AI consultancy vs legal tech SaaS: what is the difference?

The core difference is ownership and fit. Legal tech SaaS tools like Harvey, Luminance, and Clio AI provide ready-made AI capabilities that work across many firms. AI consultancies like Formulaic build custom systems designed specifically for your workflows, your data, and your practice areas. SaaS is faster to start and cheaper upfront. Custom builds deliver better fit, lower long-term costs at scale, and complete ownership. Most firms benefit from a combination of both.

Short answer: SaaS gives you general tools fast. Consultancies build systems that fit your exact workflows. Best approach: SaaS for broad capabilities, custom for your differentiating processes.

Why this question matters now

The legal AI market has split into two distinct camps, and the marketing from both sides can be confusing. SaaS vendors position their tools as comprehensive solutions. Consultancies position custom builds as the only serious approach. The truth is more nuanced, and choosing wrong is expensive.

In 2026, the SaaS market has matured. Harvey has raised over $200 million. Luminance serves hundreds of firms globally. Clio AI is embedded in one of the most popular practice management platforms. These are not experimental products. They work, and they work well for their intended use cases.

Simultaneously, the demand for custom AI systems has grown because firms are discovering that general tools do not solve specific problems. An immigration firm’s intake workflow is fundamentally different from a family law firm’s. A conveyancing compliance check requires different logic than a commercial litigation review. General tools handle the common denominator well. They struggle with the specific.

Understanding the trade-offs between these approaches is essential for making a smart investment decision rather than an emotional one driven by whichever sales pitch you heard most recently.

What SaaS tools do well

Breadth of capability. Harvey can draft, research, summarise, and analyse across dozens of practice areas. Building equivalent breadth from scratch would cost millions. SaaS spreads that development cost across thousands of users.

Speed to value. You can start using most SaaS tools within a week. Account setup, basic training, and initial workflow configuration are straightforward. For firms that need immediate capability, this speed matters.

Continuous improvement. SaaS vendors update their products constantly. New features, better models, improved accuracy arrive without you doing anything. A custom system requires deliberate updates.

Community and support. Popular SaaS tools have user communities, support teams, training resources, and integration ecosystems. Custom systems have your consultancy and your internal team.

The major players in 2026:

• Harvey: Best for general legal research and drafting. Strong across practice areas. Enterprise pricing, typically £200+ per user per month.
• Luminance: Leads in contract review and due diligence. Particularly strong for M&A and corporate work. Enterprise pricing.
• CoCounsel (Thomson Reuters): Deep integration with Westlaw. Best for research-heavy practices. Bundled with Thomson Reuters subscriptions.
• Clio AI: Embedded in Clio practice management. Best for firms already using Clio who want AI features without switching platforms. Per-user pricing.
• Smokeball AI and LEAP AI: Practice management integrated AI for smaller firms. Good entry points for firms wanting AI within their existing toolset.

What custom-built systems do well

Exact workflow fit. A custom system is designed around how your firm actually works, not how the average firm works. If your employment intake process has 7 specific qualification steps, your custom system handles exactly those 7 steps. A SaaS tool handles a generic version.

Integration depth. Custom systems can integrate deeply with your specific practice management system, accounting software, document management, and communication tools. SaaS tools offer standard integrations but cannot account for every firm’s technology stack.

Data ownership. You own the code, the data, and the system. If you want to change providers, modify functionality, or host the system differently, you can. With SaaS, you are dependent on the vendor’s roadmap and pricing decisions.

Cost at scale. SaaS pricing is per-user. A firm with 40 users paying £150 per month per user spends £72,000 per year. A custom system costing £30,000 to build with £6,000 annual maintenance costs £36,000 in year one and £6,000 per year thereafter. Over three years, the custom system costs £48,000 versus £216,000 for SaaS.

Competitive differentiation. If your AI system is the same tool your competitor uses, it provides efficiency but not differentiation. A custom system built around your specific processes, expertise, and client base becomes a competitive asset that others cannot replicate.

When to choose SaaS

Choose SaaS when your needs are general (legal research, basic drafting, standard contract review), you need capability immediately and cannot wait 4 to 8 weeks, your team is small (under 10 users) making per-user pricing economical, you want to experiment with AI before committing to custom builds, or the workflow you are automating is common across the industry.

When to choose custom

Choose custom when your workflow is specific to your firm or practice area, you have 20+ potential users making per-user SaaS pricing expensive, you need deep integration with your existing systems, data ownership and residency are critical requirements, or the process you are automating is a competitive differentiator.

The hybrid approach

Most mature firms land on a hybrid model. SaaS tools handle general capabilities: research, broad drafting, basic document review. Custom systems handle firm-specific workflows: intake for specific practice areas, bespoke document types, compliance checks against specific regulatory frameworks, client communication flows tuned to the firm’s style and process.

This approach captures the best of both: breadth from SaaS, precision from custom builds. The key is being deliberate about which processes deserve which approach rather than defaulting to one.

What we have seen at Formulaic

We are a consultancy, so we have an obvious bias toward custom builds. We try to be honest about it. When a prospective client’s needs are well-served by an existing SaaS tool, we tell them. It would be dishonest and counterproductive to build a custom system when Harvey or Luminance already solves their problem.

Where we consistently see custom outperform SaaS is in practice-specific intake systems, firm-specific document workflows, and processes where the firm’s approach is genuinely different from the industry standard. The Calder & Reid employment intake system could not have been replicated by any SaaS tool because it encodes Calder & Reid’s specific qualification criteria, their fee structures, and their routing logic. That specificity is what delivered the 70% reduction in unqualified calls.

Across 30 production systems shipped, approximately 80% were for workflows where no adequate SaaS alternative existed. The other 20% were for workflows where SaaS alternatives existed but could not integrate deeply enough with the client’s existing systems. In every case, the decision was driven by fit, not ideology.