Harvey is an AI-powered legal platform designed to significantly enhance legal professionals' efficiency and accuracy by automating complex tasks, analyzing vast datasets, and assisting with document drafting and review.
Mind Map
Click to expand
Click to explore the full interactive mind map • Zoom, pan, and navigate
hello everyone and welcome to the lth
product briefing I'm delighted today to
be joined by Jake weiner who is the head
of strategic business development at
Harvey Jake welcome thanks Nikki yeah
great to be speaking with you and
excited to show the Harvey platform a
little bit I think everyone is excited
to see this Harvey hard to believe given
people still feel that chat GPT and geni
and legal is recent but in fact Harvey's
now been around around for about 2 years
how has it evolved where is Harvey now
yeah probably the best way to answer
that is just to jump into the demo so
want to show Harvey's fault feature so
this is uh Harvey's sort of data room
tool this allows you to upload data sets
of thousands of documents and the
documents are then stored within Harvey
and you're able to then query across
them both in natural language as well
well as with dedicated purpose-built
workflows and tremendous tool for due
diligence transactional due diligence
for litigation Discovery and litigation
review tasks as well as for tracking
sort of precedents and deal points for
in-house teams particularly useful for
tracking agreements whether those are
commercial agreements or just agreements
with various sort of counterparties so
there a hugely valuable tool for working
with large legal data sets you to use
this you create a project and you can
see examples of some different projects
here and then you can jump into any one
of these and perform a query so let
start by just showing an example of one
of these so here we have a data set of a
number of commercial contracts and the
query that I I previously ran here very
simple query I just said for all these
commercial contracts tell me about the
parties any change of control Provisions
termination and red flags generally so
not particularly on this red flags point
a lot of specificity there so like
leveraging the model's knowledge to to
be able to produce this here we prompted
Harvey to produce this answer in a table
format and so you can see Harvey's gone
through every one of the agreements in
this data set it's produced this answer
in this table format specifying each of
those terms that we asked about we can
then open up any of these agreements and
we can jump directly to the terms that
we've asked about and see them
highlighted in the documents as well as
sort of Harvey Zone kind of analysis of
the terms that we asked about oh I see
there's red flags there that's
interesting are you able to customize
the data that is extracted from a batch
of documents so if you wanted say 25
very specific data points extracted you
can specify that absolutely and I can
show you exactly what that looks like
here's where you would run an open-ended
query like this you can have Harvey give
you an answer either in a table format
like we just showed or in more of a memo
format and so here if we want to get a
table you start by prompting in entirely
natural language and again you can ask
about anything you like I can say
something like tell me about the control
Provisions any supply chain risks
assignment and anything that would
implicate Georgia law right any sort of
like any term that you might happen to
care about and when we hit ask Harvey
you'll see that what it's actually going
to first do is take a a pass at
producing The Columns that that you're
likely to care about so you'll see here
it's a it's added a number of columns
related to two columns around change of
control you have the ability to edit
those so you can choose exactly what
each of these columns should should say
but again are there any provisions or
Clauses that implicate Georgia law what
are the details of those so so Harvey
gives you a sense of exactly what's
going to be in that table and entirely
flexible entirely open-ended and of
course this is powered by our large
language models so this is not using
conventional machine learning or pattern
matching it's not looking for Section
heading it's not even looking for
specific words it's actually reading the
context of the entire document to
produce these outputs and even once
you've even once Harvey's produced an
answer like this you also have the
ability to add columns right click this
button here we can add columns and add
files and then of course you have
sophisticated filtering tools so we can
filter these down and so great way to
like rapidly work through large data
sets of documents and perform these
sorts of diligence tasks okay the
example of prompting that I just showed
you is the way that you can use this
through open-ended natural language
prompting but as I mentioned we are also
building and releasing on an ongoing
basis sort of purpose-built endtoend
agentic workflows within Vault and also
throughout the Harvey platform generally
many of these are Under development
right now but within the assistant tool
within the various research tools there
are more and more sort of endtoend
workflows that don't require prompting
in the same way so you can see some
examples of these here here for instance
we have workflows provided around
analysis of Court opinions producing
email chronologies in for instance a
litigation context reviewing lease
agreements reviewing share purchase
agreements if we select one of these
create a do an LPA review or fund review
we click this Harvey specifies okay
these are the terms that you're likely
to care about you also have the ability
of course to edit any of these particular
particular
Concepts but it's setting it all up for
you you choose the documents within the
data set that you want to run this
across and then you essentially hit go
and we won't wait here while this runs
but this sort of gives you a sense of
where the Vault platform is going and
also where the Harvey platform more
generally it's going it's actually
running here but where the Harvey
platform is going more generally which
is more and more of these agentic
workflows that sort of remove the
prompting burden from the user and that
are purpose built for specific legal
tasks um conduct those tasks at really
high levels of fidelity large numbers of
these workflows are being loaded into
the Harvey platform and are just there
when you log in we also build these on a
custom basis for for customers and as
our models and LLS generally have gotten
more and more sophisticated we're able
to produce these workflows with really
pretty light engineering lift and we're
able to plug these in for customers on a
bespoke basis without a lot of obstacles
to to doing so and for those watching
who are not sure what the distinction
might be between prompting the assistant
tool or clicking a button on an agentic
workflow would it be correct to say
you're seeking a specific output you
click the button for that output and
then what Harvey does is take the
necessary tasks in order to achieve that
output or that outcome yeah that's right
that's right and I think the other way
to to think about it is many of the
tasks that that the workflows that we're
building accomplish are tasks that are
already possible in the Harvey platform
right now via open-ended prompting but
they might require very sophisticated
prompting or they might require like
multi-steps of prompting so running one
prompt taking that output then running
another prompt on top of it and then
doing that several times so what these
workflows are doing is essentially
putting together very sophisticated
prompts in addition to a software layer
um but very sophisticated prompts and
then chains of prompts so that as a user
you don't have to go through that that
process to get these sorts of outputs
got it thank you that's super helpful
great okay so next want to just mention
a few of these Harvey again like
workflow tools these are slightly
different so these are tools that again
are like purpose-built for specific
functions certainly won't dive into all
of these but tools for review of
litigation transcripts so trial
transcripts or deposition deposition
transcripts document comparison
substantive comparison of documents
whole document translation and then
certainly want to highlight this Redline
analysis tool so I'll jump into this
here this allows you to upload Redline
documents either PDF redlines or word
track changes and have our Harvey either
produce an issues list for you in the
pre-specified format again with really
no prompting or you can query this
generally here um some an amended set of
corporate bylaws and then for the prompt
just said you you represent a
prospective investor analyze these
changes flag any that might be of
concern from an investor standpoint and
you'll see that Harvey has done that
here it's extracted the changes that are
likely to be of concern for an investor
and then it's going to site back
directly to to the language within the
document here good example of sort of
the legal specificity of the Harvey
platform in terms of the ability to read
and analyze red lines okay great the
kind kind of like last point that I do
want to mention here just within the
Harvey platform before we jump to the
Microsoft Word ad in is the library
feature so we've talked quite a bit
about prompting and workflows and things
like that we are again building more and
more of these agentic workflows but
prompting and the flex ability and the
power that prompting offers with
Harvey's models is we think really
important and is always going to be a
part of the the platform and a lot of
folks are interested in understanding
how they can get better at prompting and
learn how to prompt and that sort of
thing we provide
and examples Library here we're
constantly updating this and you can
filter this across practice areas and
it's so great way to get a sense of how
to use the Harvey platform you can open
up any of these examples they include
sample documents and gives you a great
sense of how you can be using the
platform may not have thought about
using Harvey to produce
cross-examination questions in an expert
report but rvey will do that for you so
great way to get a sense of how to use
the platform and then we also do this at
the level of the actual interface so you
can load prompts and templates and you
can do this both using sort of Harvey's
own specified prompts and templates or
you can create prompts and templates
that are specific to your organization
or your practice area and so also I the
Harvey prompts the pre-loaded prompts
are such a great training tool actually
for people in how to actually prompt
properly and what good prompts might
look like I think that's right yeah yeah
you can see even the way that we lay
some of these out they're bracketed for
the user to be able to go in and edit
them a little bit so yeah they're you
can think of as helping you get to that
first step right they're like precedent
yeah so it does help you track usage and
so on in the platform from an
administrative perspective but let's
jump in and look at the at the word
addin yeah definitely yeah of course you
have the ability to track your history
and I'll just mention to like very
sophisticated controls from an admin
perspective in terms of data retention
ability of admins sort of review usage
statistics and data across the
organization we offer a client matter
number tracking and sort of
administrative tools for folks who need
to manage the use of AI tools for
specific client matters lots to be said
on like the administrative front but we
can save that for for a deeper dive okay
let me change my window here and we will
show the Microsoft Word addin so the
addin which you can see here on the
right hand pane allows you to language
directly in a Microsoft Word document so
the way that this works is hypo there
the way that this works is by letting
you select the language that you're
interested in editing and then you can
edit it in two different ways you can
either provide instructions to the
models or you can provide precedence we
think about this the same way that you
might give instructions to one of your
colleagues you might just tell them like
hey add a a particular term or make this
particular term more favorable to one
party or the other or you might actually
just give them a precedent and say hey
so you can see this kind of an action
here so I'm working in a credit
agreement I've selected the ma
definition starting with the instruction
based editing I can just say something
like make this more and you'll see that
Harvey's going to make suggestions on
the ways to revise this language you can
review these different suggestions and
see which ones you'd like to apply let's
see include an adverse effect resulting
from under any other material agreement
great we hit apply and Harvey's going to
go ahead and then make those changes in
track directly in the document again
this is powered by Harvey's legal
specific models so you'll notice that
these suggestions are I'm really quite
sophisticated and carefully tailored to
the content of what you're trying to
edit here so that's the way that the
instruction based editing works I'm
going to go ahead and just reject those
changes just so we have a blank slate
here and then I'll show the
precedent-based editing so we just click
over to precedent based here I've
uploaded another credit agreement as as
a precedent and we've just highlighted
this Mee definition which we want to
edit just hit make suggestions and
you'll see that what Harvey's doing here
is it's reading this precedent document
it's identifying the relevant language
in that precedent and then it's going to
make suggestions here based on the
precedent to bring this into Conformity
for instance specify that the financial
condition refers to financial condition
determined in accordance with gap great
we hit apply and Harvey will make those
edits in track changes here and again as
always you have the ability to view
references so you can go back and check
within the Press of where these
suggestions are coming from that's great
Jake thank you so much for showing us
all of this this is fantastic I think
for a lot of people would be really eye
opening just to see the breadth of
Harvey and the various problems that it
really can solve or the ways that it can
help lawyers I know that's really the
focus is on helping lawyers it really
feels like that so thank you so much for
sharing with us today it's great to hear
I'm an attorney myself we have many
attorneys here at Harvey as we've talked
about and so we're very passionate about
anything that we can do to help improve
the profession and improve the value
that the work that attorneys are doing
glad to hear that great we will put a
link at the bottom so anyone watching
can reach out to Jake and his team at
Harvey If you want to learn more Jake
thank you again and to everyone watching
thanks so much for coming along to the
LT product briefing and we'll see you
Click on any text or timestamp to jump to that moment in the video
Share:
Most transcripts ready in under 5 seconds
One-Click Copy125+ LanguagesSearch ContentJump to Timestamps
Paste YouTube URL
Enter any YouTube video link to get the full transcript
Transcript Extraction Form
Most transcripts ready in under 5 seconds
Get Our Chrome Extension
Get transcripts instantly without leaving YouTube. Install our Chrome extension for one-click access to any video's transcript directly on the watch page.
