August 3, 2026 - Issue #405

Hi, Charlie Uniman here, host of Legal Tech StartUp Focus ("LTSF"), the online community for everyone involved with legal tech startups. You're reading the latest digest of articles, opinion pieces, and other thoughts posted during the past week at the community.
​

If you enjoy reading this digest, please forward it to others with an interest in legal tech startups. Readers who aren't already members of the LTSF community and who wish to join can do so here. Please do send me feedback here with any questions, comments or other ideas for this digest. If you're not already a subscriber to this newsletter and would like to subscribe, please email me here to join the subscriber list.
​

Oh, and if you want to unsubscribe to this digest, you can do so by using the link in this email's footer.

Exit/M&A

◾ Legal IT Insider (LITI) has posted today, July 29, 2026, that Legora has announced it’s acquiring Wexler.
​
From the LITI post:
​
“Legora today (29 July) announced that it is acquiring London-based startup Wexler, which uses large language models to help lawyers extract and organise key data from documents relevant to their case. Wexler is used by some of the world’s leading litigation and disputes teams at Clifford Chance, Goodwin, and Herbert Smith Freehills Kramer, as well as in-house legal and compliance teams at large global enterprises. Its engineering team will form the founding team of Legora’s London engineering hub. This is Legora’s fifth acquisition in 2026.”
​
Read more from LITI at the link below about this acquisition:
​
https://legaltechnology.com/legora-is-acquiring-wexler-its-fifth-acquisition-in-2026/​

Fundraising

◾ Harvey closes another fundraise, as reported in a post by Legaltech News (LTN) today, July 28, 2026.

From LTN's post:

"Legal AI startup Harvey announced Tuesday that it received funding from Growth Equity at Goldman Sachs Alternatives and J.P. Morgan’s Growth Equity Partners. Financial details of the investments were not disclosed.

"Harvey will use the funding to bolster its growth and development, according to a press release."

Read LTN's full post here: https://www.law.com/legaltechnews/2026/07/28/harvey-announces-investments-from-jp-morgan-goldman-sachs-funds/?kw=Harvey+Announces+Investments+from+JP+Morgan,+Goldman+Sachs+Funds&utm_source=email&utm_medium=enl&utm_campaign=newsroomupdate&utm_content=20260728&utm_term=ltn&oly_enc_id=6788E2252056B4A&user_id=5a62392218ff43ec508b502b​

Honors and Awards

◾ Applications for legal tech companies to appear on-stage at the forthcoming TLTF Summit's Summit Sessions close TODAY, Friday, July 31, 2026. The Summit (November 16-18, 2026 in Scottsdale, Arizona) is one of the best legal tech conferences worldwide, and an appearance on one of the Summit Sessions' stages (Startup Showcase, Scale Stage, and Growth Track) is an unparalleled opportunity to appear before an outstanding group of legal tech innovators and investors.

Use the link below to visit the Summit Sessions page, where you can apply. Don't wait!! https://www.tltfsummit.com/the-summit-sessions​

◾ From the Axios "Pro Rata" newsletter on July 31, 2026:

"Discern, a Stamford, Conn.-based legal entity management platform, raised $10m in Series A funding led by Walkabout Ventures and Bungalow Capital." https://www.law.com/legaltechnews/2026/07/29/legal-entity-management-platform-discern-announces-10m-in-series-a-funding/​

◾ From the StrictlyVC newsletter on August 1, 2026:
​
“Vector Legal, a San Francisco startup founded this year that provides startups with AI-assisted legal services for company formation, commercial contracts, trademarks, and financing transactions, raised a $5.2 million seed round led by Base10 Partners. Business Insider has more [at the following link].”
​https://www.businessinsider.com/vector-legal-seed-funding-mitch-duncombe-y-combinator-2026-7​

LegalEd

◾ I've long believed that if one wants to get a well-grounded idea of what's going on in an industry (or any other domain), pay attention to what insurance companies are doing to provide coverage for that industry (or domain) - climate change being an example of one such domain. But we're here to talk legal tech, not climate change, so take note of this Artificial Lawyer (AL) post about "new model" law firm Crosby's insuring its LLM agents.

From the AL post:

"NewMod law firm Crosby is to provide ‘professional liability insurance for [their] agents, so that they can do autonomous legal work’, in what is an important step for the legal sector.

"We have seen moves to insure the outputs of AI before, e.g. Orbital and also Corgi insurance (see below), but this is taking things further with a focus on agents. Plus, the idea behind it is perhaps more radical than before in that it’s about allowing the AI agent to be treated as a human lawyer, and thus it needs to be insured as one. (Although, they have not outlined which insurance company will help with this, yet.)"

All of the AL post in question can be accessed here: https://www.artificiallawyer.com/2026/07/28/crosby-to-insure-its-agents-for-legal-liability/?jetpack_skip_subscription_popup​

◾ New LTSF'er, Santhosh Guntupalli, posts about combining (in a novel way) deterministic rule-based AI systems with learning-based pattern-matching LLM AI's. The "novel way" (which I refer to above) is "confining" the output that the LLM produces to an explanation (in natural language, of course) of the results of the output generated by the deterministic, rule-based AI.

"Where Should We Draw the Boundary Between Deterministic Reasoning and LLMs in Legal AI?

"Over the past year, I've been thinking about a question that eventually led me to build TriageCounsel.

"Should large language models be making legal risk decisions in the first place?

"Most legal AI products today follow a similar pattern:

"Upload a contract → let an LLM analyze it → return risks, summaries, and recommendations.

"These systems can be incredibly useful, but they also inherit the characteristics of probabilistic models. The same contract can produce different outputs, reasoning may not always be transparent, and explaining why a particular conclusion was reached can become difficult.

That led me to explore a different architectural question.

"What if we separated decision-making from language generation?

"Instead of asking the LLM to determine legal risk, our approach is:

• A deterministic reasoning layer evaluates the contract.
• Every finding is tied to explicit evidence and predefined rules.
• The LLM's responsibility is limited to explaining those findings in plain language.

"The objective isn't to replace AI.

"It's to place AI where it creates the most value.

"In this architecture, deterministic components provide:

• Reproducibility
• Auditability
• Evidence-backed reasoning
• Policy enforcement

"The LLM provides:

• Natural language explanations
• Communication
• Context
• User experience

"This separation raises a broader question that extends well beyond legal AI.

"Could this architectural pattern apply anywhere decisions need to be:

• Explainable
• Auditable
• Reproducible
• Trustworthy

"Healthcare, Finance, Insurance, Compliance, Government, Enterprise AI.

I'm increasingly convinced that the future discussion isn't simply "LLMs versus rules."

It's about where each belongs within the overall architecture.

"I'm still exploring this question, and I certainly don't claim to have all the answers.

I'd be interested in hearing how others in this community think about the boundary between deterministic reasoning and probabilistic AI.

"Where would you draw that line?

"— Santhosh Guntupalli​
"Founder, TriageCounsel.com​

◾ In his post today, August 2, 2026, on his Legally Disrupted blog, Zach Abramowitz discusses an episode of the "Zach Amramowitz is Legally Disrupted" podcast where he is joined by podcast guest Batya Nadler, a lawyer who left an eighteen-year career in BigLaw to become a partner at Norm Law.

From Zach's post:

"So what does that pitch [to leave BigLaw and join an AI first law firm] actually sound like? What is it like working at Norm? What does she [Batya Nadler] recommend to attorneys about improving AI fluency? And what has convinced her that firms like Norm entering the AmLaw 100 is inevitable? Check out the full conversation with Batya and please like, subscribe share and all that good stuff."

Read Zach's complete post at this link: https://www.legallydisrupted.com/p/norm-law-partner-ai-first-firms-joining?utm_source=post-email-title&publication_id=154342&post_id=209495809&utm_campaign=email-post-title&isFreemail=true&r=dm1ui&triedRedirect=true&utm_medium=email​

Member Introductions/Questions

◾ New community member, Santhosh Guntupalli, introduces himself to our community:

"Hello everyone!

"I'm Santhosh Guntupalli, founder of TriageCounsel, and I'm excited to join this group. My background is in AI and data engineering, and over the past year I've been researching how deterministic reasoning and LLMs can work together in high-trust domains like legal AI.

"That research led me to build TriageCounsel, where we're exploring a simple architectural principle: deterministic rules make the decisions, while AI explains those decisions in plain language. The goal is to make contract analysis more transparent, reproducible, and auditable rather than relying solely on probabilistic outputs.

"Beyond building the product, I'm interested in the broader questions around hybrid AI architectures, symbolic reasoning, knowledge representation, and where deterministic systems fit alongside LLMs. I recently had a fascinating discussion with Charlie on these topics and am looking forward to learning from the perspectives in this community.

"I'm here primarily to learn, exchange ideas, and contribute wherever I can. I'd love to connect with founders, lawyers, legal operations professionals, researchers, and anyone thinking about the future of legal technology.

"Looking forward to the discussions! — Santhosh"

Partnerships/Business Development​
​

◾ Interesting "design-driven" partnership between litigation workflow startup Advocacy and complex litigation law firm Scott & Scott, as reported in a post by Legaltech News (LTN).

From the LTN post:

"Litigation startup Advocacy AI announced Wednesday that it had launched a design partnership with Scott+Scott, a New York-headquartered firm with roughly 150 attorneys that specializes in complex litigation. The formal design partnership continues a close working relationship between the firm and the company that has existed since shortly after Advocacy’s launch.

"Under the partnership, all of Scott+Scott’s attorneys will be provided with access to Advocacy’s platform, and the firm and company will work together to build customized solutions for the firm’s antitrust and securities practices. The customized solutions will be built on Advocacy’s artificial intelligence-powered platform, which is designed to preserve and leverage case information and context across the life of a matter to streamline discovery and drafting tasks."

Access the entire LTN post here: https://www.law.com/legaltechnews/2026/07/29/inside-scottscott-and-litigation-startup-advocacy-ais-design-partnership-/?kw=Inside+Scott%2BScott+and+Litigation+Startup+Advocacy+AI%27s+Design+Partnership&utm_position=1&utm_source=email&utm_medium=enl&utm_campaign=morningupdate&utm_content=20260730&utm_term=ltn&oly_enc_id=6788E2252056B4A&user_id=5a62392218ff43ec508b502b​

Product Development

◾ Artificial Lawyer (AL) has posted today, July 28, 2026, about contract analysis company LegalOn's ambitious program for building LLM prompts for its customers.

From the AL post:

"LegalOn, the contract review specialists, have rolled out a library of 100-plus AI workflows, all available within its platform and aimed at inhouse lawyers.

"The lawyer-built collection, branded as Prompt Workflows, covers areas such as compliance, M&A, litigation, IP, corporate governance, legal research, and legal operations; and with every workflow available in US, EU and Singapore variants."

Read all of the AL post right here: https://www.artificiallawyer.com/2026/07/28/legalon-lets-100-ai-workflows-bloom/​

◾ Legaltech News (LTN) has a post updating readers on Thomson Reuters' progress in developing a legal domain-specific LLM called "Thomson." From the LTN post:

"Thomson Reuters provided an update Friday on the development of Thomson, a proprietary, domain-specific large language model (LLM) the company is building for use in its artificial intelligence-powered tools.

"Thomson is built on open source LLMs and trained with content from WestLaw, Practical Law, Checkpoint and Reuters. Thomson Reuters said it has evaluated Thomson against general purpose LLMs for use in specific domains, including legal, tax and accounting, as well as on more general tasks. The model’s development has cost roughly $20 million in compute cost to date, and it has been trained on less than 10% of Thomson Reuters content so far."

Read the entire LTN post at the link below (where, among other things, you'll get an idea about how Thomson rates on various benchmark tests.

​https://www.law.com/legaltechnews/2026/07/31/thomson-reuters-provides-benchmarking-data-for-forthcoming-llm/?kw=Thomson+Reuters+Provides+Benchmarking+Data+for+Forthcoming+LLM&utm_source=email&utm_medium=enl&utm_campaign=afternoonupdate&utm_content=20260731&utm_term=ltn&oly_enc_id=6788E2252056B4A&user_id=5a62392218ff43ec508b502b​

◾ Read this week's legal tech rundown from Legaltech News (LTN) where you'll learn of many product launches, a fund raise and an announcement by Theo AI of the appointment of a new member to its Board of Directors (shameless plug: Theo AI's CTO and co-founder, Tiago Luchini, and Head of Legal Product, Sarah Johansson, recently appeared as guests on the Legal Tech StartUp Focus podcast (episode show notes and link accessed here: https://network-295075.mn.co/posts/purpose-built-ai-tools-that-genuinely-help-litigators-legal-tech-episode-87-of-the-legal-tech-startup-focus-podcast-a-conversation-with-tiago-luchini-and-sarah-johansson-of-theo-ai?utm_source=manual)).

LTN's legal tech rundown post can be read in full at this link: https://www.law.com/legaltechnews/2026/07/31/legaltech-rundown-copysight-secures-3m-seed-round-ogletree-deakins-launches-compliance-calendar-and-more/?kw=Legaltech+Rundown:+CopySight+Secures+$3M+Seed+Round,+Ogletree+Deakins+Launches+Compliance+Calendar+and+More&utm_source=email&utm_medium=enl&utm_campaign=afternoonupdate&utm_content=20260731&utm_term=ltn&oly_enc_id=6788E2252056B4A&user_id=5a62392218ff43ec508b502b​

Purchasing/Using Legal Tech

◾ Legal IT Insider (LITI) posts about Intapp’s Celeste “business of law” AI agentic tooling becoming generally available to its customers. As explained in the post, this set of AI agents is aimed at improving firm-wide business-related workflows such as conflicts checking, lateral hiring, and client onboarding generally.
​
From LITI’s post:
​
“Intapp this month announced that its agentic ‘AI coworker’ Celeste, previewed in February, is now generally available. The listed company positions Celeste as part of a new category of ‘firm AI’, focused on improving law firm performance rather than individual lawyer productivity.
​
“Celeste sits in the business of law space that Intapp has operated in since its inception, helping firms with deal screening and conflicts clearance, to business development and lateral hiring.”
​
Read the entire LITI post here: https://legaltechnology.com/intapp-bets-on-firm-ai-with-general-release-of-celeste/​

◾ Legaltech News (LTN) has put up a post today with a helpful analysis of the hot topic of Chinese-based open-weight LLM's and what those models can mean for legal tech (particularly when it comes to LLM use and development by law firms themselves).

From the LTN post:

"The promise of open-weight models lies in more than just their top line performance. Because users can download these models and adjust their parameters, they are able to customize them for specific domains like the legal industry."

Further to the above, and as noted in the LTN post, open-weight models are often cheaper to operate than US foundation models. Moreover, the customization of the open-weight models includes not only the ability to "tune" them for the legal industry domain, but also (if a law firm has the talent and funds to do so) to tune them on a law firm's own proprietary data.

Read the entire LTN post here: https://www.law.com/legaltechnews/2026/07/27/what-chinese-open-weight-ai-models-mean-for-legal-tech-/?kw=What+Chinese+Open-Weight+AI+Models+Mean+for+Legal+Tech&utm_position=1&utm_source=email&utm_medium=enl&utm_campaign=morningupdate&utm_content=20260728&utm_term=ltn&oly_enc_id=6788E2252056B4A&user_id=5a62392218ff43ec508b502b​

Startup Management

◾ Bob Ambrogi over at the LawSites blog has posted today, July 30, 2026, about "[t]he Q2 2026 Legal Tech Buyer Intelligence Report was produced by FlyTech, a marketing agency specializing in advertising for legal vendors, in partnership with LawSites."

From Bob's post:

"In April [2026], I reported here on a dramatic first-quarter 2026 increase in the numbers of legal tech demos law firms were scheduling, with the numbers going up across nearly every major practice area by as much as 40% to 50%.

"Now, it appears that trend has reversed, according to a newly released report on legal technology buying behavior during the second quarter of the year, suggesting that the first quarter’s surge in attorney engagement was driven largely by seasonal evaluation cycles rather than a permanent shift in demand.

"But the report also introduces new buying-intent data that tells a more granular story, indicating that demand for legal technology is not evenly distributed across the market. It is heavily concentrated in a handful of product categories and practice areas – led, by a wide margin, by personal injury firms."

Head on over to LawSites by following the link below to read the complete analysis of the Report that Bob has on offer. It's very much worth reading.

​https://www.lawnext.com/2026/07/q2-report-on-legal-tech-advertising-finds-q1s-buying-surge-was-largely-seasonal-but-reveals-where-demand-is-now-concentrated.html​

◾ Startup investor, NfX has a post about how to run startup board meetings that, to put it bluntly, "don't suck." From the NfX post:

"Most board meetings suck.

"They are too long, too often, and a total waste of time for founders, the team, and investors.

"This is why they have a bad rep. And many investors have argued that even having a board at all for an early-stage company isn’t worth it.

"We disagree: if you do a board meeting right, it can be one of the best exercises you do as a company."

And even more from the NfX post:

"There’s a template for board meetings that I’ve seen work extremely well. If you run it this way, you’ll actually learn something, get high-leverage help, and keep everyone engaged.

"Here’s how to run it"

Read the complete post here to see the template in its entirety: https://www.nfx.com/post/how-to-run-board-meetings-early-stage?utm_campaign=NFX%20Newsletter&utm_medium=email&_hsenc=p2ANqtz-9yZJvSw55fgfC0_M54kT6KQgruynOA2mvtvQtLO7izfRJTr16MYMQg8eaqxx0yCh-q2oCuqzXNvelxp6SBZ2f-95IngQ&_hsmi=430868840&utm_content=430868840&utm_source=hs_email​