The Keys to the Secret Garden: Unlocking the Potential of AI in the Enterprise

At this point, enterprise generative AI applications for  u003ca href=u0022https://github.com/features/copilotu0022u003eu003cspan role=u0022linku0022u003esoftware developmentu003c/spanu003eu003c/au003e, u003ca href=u0022https://www.neuronsinc.com/u0022u003eu003cspan role=u0022linku0022u003emarketingu003c/spanu003eu003c/au003e, u003ca href=u0022https://fortune.com/2023/09/20/morgan-stanley-ai-assistant-answer-investing-personal-finance-queries/u0022u003eu003cspan role=u0022linku0022u003ecustomer/employee serviceu003c/spanu003eu003c/au003e , and u003ca href=u0022https://www.framer.com/aiu0022u003eu003cspan role=u0022linku0022u003eproduct designu003c/spanu003eu003c/au003e are proliferating as the standout use-cases early adopters are focusing on. From our position working with clients to develop and enable u003ca href=u0022https://blanclabs.com/capabilities/ai-analytics-automation/u0022u003eenterprise data + AI solutionsu003c/au003e, we’ve gained perspective on several of the foundational characteristics enterprises will need to develop to realize on the promise of this u003ca href=u0022https://txt.cohere.com/unlocking-productivity-with-generative-ai/u0022 target=u0022_blanku0022 rel=u0022noopeneru0022u003eu003cspan role=u0022linku0022u003e3rd wave of the human-computer revolutionu003c/spanu003eu003c/au003e.rnu003cdiv class=u0022scriptor-paragraphu0022u003eEven though it is still early days we are very enthusiastic about the prospects of LLMs and powerful ML capabilities to unlock vast economic value and growth for progressive enterprises. Some are estimating an impact of u003ca href=u0022https://arxiv.org/pdf/2303.10130.pdfu0022 target=u0022_blanku0022 rel=u0022noopeneru0022u003eu003cspan role=u0022linku0022u003e$1 trillion, or 4% of GDP, in the U.S. aloneu003c/spanu003eu003c/au003e and the generative AI market is expected to see rapid growth, rising from u003ca href=u0022https://www.cmswire.com/digital-experience/what-are-large-language-models-llms-definition-types-uses/u0022u003e11.3 billion USD in 2023 to 51.8 billion by 2028u003c/au003e.u003c/divu003ernu003ch3u003eBut First, a Disclaimeru003c/h3u003ernu003cdiv class=u0022scriptor-paragraphu0022u003ernrnBlanc Labs focuses on the financial services and healthcare industries. We see both sectors as having massive potential in terms of the benefits offered through the application of these transformative technologies.  Both healthcare and financial services are hugely complex, highly regulated, and depending on who you speak to, resistant to change. We believe the stage is set for a broader divergence of winners and losers: organizations that build capability and competency in adopting new tools and leverage the power of their enterprise data will stand to reap the benefits of their investments.rnu003cdivu003eSo with all the hype surrounding generative AI and the rapid advancement of a field that feels to moving forward at the speed of light, it is important to understand why some organizations will be better suited to create value in this field than others, and what organizations can do to ensure success.u003c/divu003ernu003ch3u003e⛔️ Cloud Infrastructure Only ⛔️u003c/h3u003ernu003cdiv class=u0022scriptor-paragraphu0022u003eAWS Enterprise Strategist and Evangelist, Phil Lebrun is a u003ca href=u0022https://aws.amazon.com/blogs/enterprise-strategy/how-technology-leaders-can-prepare-for-generative-ai/?trk=567f905f-0336-452c-b82a-38b2673a1458u0026amp;sc_channel=elu0022 target=u0022_blanku0022 rel=u0022noopeneru0022u003eu003cspan role=u0022linku0022u003epretty smart guyu003c/spanu003eu003c/au003e. In one of his posts about the topic, his flagship advice is u0022Don’t try this without the cloud.u0022 You want your teams focused on problem-solving and innovation, not on managing the underlying complexity and cost of enabling infrastructure and licenses. The cloud is the enabler for adopting AI, making available cost-effective data lakes, sustainably provisioned GPUs and compute, high-speed networking, and consumption-based costing.u003c/divu003ernu003ch3u003eEnterprise Platforms are a Lynchpinu003c/h3u003ernu003cdiv class=u0022scriptor-paragraphu0022u003ernrnIf we consider that having completed a cloud migration or cloud native approach is table stakes for AI  enablement within an organization, the next big question will be where your enterprise data resides. The organization’s technology infrastructure and enterprise platform ecosystem will in many ways determine what toolsets may be available to facilitate experimentation and development of AI enabled use cases.rnrnIf you are a CIO, CTO or CDO, start by taking stock of the platforms that house your enterprise data and evaluate the toolsets offered by these power technology-ecosystem juggernauts. u003ca href=u0022https://azure.microsoft.com/en-us/solutions/ai/dev-resourcesu0022 target=u0022_blanku0022 rel=u0022noopeneru0022u003eu003cspan role=u0022linku0022u003eMicrosoftu003c/spanu003eu003c/au003e and u003ca href=u0022https://azure.microsoft.com/en-us/solutions/ai/dev-resourcesu0022 target=u0022_blanku0022 rel=u0022noopeneru0022u003eu003cspan role=u0022linku0022u003eAWS u003c/spanu003eu003c/au003eare emerging as clear leaders in terms of providing their customers powerful tools to apply ML and AI to their enterprise data, but Salesforce, SAP, Pega and OpenText – amongst many others will offer tooling to access, manage and apply AI to enterprise data.rnrnIt is impossible to predict who the big winners and losers of this space will be, partially because they are such massive and well entrenched players but keep an eye on this space over the next few years.rnrnu003c/divu003ernu003ch3u003eThe 👿 is in the Data [Taxonomy]u003c/h3u003ernu003cdiv class=u0022scriptor-paragraphu0022u003ernrnBuilding an enterprise data model across a complicated operating environment like a bank is no small feat. Standardizing data definitions between different lines of business in a bank involves tackling the complexity of enterprise (+legacy) platforms used to store and handle data. This complexity extends to accommodating the diverse needs of various stakeholder groups with distinct preferences for accessing, manipulating, and analyzing data.rnrnIn many ways, risk is a unifying theme and responsibility within regulated industries like the banking sector. We see conformance to u003ca href=u0022https://medium.com/@willempretorius/an-introduction-to-credit-risk-in-banking-basel-ifrs9-pricing-statistics-machine-learning-76f9eafb8ec1u0022 target=u0022_blanku0022 rel=u0022noopeneru0022u003eu003cspan role=u0022linku0022u003erisk measurement and regulatory reporting requirementsu003c/spanu003eu003c/au003e as a lightning rod for action amongst FI’s to get their u0022data houseu0022 in order but proactive approaches amongst banking and technology executives are few and far between.rnrnIn a recent interview with u003ca href=u0022https://www.tech-execmagazine.com/interviews/speaking-of-synergy/u0022 target=u0022_blanku0022 rel=u0022noopeneru0022u003eu003cspan role=u0022linku0022u003eTech Exec magazineu003c/spanu003eu003c/au003e, VP of Data and Adaptive Intelligence at Munich Re Canada Branch (Life), u003ca href=u0022https://www.linkedin.com/in/lovellhodge/u0022 target=u0022_blanku0022 rel=u0022noopeneru0022u003eu003cspan role=u0022linku0022u003eLovell Hodgeu003c/spanu003eu003c/au003e encapsulated the shift in how progressive organizations are thinking about their data:rnrnu003c/divu003ernu003cblockquoteu003ernu003cdiv class=u0022scriptor-paragraphu0022 style=u0022padding-left: 40px;u0022u003eu003cemu003eu0022Over the years, there has been a realization that data has an inherent information value to the organization. As the years go by, we’re rapidly producing more data at an increasingly and somewhat alarming rate. So, what a lot of folks in the industry have realized is that there is not a huge space between theoretical concepts around information from data, and how they can use that information to simplify internal processes and also benefit clients.u003c/emu003eu003c/divu003eu003c/blockquoteu003ernu003cblockquoteu003ernu003cdiv class=u0022scriptor-paragraphu0022 style=u0022padding-left: 40px;u0022u003eu003cemu003eSo, you’re seeing that interdependence starting to mature. And because of that, many organizations are looking at their data not as just something they have to store, but as an asset they can leverage. And once you start thinking in that sort of way, then you start to formulate methods by which you can gain value from data, and it becomes an important part of your corporate strategy.u0022rnu003c/emu003eu003c/divu003eu003c/blockquoteu003ernu003ch3u003eApp Design and Development Now Occurs ‘In the Data’u003c/h3u003ernu003cdiv class=u0022scriptor-paragraphu0022u003ernrnHistorically, software design and development occurred in a sandbox that was devoid of enterprise data. In traditional dev environments we piped in dummy data to test parameters and during QA tested connectivity and security. But with the explosion of sophisticated ML and AI tools, application development will very much occur at the nexus of querying, manipulating, analyzing and visualizing proprietary data sets.rnrnThere is a tremendous amount of opportunity for these new tools to u003ca href=u0022https://www.youtube.com/watch?v=3uTTGWvvb90u0026amp;ab_channel=DataCouncilu0022u003eu003cspan role=u0022linku0022u003edemocratize data u003c/spanu003eu003c/au003efor business users through the use of conversational/natural language query tools to conduct analysis and derive insights from enterprise data sets.rnrnAnother big change will be the blurring of the lines between the role of developers and data scientists and in many cases, these resources will be working directly alongside each other.  Technical resources will require access to live data sets to facilitate model training and to leverage powerful data analytics toolsets. This is going to require a significant shift (from reactive to proactive) in terms of access management and data security.rnrnFirms that work with external partners will require them to adhere to an extremely high standard of data security protocols i.e., u003ca href=u0022https://www.iso.org/standard/27001u0022 target=u0022_blanku0022 rel=u0022noopeneru0022u003eu003cspan role=u0022linku0022u003eISOu003c/spanu003eu003c/au003e and u003ca href=u0022https://www.auditboard.com/blog/soc-2-framework-guide-the-complete-introduction/u0022 target=u0022_blanku0022 rel=u0022noopeneru0022u003eu003cspan role=u0022linku0022u003eSOCu003c/spanu003eu003c/au003e 2 compliance certification with regular audits.rnrnSpecific sector, domain, and knowledge of business context has always counted for a lot but increasingly developers won’t be able to rely on detailed functional, technical and design requirements documentation to define the specs of what they are developing.  It will require a much more experimental, test and learn approach.  Toolsets are only going to get more user friendly to work with but developing high impact apps leveraging advanced AI/ML techniques will require a detailed and nuanced understanding of user needs, workflow, data inputs, command prompts, prompt modelling strategies, and a continuous improvement approach to get to a place of lasting and tangible impact for businesses and teams.rnu003ch3u003eIn Summary:u003c/h3u003ernThere is much still to be learned through this large-scale adoption of a suite of technologies that are evolving at such a rapid pace. This is by no means a complete list of the things business teams and leaders need to consider as they look to capture competitive advantage from these powerful new tools. At Blanc Labs, we are excited about the future and look forward to working with companies to develop their capabilities to create business value by harnessing the tremendous potential of their enterprise data.rnrnu003c/divu003ernu003c/divu003e