Town Hall: Strategic Plan Updates and Staff Engagement

Kathy Giacomini: Okay. Good afternoon, everyone, and welcome to our fall Town Hall of 2024. I am very happy to see people in the room, so we have a real in person meeting as well as all of you on Zoom. So today should be a fun day. I'm going over the agenda here. So first we'll start with [the] dean's introduction, and then Sharon's going to give us an update on our Strategic Plan, which is

truly is a living, breathing document for our school that we're using, and then we'll have a [talk] about a school of Pharmacy Faculty Council. One of the most important things we're doing this year is increasing engagement and belonging in our school, and that is near and dear to all of our hearts. So we'll hear about a new pharmacy staff engagement.

And then here comes the fun. We are going to have Joe Owens, who I assume is in the room, but maybe not. Maybe on Zoom.

Okay. Anyway, we will have. Joe is going to give us a demonstration of Versa, so that should be fun, and then we'll have an open question and answer time. So before I begin, I want to remind people that every three years UCSF does a Climate Survey. That's different from the Gallup Poll. We do the Gallup Poll every year which engages staff, which they're involved with.

But the climate survey involves our faculty, our staff, our students—everyone associated with the University, I think at UCSF Health as well as on the campus. So we like the one thing we like to report is that we do very well in terms of people at least taking the survey. So look out for that. I heard today it should be distributed sometime in January, and it's a 20-min survey only. So, not too long.

Okay.

Bob Day, as many of you are aware and read about passed away last fall. So we are

planning to do something in honor of Bob. Bob, as you know, was our resident historian, our resident clinical pharmacist, our resident mentor for all—for many of us over many years. So Sharon Youmans and LeeAnn Mutchler are organizing a short celebration of Bob's life, and we're going to do it on Alumni Weekend

which I think is around April 25 [and] 26, so we'll be having one there. We'll probably do it at a luncheon, and we'll have people speak in honor of Bob.

And now here's some really good news. Jonathan Watanabe. Our incoming chair of the Department of Clinical Pharmacy is weeks away from coming. He will be coming here on January 2nd. I'm very excited, Jonathan attended Sharon, and I

I think Sharon and I, Lawrence—a few of us were at ASHP in New Orleans yesterday, the day before, and we had a reception there a UCSF alumni reception Jonathan attended. I introduced him to the alumni. Oh, he really hit it out of the park. He's very friendly, open, and wants to meet all of you. So as soon as he comes he'll be making those meeting times, and I think it's going to be great.

I want to thank the search committee. They're probably not on this call, but some of you may be—for really doing a spectacular job in raking in some really good candidates, and we were lucky [and] fortunate enough to get Jonathan in particular. Thanks to Claire Brindis and all of you who served. And I also want to once again thank Jen Cocohoba for being the interim chair of the Department of Pharmacy. She, when I remember when

I remember when I asked Jen, I said, Jen, I'm going to ask you a favor, and I was expecting a big No, and she said, I will do it. And she did an amazing job, because usually when you're an interim, it's like I'm not going to be here. I'll leave that for someone else—not her. She took on problems, and she did a fabulous job.

I also want to welcome some new faculty in BTS. I don't have time, I think, to read everyone's name. I'll just leave it up there for you to see. We've got a variety of new—variety of different faculty that have come join BTS in the last—

year, really. I mean, Joanne's been here a while, but she didn't get her faculty position, I think, until this year. So all of them: welcome to our School of Pharmacy. And if any of you see them in the hall, or anything like that—make them feel welcome. You know how it is, or maybe you forgot, because I certainly forgot. It's been so many years since I transferred jobs. But I forgot how it's like to be new when you're new, just having someone say, "Hi, I'm so and so" really helps.

And then we've got new teaching faculty. Yessica Gomez, who will be in Pharm Chem, or is in Pharmaceutical Chemistry, as well as Tristan Storm. So

he's joining. I think he may have already joined. So we're going to have both of them around. They will be integrating the departments into the PharmD curriculum. That's their primary job. They'll also be doing other teaching. You know. They'll have other responsibilities, for example, teaching in our master's program. But their primary responsibility is integrate our basic science faculty members into our curriculum, which

has been a major gap for many years.

I also now need to welcome a lot of new staff. So here we all are. Is anybody in the room that's new and on this list? Oh, Alyscia's there. Katherine's there. So we've got new staff even here, and all of you online who are participating in this Town Hall. Welcome to the School of Pharmacy, and you can see all the different departments and the Dean's Office have new staff.

We've been working on a National Academy of Medicine workshop, and I'm happy to announce that the NAM has adopted it. We raised, I don't know, almost $400,000 to put on this workshop. It will focus on the pharmacy workforce. When the National Academy of Medicine speaks, Congress will listen, and I have the honor today of

participating in the first planning committee. They have a planning committee now, and they're scheduled it. Well, I'll tell you the date in a minute, but it's chaired, amazingly, by Jonathan Watanabe. He's chairing the planning committee, and so I had the honor today to speak to them. So I'm really looking forward to—. They were going to help tackle a lot of the different

different issues that we're all facing as a profession. And here's the date: May 29 and May 30. And I will try to support as many of you to attend as possible. We want to be there and voice

our enthusiasm as well as highlight some of the issues that we're all facing. The AI—the artificial intelligence and computational drug discovery and development program was launched this year, and they took applications for next year. We'll start another class in 2025, and they got over 150 applications. So they are really hitting it out of the park.

I think the judgment day will come when we see how they're employed, where they go. We're going to track those students and see what happens to them. But there's the cohort. And again, once again a big thanks to Joanne Chun.

Sharon will talk about possibly updates. And so the PharmTech to PharmD Pathway Program that Sharon is responsible is launching for is launching in February. So we are going to recruit our first farm techs.

They're going to start participating in the program. And the idea is we are going, we are going to upskill them. We're going to prepare them, and they will be applying to our PharmD program. And that will be just fabulous. And it's being funded by different people who are helping us out. And Igor is working very diligently on this pipeline program—I now call it a partnership program—where we've partnered with the minority-serving institution

UC Merced, and we're going to do a 3 plus 3 BS to PharmD program. And Igor is also meeting all his milestones. So that's going on. The Benet Symposium. We raised the endowed chair. We got the million dollars. We went over the top with mostly donations like crowdsourcing as well as some key donors who really

contributed to that professorship. So that has been a fabulous happening because we now have the first Benet professorship, and then the Val Louie awards. Those are awards given out by UCSF Health Pharmacy, and they usually give two awards. But this year we partnered with them, and we decided to expand the award. So a pharm—of course a distinguished pharmacy tech,

a distinguished pharmacist, and then many other awards were given, so we had a total of about, I think 8 awards were given out.

and then this just tells us how our research enterprise continues to grow. Jaime Fraser got one of the first ARPA-H large grants, I mean, in schools of pharmacy they're hardly there. And they're very competitive. And this is going to speed drug development with AI, and so that started this year, and so congratulations to the whole team.

And this is just reminding us that we had a wonderful faculty and leadership retreat, and that went on well, and Sharon will be telling you about some of the updates there on strategic planning. And now I'm going to go to a lot of awards, and I'm not going to mention one name. I'm just going to put it up and and say, give them five seconds, or else the timer is going to pull me off the stage. But we've gotten a lot of faculty awards in all the different departments.

Many of our staff have received SPOT Awards which is also fabulous, recognizing their contributions to the missions of the school. And here's some more SPOT Award winners. Congratulations to all of you, SPOT Award winners. Here's another group, one without a picture, and then

again, not mentioning any names, because I can't probably pronounce them—unless I practice—student awards. We have won a lot of student awards and two that are not up there right? Oh, yeah, they're up there, the last two that won at the ASHP meeting and final one

and ending with happy holidays. Did I make it or no? [laughs] Okay, who's coming up next? Sharon?

Okay, thank you.

That's a good lead.

Sharon Youmans: Well, good afternoon, everyone. It is my pleasure to talk about our Faculty and Leadership Retreat, and to give you some highlights of what went on in that retreat, realizing that not everybody was able to attend and also mainly focus on the outcomes for our Strategic Plan and how we're going to move forward in 2025.

So our strategic retreat—well, it was a Faculty and Leadership Retreat—took place in September, and what's important for all schools and all organizations is about the strategic plan. So here is just a quick reminder of our mission and vision. And we do have a website that contains our Strategic Plan. And I'll

hopefully be able to show it to you at the end. But these are our strategic plan priorities that you can see there in a different—color-coded research, education, patient care people, DEI, strategic support and transformative partnerships. So those were the areas where we focused the discussion of the retreat.

Some of the things that we did that were different and really well received is that we started off with what's called a fireside chat that occurred with the chancellor and the dean, and a lot of folks were surprised that the chancellor took time up to come visit with us, but it was really great, and he sort of gave his perspective on where AI was going on the campus. Kathy asked him questions, so it was a wonderful exchange, and then we opened it up to the audience.

Most of the day was spent in small groups, and going around looking at our different

posters, we decided that the strategic priorities probably were best presented in a poster form as opposed to having people getting up and talking at the audience, so that way folks could go to whatever areas they were interested in or tugged at their hearts, and ask questions, and look and see all the wonderful accomplishments that the working groups had did over the years. So that was very interactive. And we did that twice, once for the updates.

Just in general. And then 2 after they heard the speed talks on AI in 3 different areas. They went back to offer suggestions and feedback on that.

So, speaking of the 5-min power talk, so that was well received, and people thought it was very interesting, and they received a lot of information. This snapshot is of our faculty in the research area, so they gave 5-min of how they were using AI tools to help in their research area. So that was very good. And we also had, like 4 or 5 people talk patient care.

And then about 4 people talk on education. They were not only our internal faculty, but we also had invited guests from industry, some of our faculty who teach at our other sites, and from School of Medicine we had a person come and participate and talk about what they were doing. So it was

a lot of variety. Very interesting, people were up and down so they weren't sitting all day, and I think we'll try to keep that same pattern of making it as interactive as possible.

Of course, at the end we always have to have a reception. You have to have a glass of wine at the end of the day, and we had a poster session, and we invited the postdocs and PhD students, and even some of our pharmacy students, even though we knew that they were probably in class to present their posters and their research, and how they were integrating AI into those areas. So it was a great way to meet

folks, because I usually never get to see the grad students as often as I would like. So it was very nice. And again, another opportunity to interact with people that we usually don't.

So I'm not going to go over every word in these upcoming slides which were really to provide some of the suggestions. Now, these slides are going to be made available to everyone, because we really want this process to be as transparent as possible. So you'll be able to read the different suggestions as people [who] went around to the posters gave to the individual folks, and I have the co-leads listed.

So if anyone has feedback or questions, or want to learn more. They know who to go to, or if they're interested in being on the work group, they can contact those people. So again, we have Education which is already taken off

under the leadership, at least in the pharmd curriculum under Conan MacDougall, and coming up with policies. But how best do we try to train our students to our learners to use AI to help enhance the teaching, the learning, and everything under research.

Robin Corelli and Tanya Kortemme. And it's already happening in terms of research, and how they're using AI and Kathy just mentioned how one of our faculty just got a great grant on using AI to speed up drug discovery.

in patient care. Our colleagues, Lisa Kroon and Kathy Yang. So we had folks from Genentech. We had a participant from Kaiser. Come and share with what they're doing with AIs to try to figure out how can we enhance patient care, how we can use patient care information to help us make decisions, etc. How can we use AI to make routine tasks a little bit easier? And I think that cuts across

all the different divisions and activities that go on in our school.

for the people in DEI

sections. What we came out of that, and I had been thinking about it a lot. Is that for efficiency, and because we had an overlap that one of the things we're going to look, do, and for 2025 is combine those two. There are a lot of overlap

in terms of the goals and sub goals, and it's really hard to talk about DEI and not talk about people, and vice versa. So we're going to move to do that. And one of the comments that stuck out to me was this one here? There's a lot of mention of work-life balance yet that—it hasn't materialized or people haven't seen it. So we need to look into that to find out. What does that look like? What are?

I don't know who wrote that. But what would that look like? What does that mean? And I think that's a great area where we can even pull in more staff to help us with that.

Strategic Support and then Transformative Partnerships. You know. One of the things about the strategic plan being a living document is that you start one place, and then over time, you might discover. Oh, we need to be more clear about this. Or maybe we need to change the definition or be, you know, sort of maybe narrow the definition. So we can get our arms around the work. So especially in Transformative Partnerships. That's what we've decided to do. So where should our focus be

and then move from there. So you don't have to feel that you write something in 2022, and you're stuck with it for 5 years just because it's a 5-year-plan, but it's supposed to be a dynamic and transformative document, and it will teach us what we need to do and what we need to listen to, and that's why we have the retreats and have the feedback.

So the next steps for the strategic plan is that all of the co-leads have received this information, received the feedback. They're going to go back. Look at their work groups, do we? Who are we missing? Who should we add? They're going to look at their goals and sub goals. Which one should we delete because we're not never going to be able to do it? Or which one should we add and then prepare. The co-leads have been given the charge that they need to give us a report

out in February in terms of where they are, and our next retreat is scheduled for September of 2025, which will be here before we know it.

And I wanted to. And so we're gonna see how this works. I wanted to show you the

actual website. Yes, it's working.

that you can go to. And again, the link will be in the slides. And just for example. So here we have on the website, and we have basically made it available. See how come I can't

to. I want to go down. Oh, here we go. It's working. Let you know that we have on the website. We posted the posters. So if you want to go and read and read, you can't read it. Oh, they're not seeing that. Okay, why is it not showing?

Oh, I think it's because—I knew this was gonna happen.

Yeah. And I was afraid. Yep, yep, he's

she's so wonderful you never have to ask to help me.

But what really, what I wanted to do was to show you—excellent—where the information is from the strategic plan. We also have a strategic site a retreat site. But if you go to the strategic plan, this is where the posters are. And so you

oh, gosh! What did I do? And you can read more specifically in terms of what the goals and accomplishments were. For the 2023, 2024. What our challenges are, what our future directions. So please go to the site. Pick whichever one you want and read and ask questions. And we think that'll be fantastic. Okay, I'm going to get out of that.

Oh, perfect! All right so. And but I also put the link up to the retreat which has photos of all the people that were there. And it's really important because these slides and these posters serve as documentation of our progress in the strategic plan. So that helps us because it is an accreditation

requirement for the school. So with my 30 seconds left, we have a QR code that you can sign on to to provide feedback about what you just heard. If you have any information, questions, or advice that you want to give, please please

don't be shy. Also work through your CAOs, or whoever your staff managers are to get information, you can always feel free to reach out to me. I think most of you may know who I am, but I'm happy to answer. So with that, my time is up.

and thank you for your attention.

[…]

Oh, we do have 4-min for questions. Oh, okay, we do have some questions.

if there's any on Zoom, or if not.

yes, ma'am, yeah, Sharon, I realize that we're kind of, because [unheard]

let them know it's being updated.

Sorry, Susan, I didn't know. Okay, should I repeat it? Yeah. So I realize that for accreditation, one of the requirements is, each school needs a strategic plan, but when we update it, do we need to let them know that it's been updated? So the answer to that is no. So when we write our self study, which is not going to happen until 2028, we will have a trail of

documentation of updates. And what we changed and whatever. So that's what we need to do. Everyone knows that

the strategic plan is living, but ACPE doesn't need to be known when we make changes.

Question Paul, content?

I can. They're right now. We're.

I know, but I'd like them to know. Tell them sorry I'm asking Katherine to add one more thing.

Hi, there! I'm Katherine Tam, Communications Director. We are also launching a new Dean's suggestion box. And if you go to our website at pharmacy.ucsf.edu, and you scroll down to the footer. There will be a suggestions link where you can easily access it. It is an anonymous form. If you would like to provide your name and self identify, you can. But you also don't have to.

any questions on Zoom. We don't know. Oh, we do no questions on Zoom, maybe.

Sure. So I know that's well Joel is here, although I won't put him on the spot. But we're happy to say that for our PharmD application cycle that we have seen an increase that was above the national average. So we're very pleased about that. It's the talk of the town across the country about the declining, but there, it seems to be

somewhat of an upswing, and we've been fortunate that we've always had good people, they just seem to be less of them over the years. But we're we're encouraged, very, very encouraged. And I think all of these pathway programs are going to help. In fact, a couple of people applying this year

worth because we they heard about our PharmTech to PharmD program, and some of them were ready to go right away.

and some might take them a year or 2 years, or whatever. But that's the whole point is to do an assessment. They're somewhere on the continuum, and I told this one. Young man, I said, you're ready to go now. You don't need to wait for this program, and so he's applying.

Joel didn't want to say anything. Joel.

Okay, anything else?

Alright. Well, we will move on to the agenda.

Thank you. Thank you for your attention.

Thank you, Sharon. Next. I would like to introduce Kim Cantero, Jacqueline Fabias, Karen Hamblett, and Shondell Moody, who will be talking a bit about a new staff Council.

Kimberly Cantero: Hi! Good afternoon, everyone.

Kimberly Cantero: Thank you for joining us. Today we are excited to announce the creation of a new staff council within the School of Pharmacy.

Kimberly Cantero: The initiative is designed to bring together staff members from across the school, including Clin Parm, Pharm Chem, BTS, QBI and the Dean's Office.

Kimberly Cantero: The purpose of this council is to facilitate communication and collaboration, to support well-being, and to advocate for staff, to ensure that their voices are heard by school leadership.

Jacqueline Fabius: The Staff Council will serve as a platform to address various issues that staff members face in their day-to-day work by bringing these concerns to the attention of the school leadership, we can work together to find effective solutions. The council will also play a crucial role in proposing new initiatives and policies that can enhance our work, environment and improve overall productivity.

Jacqueline Fabius: Our goal is to create a supportive and inclusive atmosphere where everyone's input is valued.

Shondell Moody: To ensure diverse representation, staff members from each unit—Clin Pharm, Pharm Chem, BTS, QBI and the Dean's Office

Shondell Moody: will be nominated to serve on the Council. These nominations will be conducted within each unit, allowing staff to select their peers, who, they feel will best represent their interests and concerns.

Shondell Moody: Serving on the council is a valuable opportunity for staff members to engage actively in shaping the future of our school.

Karen Hamblett: And to keep everyone informed and engaged. The Staff Council will present updates to the CAOs on a quarterly basis. These updates will include the issues you've discussed with proposed solutions, any progress updates, and by maintaining this regular communication we aim to ensure transparency and accountability in our efforts. We encourage all staff members to reach out to their caos or unit representatives with any concerns or suggestions that you may have.

Karen Hamblett: and just to wrap up the creation of the Staff Council marks an important step towards enhancing staff involvement, creating a sense of belonging and improving our school environment, and we look forward to working collaboratively with all of you to address the needs of our staff.

Karen Hamblett: Together, we can make a positive impact on our school's community.

Karen Hamblett: Thank you for your attention, and we're happy to answer any questions you might have. I think we have some time right now.

Excuse me, we have a couple of holdovers from the last talk here. We had a question about the patient care slide.

And there was a correction there that it should have read PGx testing for pharmacogenomic testing. And then there was also a follow-up question about whether we could provide more specifics on the applications, but I thought I would toss those in, and a late breaking question, what kind of time commitment will serving on the Staff Council require?

Karen Hamblett: So I can take a stab at answering that. We're still sort of forming the details about how much time who the who it would be, and how it would all work. So

Karen Hamblett: the group can kind of come up with those suggestions together and bring it to the CAOs for the first meeting.

If there are questions in the room, feel free to raise your hand, and we'll bring a mic over. If you are joining via zoom feel free to type your question in the Q&A box.

Thanks, guys, for presenting. This is Kathy again. But thank you for presenting and announcing this exciting new Staff Council? It will be chaired. Is it chaired by Alesia, or is there a staff person chairing it, or how does it work? How will it work?

Jacqueline Fabius: I can take a stab, and maybe Karen can chime in as well. So we think staff will be will nominate someone to represent them.

Jacqueline Fabius: They will then meet with other people who have been nominated from different departments, and they would bring issues or solutions to the CAOs. The CAOs in turn would then discuss with Alesia.

Okay, alright, and when they meet as a council they'll meet like maybe monthly, or whatever they meet they meet as a council. Will. Will you guys be in the room? Will the CAOs be in the room, or it's just, oh, so just the staff. Okay, great.

Karen Hamblett: We're really viewing this as a way for staff to be more involved and also have leadership opportunities. So they might take turns coming to the CAO meetings. They don't have to all come at one time, but definitely still, and they have the opportunity to form the details about how they want to work as well. So.

One other question. Are there staff councils in the other schools, and will our staff council somehow meet with those staff councils? And then, you know, share best practices or share issues.

Kimberly Cantero: I'll take a stab at this one. I don't think there's a staff council that I'm aware of in other schools, although there is a staff council that represents all of UCSF to the larger UC campuses. So that could be another avenue that this council could provide input to. If there are issues in the School of Pharmacy that should be addressed

Kimberly Cantero: UC-wide.

Are there any additional questions?

Kimberly Cantero: Okay, great. If there are no other questions, I'd like to introduce our next speaker, Joe Owens, a product leader for enterprise AI infrastructure at UCSF. Who will provide a demonstration of Versa.

Joe Owens: Can you hear me? Anyone in the room. Hear me?

Hello! Yes, we can hear you.

Joe Owens: Okay, perfect. Sorry. That was long delay to get a response on that.

Joe Owens: wasn't sure if I was being heard. I'm Joe Owens. Thanks for taking time to learn a bit about Versa today from me. I am presenting on behalf of a very large group of folks across UCSF.

Joe Owens: All mission areas. We're calling. We've called ourselves since last summer the UCSF AI Tiger team, who is a collaborative group that came together to address the really once-in-a-generational speed of advancement in AI that has come about over the last couple of years.

Joe Owens: Now we're going looking back all the way to 2022 fall.

Joe Owens: Today in a brief time today I'm gonna focus mostly on some demos of the platform that we have called Versa. I'm gonna be introducing that, showing some how to get involved information there and then, if time allows or questions

Joe Owens: drift towards that. I can talk a little bit about the platform roadmap as well as some of the newer things we've introduced. I did. We work with the meeting organizers to come to select 3 specific demos

Joe Owens: for this group: recommendation letter drafting, summarizing information and generating interview questions. I think those are probably broadly applicable. But hopefully, as I'm demoing those we can highlight sort of what the thematic element to it is. Generative AI tools like Versa are incredibly generalizable to many, many different tasks. Anything that's really using

Joe Owens: written language. These tools apply to so the number of things you could use them for is is quite vast

Joe Owens: and really just being explored. So this is my outline in terms of what I want to talk about assuming no complaints about this agenda. I'll move forward

Joe Owens: Perfect.

Joe Owens: Just at the outset, I like to put this up here for anyone who wants to just dive in before they've even seen what I'm talking about. If you want to go ahead if you haven't gotten involved yet, and you want to get involved or learn more the first place to start is the ai.ucsf.edu site. That is now a centralized front door for many, many things. AI across UCSF.

Joe Owens: And one moment I'm going to move my presenter tools and apologies for presenting from the Powerpoint slide. I'm going to be moving back and forth into into the web browser, so let me know if this is not big enough. But I really encourage folks to come over to the UCSF AI site and specifically under Platforms, Tools, and Resources. There's a lot of information here, for example, last week the UCSF AI Scribe Program launched for ambulatory physician—ambulatory attending physicians.

Joe Owens: So that's here. There's a lot of information about Versa that's on this page, including how to request access to it, and how to get your own assistant built, or how to get access to Versa API, but many other AI platforms that are available at UCSF are available on this page, and then many materials are related to training.

Joe Owens: and others are, and ethical use of ethical and equitable AI use, which is UCSF is a leader on, is also on the site, so I encourage everyone to really start there. But then, as you dive deeper with

Joe Owens: generative AI community, there are every-other-week office hours as well as Slack channels and teams, channels depending on which platform you like to be on where there's a really a thriving community of AI practitioners at UCSF who post on a lot of the topics

Joe Owens: To dive into Versa there is a required 10-minute UCSF training on you on SumTotal. That's really the only prerequisite to get access to the versa platform at UCSF. And when you go through the process of clicking through like I just showed you, you would be encouraged to take that, and you'll be given access to Versa. Once you filled out a Versa request, and taking that training, and then the team can be reached at [email protected].

Joe Owens: So what is Versa? I'm not going to give a big background about what generative AI is, except to say that it's as big a paradigm shift in software, engineering, or or really compute computational power, or really any of those sectors. As it was when we switched from all the computers being in a big building, aka like a mainframe, to personal computers.

Joe Owens: And there were many people in the 70s and even 80s, saying, Why would anyone ever need a computer on their desktop? I can't imagine the demand for that. Who would want something like that? Generative AI is as big a shift. It's looking like it will be as big a shift as that shift. Many of the pieces are up in the air, and we're probably about to get into the trough of disillusionment, as they say. But for those that are following this closely, we anticipate that this will be quite a transformation in the technology sphere.

Joe Owens: So what what did we do? We brought Versa—

Joe Owens: the IT Tiger team built a platform to basically build a protected area for using generative AI. There's a lot of reasons I won't get into here about why you need to protect data and generative AI, and we built Versa chat as an alternative to say, a ChatGPT. And then we built Versa APIs for programmers and coders and data practitioners who want access to these models directly. And then we built a third product more recently, which is Versa Assistants. And that's a way to

Joe Owens: have the generative AI models and tools available as well as specific access to specific data that you can give them access to. And so that makes up for some of the weaknesses of these tools. But I'll explain more about that.

Joe Owens: So I want to jump into the demo really fast. And I always put, you know,

Joe Owens: a placeholder image here just in case we have any issues with access or anything like that, there have been a number of network outages this week. So hopefully, we'll not have any issues. This is the Versa window. So if you haven't been to Versa yet, you need to be on the VPN.

Joe Owens: And it's just versa.ucsf.edu, and if you go there and you haven't requested access or done the training, it'll just tell you, hey, it's forbidden, or if you're not on the VPN you won't get access, but if you have, you'll get this screen, and when you come into here you probably would get

Joe Owens: well, you probably get this as your default. And so this is showing that it's connecting to OpenAI's large language models, and it's showing you GPT-4o as the base model here. And just because I was playing around with it a little earlier, and to highlight that we're putting this out, we now have

Joe Owens: models—so there's another company that makes models called Anthropic, and we have this available through Amazon. So we now are dual-platform both through Microsoft and Amazon, which really is a big step forward. But I'm just gonna jump into a demo here.

Joe Owens: So one of the tasks that I mean, really any writing task can be accomplished in generative AI tools quite well, but one of the ones I was talking to the organizers about is writing letters of recommendation.

Joe Owens: And so one of the things you might

Joe Owens: have happen is, hey, you know, I need some advice about writing letters of recommendation. So a really good way to start with these tools is just to say that.

Joe Owens: I need some advice about writing…

Joe Owens: academic—it handles typos pretty well, by the way—academic letters of recommendation.

Joe Owens: and I'm gonna butcher some spelling there intentionally.

Joe Owens: and it'll think for a second. You know, these things are not instantaneous. It's doing quite a lot of compute in the background. So, having a little patience for it is worthwhile.

Joe Owens: And here we go. So it's going to stream these responses. And so what I wrote to it was this on my "You" line, and then it pulled from this very large language model, and when they say large language model, they really should say, like stupendously enormous, enormous. These have tens of, or hundreds of billions of parameters in them, and it gave me back this information. You know I've written some letters of recommendation before, and this looks pretty good, so I'm going to assume that it, you know, has a general sense of how to do this.

Joe Owens: So say I have an actual letter of recommendation I want to write, and I've gone ahead and pre-written up a little bit of—

Joe Owens: now I'm gonna expand this window a little bit—

Joe Owens: I've gone ahead and pre-written-up some information about in this case, Jane Doe, and I've I've noted that Jane is an independent research scientist. I've noted her research area number of peer reviewed articles, etc.

Joe Owens: and I've you know I can drop that. And then I would write something like, Please draft the letter

Joe Owens: for me, based on the following.

Joe Owens: And then you just hit go. It's really kind of wild. So I'm in a conversation with it. I started with some, hey? I need some advice. I have reviewed its advice. I could have gone back and forth asking it more questions about some of that advice if I wanted to. But for the point of this demo, I'm just gonna go forward and then I dropped in some more context about the specific person that I wanted to write letters of recommendation for.

Joe Owens: And then here we go.

Joe Owens: Here's its response. Here's the draft of the letter of recommendation, based on the points provided. It encourages you to put it on letterhead. It tells you the date, you know. It seems to know what it's doing. "Dear Search Committee…" and then it starts off in the first person, you know which is good practices. It mentions their full name in the first sentence, which is also good practices, and then it refers to them more formally as it goes forward, and then here's a letter, and then it encourages you [to] put your name, title, position, institution, etc. That's pretty good.

Joe Owens: I don't know if anyone has ever tried doing something like this in here. But one of the things that's really good to do at this point would be to say,

Joe Owens: please give me some criteria for how to best write such a letter.

Joe Owens: So I'm gonna

Joe Owens: ask it to give me some criteria to evaluate this letter based on basically, at least, that's what I attempted to do with this prompt here.

Joe Owens: And after that, it thinks for a second. What I'm going to do is I'm going to ask it to evaluate the letter it just wrote, based on this criteria. And this is a good way to get it to quote unquote, think about what it has just done. And so it gave me some criteria

Joe Owens: that it thinks are best practices, and I, without really reviewing that. But if I were doing this in real, I would review that more carefully, and I might revise it.

Joe Owens: Okay, now, please use those criteria to evaluate the letter you provided and provide

Joe Owens: for all of the criteria. It's good to tell it to be exhaustive and provide recommendations about how to improve any of

Joe Owens: the them that are deficient.

Joe Owens: And so here I'm continuing to reason with it, quote unquote reason. And so I've asked it to come up with this list firsthand that way. I could evaluate what it thought a good list was. I could just ask it to evaluate it, whether it was a good letter or not, but I've asked it to come up with what it thinks are a good criteria first, and I might refine those, and then I asked it to evaluate the letter based on that.

Joe Owens: And so it was showing me relevance, and so I might go back and say, Oh, it was a little lacking context. Before I use this letter, let me go add some more context. That's nice. It looks like,

Joe Owens: oh, academic potential or just potential for this person doesn't discuss their future potential. If I wanted to raise that a little bit more, right? So this is just an example of, you know, kind of

Joe Owens: jumping in to a task here. I didn't work a lot on revising these prompts, you know, to make them really good if I was doing this more robustly I might spend a little more time going back and forth and revising each of these. But this demonstrates the concept here.

Joe Owens: I'll pause for questions at the end. But I'll just do 2 more quick demos here. I went a little further on that one than I meant to. Another really useful thing you can do here, with these tools is, if you were, say, on a YouTube or a Zoom Meeting. Say, for example, these UCSF School of Pharmacy Class of 2026 White

Joe Owens: Coat Ceremony. And you wanted to get a summary of that. Say, you didn't want to. Listen to the hour and 6-min that this this video is, you could. And actually, this is kind of funny. I couldn't remember how to get a transcript in YouTube. And so I asked before this meeting, I asked the model for advice on how to find the transcript button. It's actually buried down here.

Joe Owens: And YouTube and a model gave me that. And then, after a little bit of selecting. It takes a little bit of effort. You can. You can pull everything out of the transcript. And I went ahead and threw that in a different

Joe Owens: document just for speed of this. But so please, or actually here.

Joe Owens: I'm going to give you a transcript from a meeting.

Joe Owens: Can you summarize the current script?

Joe Owens: It's from a YouTube recording.

Joe Owens: Please also use best practices of summarization in an academic medical center setting.

Joe Owens: And I'm not. I in like we did before.

Joe Owens: I would have maybe asked what it's criteria for being best practices. But I'm just gonna guess that maybe it has a sense of how I might want to do that. And so I told it what I'm gonna do.

Joe Owens: And then here you go.

Joe Owens: and I just paste that in there, and you can see that I pasted a pretty ugly thing in there, you know. It's even got applause, music, timestamps—things like that. I just pasted in very crudely.

Joe Owens: And it's gonna chunk on that for a minute. That's you know, a couple of 100 lines of text that I just threw in there, and it chunks on all that.

Joe Owens: and there we go. I don't know if anyone in this room attended that meeting and can verify. Note that everything about generative AI is probabilistic and not declarative. And so you always want to heed this warning down here that says, please evaluate responses. But if you wanted to, at least in a minute or 2, get a sense of what happened in that hour and 6-minute meeting. This is not a bad way to jump in.

Joe Owens: So you can see that really, what we're talking about here is information transformation. And, in this context, distillation. So if there's any large tranches of text that you need summarized, or at least a first pass, it summarized for these tools work incredibly well.

Joe Owens: I'm gonna start one more new chat real fast, which is

Joe Owens: I would like to generate some

Joe Owens: interview questions for a role. So I'm going to

Joe Owens: generate interview questions for a role.

Joe Owens: The role is

Joe Owens: …at UCSF…

Joe Owens: Here is the JD.

Joe Owens: And you know you go over to our our

Joe Owens: Brassring and you can. You can. You know it's kind of people. Ask me, you know. Can you just control all and pop, pop pop that in you generally? Can. You could go in here and be more exact and select everything. But these models are pretty good at ignoring page context. Unless there was something really wonky, you might want to just look at the whole page and make sure there is nothing too wonky that you're including

Joe Owens: and then

Joe Owens: I'm just gonna go ahead and throw that in there and just do it as a one-shot. So that's everything I gave it. I'm just scrolling around really fast, so you can see how much I threw in there. It's not very formatted, as you can tell. And my prompt was at the very beginning. It's gonna parse that out. And then it's gonna look through the rest of this.

Joe Owens: And so there are some interview questions

Joe Owens: based on this role. And I can tell that it's tailored this pretty well, because it's

Joe Owens: I might want to go check this. But if I was interviewing for this role, I imagine I would understand what USP 797 is—regulations. And I might say, you know, I don't know. I'm planning on interviewing somebody for only an hour. That's looks like good questions. But

Joe Owens: how about make that 3 to 5 more?

Joe Owens: Let's just say tighter, tighter questions

Joe Owens: covering the full range of the role.

Joe Owens: And you know, maybe it's good to have these 15 questions in your back pocket.

Joe Owens: But this is a faster, you know, a good way to to narrow something down. So I'm asking it for information, and then I'm asking it to distill even the information that it had.

Joe Owens: Now

Joe Owens: one of the things I might try to do here is, I might say, Hey, it'd be really good if I could refine this based on

Joe Owens: you know, some UCSF guidelines. For example, we have our PRIDE values. So I could say, go over to the UCSF website, and I could retrieve

Joe Owens: the information from the website. And I could

Joe Owens: overlay our values from this page onto these 5 questions.

Joe Owens: And I'm retrieving some information

Joe Owens: augmenting my prompt with that information. As you can see here, I just threw in that web page along with that, it's keeping the whole context of the whole conversation going here.

Joe Owens: and I've augmented that. And now, I'm getting generated generation that is based on that. And so basically, you know, I was not very fancy with my prompt. So it basically used our pride values as a very cut and dry rubric to organize those questions.

Joe Owens: So what I did there is

Joe Owens: exactly the same as what happens in the assistants that we have. We just have access to much larger amount of information. So I can. I can show you what that is. Or I can encourage you to go to our

Joe Owens: Versa wiki, and learn more about the assistance. So I I will actually pause for questions, because I think I'm at my time.

Joe Owens: but that is a very quick 3 demos of what's going on in Versa.

Hi there, we do have one question. Is there a fee for UCSF Versa Chat and API? It seems that information is required prior to provisioning access.

Joe Owens: Yeah, so on the learn more about versa. Page.

Joe Owens: There's an introduction to Versa, and Versa Chat is free.

Joe Owens: And it used to say it more cleanly at the top of that that it was free.

Joe Owens: I'm trying to remember where that went. We used to say that as a higher level, so I can see why you didn't find that Versa Assistants are also free, and Versa API is oh, here it's on the Versa pricing page. Sorry—Versa Chat is free, Versa APIs are subsidized.

Joe Owens: For those in the know, these models can get expensive, especially if you call the more recent Frontier models with large jobs. We've seen researchers run up several $1,000 in a couple of hours of charges. So we're subsidizing. IT is subsidizing each account. Basically, each person at UCSF who signs up for an API

Joe Owens: Versus API account at $200 a month.

Joe Owens: I'd say, 80% of users when they're first learning about it will not exceed that. It's when you go to do production-type jobs when you're doing big research jobs with it that you would start to exceed that. And you'd want to include that in your grant. Everything in Versa Chat is completely free.

Joe, thank you very much. This is Kathy. A great job. And it's very interesting to see the 3 demos that you gave us.

Yeah, I'm wondering why anyone would want to use Versa rather than well, if I'm using ChatGPT, or I have another chatbot, can you explain the advantages of just the Versa Chat over.

Joe Owens: So I'll tell you the 1st advantage.

Joe Owens: which is not an advantage. It will not keep up with the roadmap of ChatGPT. ChatGPT will always be fancier and be able to do much more than Versa, except

Joe Owens: any information you give out to the Internet. You have given to tech companies.

Joe Owens: So UCSF—

Joe Owens: I you know I kind of I don't know where I got this phrase, but we do not try to be good stewards of our patient data, our research data,

Joe Owens: our financial data. We are excellent stewards of that data, any information that I go to ChatGPT with and put it into ChatGPT, Google Gemini, any of these—Anthropic, Claude, any of those things, anything I drop in there? I have given away to that company by the terms of service you sign up for when you're using those things.

Joe Owens: There's various things you can do to make your accounts with them less—let's call it data retentive—but there's nothing you can do to basically not be giving away your data to those companies when you're using those tools.

Joe Owens: We're working with those companies to set up direct access to those. But we have not been successful thus far in contracting with any of the model developers directly in a way that we can have protected access to UCSF data or protected access to the large language models

Joe Owens: that would allow, which is the underlying technology here that would protect our data. We are working with OpenAI directly, and UCOP has an agreement with them, but they actually just recently were in breach of that agreement. So we're really happy that we're rolling out access to large language models and Versa to Amazon. Right now, currently, all of our models that we have access are through Azure

Joe Owens: which is a Microsoft product. So we're really de-risking our reliance on any one of these vendors. That would be the other benefit in Versa, but in terms of feature, rich and all of those things, you know, this interface is never going to be that fancy, because we're a small team. And we're developing this to basically meet the number one ask which was access to these large language models in a protected way. And so that's why you'll always see UCSF Versa—

Joe Owens: Sorry. That's why you always see UCSF. And our branding on anything that's Versa, and that'll tell you that you can use P3 and P4 data in Versa. Sorry, there's a very long answer. But it's an important aspect of Versa.

We do have one more question in the room.

Hi! My name is Kayla Mowry, and I'm a 3rd-year pharmacy student. This question is for Versa and also the School of Pharmacy. So I'm

I'm glad that the school is innovative in their technologies. However, noting that large language model data processing can be quite energy and water exhaustive. What assessments are being done to evaluate the environmental impact of us more heavily integrating AI into our work? And will there be transparency into how Versa impacts the School of Pharmacy's carbon footprint.

Joe Owens: Yeah. So to date. Well, I don't know that we have an exact carbon footprint for specifically Versa. But we do have reporting from Microsoft through Microsoft Azure and our cloud services.

Joe Owens: The amount of compute that's being used for these does exceed, say, a Google search. But it doesn't necessarily exceed similar carbon outputs, as you would have for doing those same tasks in terms of using the computer for the amount of time that you would have been otherwise doing it. So

Joe Owens: there's a lot of things being thought about in terms of the carbon footprint of these things as you start to scale them. So I would say at the early research phase where people are looking into how we might apply these, it's on the lower end.

Joe Owens: When we start thinking about having running always-on jobs where agents are doing evaluation on, say, all of the patient data that comes in all of the time and running multiple of those. That's where this is going to start to get large. And you're totally correct that it is a non-negligible component of the carbon footprint.

Joe Owens: The chancellor is looking into those issues and is meeting with the various exec leaders about AI every single month, and that is a line item that they're considering in all of those meetings.

Thank you so much, Joe. I know we're about—

Joe Owens: Yeah.

We're just at about time. So if if you are in the room and you have an additional question, feel free to write it down and give it to me. And if you are joining us online, you can type it into the Q&A box. What we'll try to do is, we'll try to respond to the questions via email and get back to you. And now I'd like to pass it to Dean Giacomini as we close out.

[…]

Yeah. So thank everybody for attending. And I want to thank Katherine, Sahru. Who else do I need to thank for? Organize Suzan for? And Frank for organizing this Town Hall. I also want to wish everybody happy holidays, and thank all of the speakers and enjoy the rest of the month. Thank you.