Legaltechnology iconLegaltechnologyOct 1, 2026 ~3 min source read

September Orange Rag: What Law Firms Are Actually Doing with Legal Tech

A concise roundup of adoption trends, vendor moves, cost risks, security concerns and hard numbers from recent legal-technology activity described in the September Orange Rag.

The September Orange Rag Is Here!

Share this story

Send the public story page.

Useful takeaways from this story.

GenAI dominated firm efforts: 40% of surveyed firms focused on generative AI over the past year while cloud apps/SaaS were second at 25%.

Adoption breadth lags: many tools are in use but few firms have deployed them to more than 30% of staff.

Vendors and firms are shifting to proprietary stacks and in-house model work, driven by rising token costs and control concerns.

# What the September Orange Rag reports

This month's Orange Rag compiles survey results, vendor moves and operational realities that should shape legal-technology planning for general counsel and firm leaders.

Adoption priorities and the reality inside firms

An ILTA survey asked 500 law firms which technological area they concentrated on over the past year. The headline numbers:

  • 40% prioritized generative AI.
  • 25% prioritized cloud applications and SaaS.
  • 52% still keep their accounting systems on-premises.

Those percentages show where attention and budgets went, but they also reveal a gap between interest and broad rollout. Usage metrics for many tools remain concentrated among early adopters rather than across most staff.

Which tools are installed, and how widely?

Market penetration numbers cited in the report:

  • Microsoft Copilot: 76% of firms using it.
  • Anthropic Claude: 44%.
  • Thomson Reuters CoCounsel: 44%.
  • Harvey: 43%.

Vendor moves and in-house model work

Frontier AI providers intensified focus on the legal market:

  • OpenAI launched Astra for Law.
  • Google released Gemini Enterprise for Legal.
  • Anthropic hired Robert Mahari to lead its legal vertical.

At the same time, established legal-data vendors and firms are investing in model development and capacity:

  • Thomson Reuters invested $40m to train its own legal LLM, called "Thomson."
  • Harvey and Thomson Reuters are building open-weight or open-source model capabilities.
  • Latham & Watkins purchased Nvidia GPU servers to fine-tune open-weight models in-house.

These developments reflect a shift in thinking: firms and vendors are weighing which stack to depend on and whether to retain control by building or tuning models internally.

Cost and security pressures to plan for now

Two operational pressures are driving behavior.

The Orange Rag collects several snapshot numbers relevant to strategy discussions:

  • $15.5bn: Harvey's valuation after a $550m round (ARR $400m).
  • $40m: Thomson Reuters' investment to train its legal LLM.
  • $1bn: Morgan & Morgan's committed tech and AI spend over the next decade.
  • 63%: Share of firms still pricing matters with traditional methods (BigHand).
  • 14x: Reported rise in token usage at Harvey over six months.
  • 5,700: Attendees at ILTACON 2026 in Nashville over four days.

What to do next

For legal leaders: treat GenAI projects as business-change efforts, not point tool installs. Clarify cost models, test security for any agent or integration, and plan the work required to move pilots into broader adoption if outcomes justify it.

More context around this story.

Loading more related stories...

Keep reading in the app

Open the app view to save this story, compare related coverage, and continue from the same source.

Open in app