The State of Gen AI insiders Report FINAL 0724 - Flipbook - Page 12
Our roundtable agreed that companies will
need to build a new full architecture around
LLMs for developer use by the end of 2023. It
INFRASTRUCTURE
TOOLS
FOR FAST
ITERATION
INFRASTRUCTURE
TOOLS EXAMPLE
became clear in Q2 that companies do not
For companies developing chat use
even need a machine learning expert to
cases, models are being iterated and
extract the value of machine learning.
deployed every week compared to
Software developers can now build generative
previous applications being deployed
AI applications on their own by leveraging
every 2 months.
LLMs, thus bringing the cost of developing an
AI application down.
LLMOps has emerged in the last few months
as an evolution of existing machine learning
Tools that can help companies with building
operations (MLOps as a fast iteration tool
their generative AI infrastructure include
to enable software developers to build
LLMOps, observability platforms, and
applications on top of LLMs. Key to LLMOps
model tuning and training tools.
is ensuring that applications are reliable,
auditable and reproducible. Developers can
explore data and seamlessly experiment with
machine learning building blocks.
THE STATE OF GENERATIVE AI INSIDERS REPORT
12
