I'm playing some catch up today, but I'll get this going for you in the morning and we can hop on a call to discuss.
@Ryan No worries, this is awesome, thanks so much!!
• Call Center Modeling for LLM chatbot ◦ No financials ◦ LR Leads plus all billable ◦ Intake Template Q&A ◦ Goal: ▪︎ People ▪︎ Places ▪︎ What ▪︎ Why ▪︎ When ▪︎ How ◦ Private to CC & Ryan for now.
• MediLens usage research validation • Predator usage research validation
• MediLens ◦ PubMed ◦ NLM ◦ FDA ◦ JPML ◦ LexisNexis AI • Predator ◦ LexisNexis AI ◦ Google Search ◦ Duck Duck Go ◦ Google News ◦ Bing / MSN
• MediLens ◦ PubMed ◦ NLM ◦ FDA ◦ JPML ◦ LexisNexis AI • Predator ◦ LexisNexis AI ◦ Google Search ◦ Duck Duck Go ◦ Google News ◦ Bing / MSN
Hi @Ryan. Based on our call center data & LLM chatbot requirements, combined with the existing google cloud/bigquery setup, I think vertex ai agent builder would be the best bet. From what I've read, it integrates directly with current infrastructure, can query call center data natively, works with our looker dashboards, and it definitely gets us to a working solution fastest. I will get started on a few courses I found online to get up to speed on Vertex AI. It is scalable, so we can always migrate to vertex ai + RAG later if needed. LMK your thoughts
@CC Kitanovski, I love that. I'm Google centric already since they are winning the data game (Amazon trying to catch up) and lets move forward with it.
Sounds good, I'll get going on the courses after the meeting, put together some notes, and send them over when finished.
Morning @Ryan! I think it might be helpful for you to chime in on the LLM email chain and redirect based on business needs/end goal. Want to make sure we've got our tasks clearly defined for tomorrow, our plan for the week is set, and no one is wasting any time 🙂
Thanks, I was just wrapping up daily weekend financials. 🙂
LOL, I've been working weekends so darn long, I consider Friday nights my only real free time! 🙂
Siri, find the “that’s going to be a ‘no’ from me, dawg” meme.
(I type as I am also working on Sunday…)
@Ryan LMK what Cam says re: the EtO brief. PFAS & EtO are the two things that the chemist in me would be most excited to write briefs about... neither of which have been requested. I can also write one big brief that covers forever chemicals, in general 🤷♀️
Perfect, I meet tomorrow with him.
BigQuery Live Sheet showing billable statuses for Retainers we use in the Page 0, Real Time Retainers dashboard.
https://docs.google.com/spreadsheets/d/1972o3wcELj80ppsPG9Qvkf2lJ8JaGAeTH4QTFd59EEg/edit?usp=sharing
@Ryan I met with Brittany to review her current LR report generation process, then we walked through shieldlegal.dev, the LawRuler Questions Report, and LawRuler Scheduled BigQuery functionality. Then I met with Dustin on the details.
Required Features for exclusive use of shieldlegal.dev: • Column customization: Brittany's team needs the ability to select specific columns and control their order based on individual lawyer requests • Automated scheduling: Currently ~ 240 reports are sent out each Monday. The existing LawRuler report puller automatically generates the vast majority of these reports Sunday night for the team to finalize & then send out on Monday morning distribution, but shieldlegal.dev requires manual report generation—this volume is unsustainable for manual handling Development Timeline (per Dustin): • Column customization: 2 weeks (pending no critical issues) • Scheduling functionality: 4 weeks (pending no critical issues) Transition Constraints: Brittany cannot transition to shieldlegal.dev before September 8th due to: • AFFF filing deadlines taking priority • Overtime scheduling limitations (weekend staff cannot legally work >14 consecutive days) Idea: Can we implement staggered report delivery (Mon/Tues/Wed distribution) instead of concentrating all 240 reports on Monday? This would: • Reduce weekend workload pressure during the transition period • Alleviate overtime scheduling conflicts • Make the manual shieldlegal.dev process more manageable until automation is built The timing should work well. Dustin's features should be ready around when Brittany's team has bandwidth to handle the transition from their current automated system to the initially manual shieldlegal.dev process. LMK your thoughts.
@CC Kitanovski , this is a great summary. Thank you for this. I think we should definitely follow this transition plan. Dustin is already doing the above? And Brittany’s thoughts on your staggering pattern?
Dustin is confirmed on the timeline above. I triple-checked and will stay on top of it. Brittany seemed very relieved at the idea to stagger the reports, whether it was M/T/W or even just M/T. I briefly spoke to Carter yesterday before I left and he said there are a few law firms he thinks would be fine with getting their reports on Tuesday (woo!). I will talk more to Carter today to see if I can get a list of which ones may be most amenable.
Nice work. Anything to improve efficiency and effectiveness. Thank you.
Thank you, happy to help! I'll keep you updated on anything else I find. Let me know if there's anything else I can help with.
@Ryan I went onto FAERS and pulled the data manually to answer Cam's question the long, manual, human way using good ol' fashioned Excel 💀 Like James/you said, all lymphoma-related adverse events fall under the reaction group "Neoplasms Benign, Malignant And Unspecified (Incl Cysts And Polyps)".
I totaled all adverse events where the word "lymphoma" appears in the reaction name, including Cutaneous T-Cell Lymphoma, Non-Hodgkin's Lymphoma, B-Cell Lymphoma, etc.
So far in 2025, we've got 10 different lymphoma adverse events totaling 61 cases out of 86,898 total Dupixent-related adverse events. Within the 2025 "Neoplasms Benign, Malignant And Unspecified" category (646 total cases), lymphoma events represent 9.4% of all neoplasm-related adverse events.
If we use all available FAERS data (2017-present), there are 32 different lymphoma adverse events totaling 365 cases out of 358,343 total Dupixent-related adverse events. Within the 2017-2025 "Neoplasms Benign, Malignant And Unspecified" category (2,838 total cases), lymphoma events represent 12.9% of all neoplasm-related adverse events.
@Ryan I went onto FAERS and pulled the data manually to answer Cam's question the long, manual, human way using good ol' fashioned Excel 💀 Like James/you said, all lymphoma-related adverse events fall under the reaction group "Neoplasms Benign, Malignant And Unspecified (Incl Cysts And Polyps)".
I totaled all adverse events where the word "lymphoma" appears in the reaction name, including Cutaneous T-Cell Lymphoma, Non-Hodgkin's Lymphoma, B-Cell Lymphoma, etc.
So far in 2025, we've got 10 different lymphoma adverse events totaling 61 cases out of 86,898 total Dupixent-related adverse events. Within the 2025 "Neoplasms Benign, Malignant And Unspecified" category (646 total cases), lymphoma events represent 9.4% of all neoplasm-related adverse events.
If we use all available FAERS data (2017-present), there are 32 different lymphoma adverse events totaling 365 cases out of 358,343 total Dupixent-related adverse events. Within the 2017-2025 "Neoplasms Benign, Malignant And Unspecified" category (2,838 total cases), lymphoma events represent 12.9% of all neoplasm-related adverse events.
@Ryan The good news is... this validates the Medilens tool!!!! I also got 646 total anything-with-"lymphoma"-in-it adverse events for 2025!!!!!!!
@CC Kitanovski, that is excellent. Are your numbers indicative of a law suit or still same is true as we said before?
@Ryan That depends entirely on what I find when I search for published research the manual, long, human way. If I can't find a handful of papers in the last decade with an OR > 1.7ish, it would still be a "watch." But, the fact that other law firms are already advertising for these cases makes me wonder if they know something I don't. If yes, what do they know that I don't, and how did they find the information.
Am I allowed to pick Abe's brain on that and ask him if there are other things aside from published research and FDA adverse events that lawyers in this space are using to find cases & decide they're worth litigating?