents are usable
  

Then narrow the market to a small shortlist and test with purpose.

If you get those basics right, the technology becomes far less overwhelming and a lot more effective.