Neos Demands Plus on LawNext: Q&A Recap with Jonah Paransky

Assembly TeamNov 17, 2026

Bob Ambrogi had Assembly CEO Jonah Paransky on LawNext PR to talk about Neos Demands Plus, the AI that builds medical chronologies and drafts demand letters inside Neos.

The short version

  • Neos Demands Plus builds the medical chronology from the records already in the case, then drafts the demand letter from that chronology.
  • Every fact in both the chronology and the demand links to the page it came from, so checking the work does not take as long as doing it.
  • The demand letter follows your firm's own examples, so attorneys review a draft in the firm's voice instead of rewriting a generic one.
  • The chronology updates as new records come in, and it flags treatment gaps and missing records.
  • Jonah's view on PI firms: AI supports the contingency model instead of threatening it, which is why adoption is happening at firms of every size.

What did Assembly release?

Jonah framed it around where personal injury work actually goes. Two things eat the most time: understanding the medical facts of a case and writing the demand.

"It is not uncommon to see firms spend weeks on building a medical chronology, because there can be thousands of pages of medical documents you have to go through to understand what the fact pattern is of the case. And certainly it's always days."
Jonah Paransky, CEO, Assembly Software

Neos Demands Plus uses AI inside Neos to build the chronology and then the demand letter, so the firm reviews a draft instead of starting from a blank page. He called it the first Assembly product that takes on the work of the firm itself, supervised by the firm.

Why build it inside Neos instead of as a separate tool?

Jonah gave three reasons.

  • No integration project. Most PI firms do not have large IT teams to stitch together separate tools.
  • Adoption. Staff already spend their day in the case management system. "Our users sit in Neos for eight to ten hours every single day," he said. A new tool to learn works against the efficiency you are trying to get.
  • More of the case to work from. Standalone tools usually have you drag documents in and wait. Inside Neos, the AI sees every document in the case and the notes on it, and updates the chronology and demand when a new document arrives.

Medical records arrive as one huge PDF. How does it make sense of them?

The AI finds the documents in the case that contain medical facts, reads them (including running OCR on poor scans), pulls out the facts, and organizes them into a chronology visit by visit and facility by facility.

From there you can see things that are hard to catch by hand: treatment gaps, what the client was dealing with before the injury versus after, and questions about the care itself. Jonah noted that records often disagree with each other, such as a note that mentions an appointment you do not yet have records for.

He also made a point about which firms benefit:

"Almost all of them wish they did full chronologies on every case, but they're so time-consuming that often they don't have the time to give a case a full chronology workup."
Jonah Paransky

So it saves time for firms that already build chronologies and gives a chronology to cases that would not have gotten one.

What happens when more records come in, and how do you trust the output?

The chronology updates as new documents land in the case, and Neos can notify the team that it changed. Firms can choose how they want that handled. That matters when a motor vehicle case runs eight or nine months and a medical malpractice case can run several years.

On trust, Jonah was direct that dropping records into a general chatbot is not the same thing:

"Every fact has a citation that will take you right to a sidebar page that'll show you where in the document it came out, so that the act of checking the work doesn't now take you as long as it would have taken to do the work in the first place."
Jonah Paransky

Does the chronology tell you anything beyond the timeline?

Yes. Neos Demands Plus also builds a case briefing from the chronology, covering strengths, weaknesses and what you might still want to gather, to help the team as it prepares the demand and plans next steps.

How does it write a demand letter that sounds like your firm?

Jonah drew a line between the two outputs. People mostly agree on what a good chronology looks like. Demand letters come down to taste: some firms want three pages because they think adjusters will not read more, others want twenty pages with photos.

So an administrator at the firm loads example demand letters for each kind of case, such as motor vehicle cases in different jurisdictions or medical malpractice. Neos Demands Plus builds a template from them, and new drafts follow the firm's structure, focus and voice.

"They get a perfectly adequate average demand letter, but they don't get their demand letter."
Jonah Paransky, on what attorneys often get from other tools

How much attorney review is still involved?

Bob asked whether the process is iterative, with attorneys checking the draft against the source documents. Jonah said yes, and that the review is where the time savings show up:

"It can turn a half a day to a day to two days of demand letter writing into a 30-minute to 45-minute long review experience for the attorneys."
Jonah Paransky

Every fact in the demand has a citation that opens the source page with the fact highlighted, and the medical facts come straight from the chronology. He said the biggest factor in cutting review time was letting firms set their style, since otherwise attorneys spend their time rewriting the draft into their own voice.

What did the beta teach Assembly?

Assembly ran the beta as co-creation with customers and outside medical records experts. Jonah named the main lessons:

  • Demand letters have to match each firm's style, voice and structure. That was the biggest one.
  • Medical chronologies changed how firms understood case value, whether or not they built chronologies before.
  • Export format and how firms want to see fact patterns mattered more than expected.

He also described how fast it caught on. At some beta firms, within days of chronologies being turned on, hundreds of people were reading and working with them every day.

How are personal injury firms adopting AI?

Jonah's view is that the hourly billing problem large firms face with AI does not exist in PI. Contingency firms get paid on outcomes, so doing the work faster at the same quality lets them take more cases and run more profitably.

"AI tooling isn't an attack on the economic model of a PI firm."
Jonah Paransky

He said Assembly sees buyers across every firm size, from one attorney with three staff to firms with 500 to 750 employees.

What's next?

Jonah called Neos Demands Plus the first example of building with customers to improve outcomes for firms and their clients, and credited the beta customers for getting the product where it is.

Learn more

See how Neos Demands Plus builds the chronology and drafts the demand inside the case, or read the launch announcement.

Most Recent Blogs

Blog

Blog

Nov 13, 2026

The Reason AI Adoption Stalls at PI Firms (It’s Not the AI)

AI for personal injury lawyers pays off on records and demands, then stalls after the pilot. How to compare the options and get your firm past the stall.

Blog

Blog

Nov 10, 2026

Medical Record Review for Law Firms: Three Ways to Get It Done

Medical record review for law firms eats hours on every case. Compare review services, standalone AI, and review inside case management, plus what to check.

Demands Plus demand letter draft with numbered source citations and a source card linking a claim to its record
Blog

Blog

Nov 5, 2026

What Goes Into Personal Injury Demand Letters and Where AI Fits

Drafting a personal injury demand letter? The sections it needs, how the demand package fits, and what AI can draft versus what your attorney decides.

Blog

Blog

Oct 21, 2026

5 Questions to Ask Before You Bring Legal AI Into Your Case Files

Before your firm adds a legal AI tool to casework, ask where it works, where its output lives, and what it needs from your case file. Five questions to start with.