Getting your AI Strategy Right – A Programmatic Approach for Data Management
In May 2022, Google announced ‘Imagen’, its Generative AI solution which covers text inputs to images. This was few months before the frenzy of Generative AI solutions began after Chat GPT was launched in November. Imagen creates realistic images depicting any text entered by users – something as ambiguous as ‘An alien octopus floats through a portal reading a newspaper’ gives a pretty accurate image.
The not so famous fact is that Google has been investing in developing several image datasets in the last decade such as Open Images with 9 million annotated images. Google partnered with Cornell and CMU universities to create this dataset in 2016. Imagine the amount of work invested into creating this data – annotating images varying from silverware to staircases; faces to mountains! In this case, the AI works well and has the potential to benefit many of Google’s future offerings. As an example, this has the potential to transform the Graphic Design industry – designers can write what they want and the machine will generate the designs. Google might make in-roads in the Graphic Design software space or completely change the way graphic design is done.
Managing the data lifecycle is often the biggest critical success factor for AI projects. Its importance unravels once execution starts. It is no surprise that almost 50% of the time in an AI project is spent sourcing and managing data.
Your AI projects need data for initial training and then the models need to be retrained based on updates made to the data on a recurring basis. We refer to this as ‘Foundational Data’. While the importance of ‘Foundational Data’ is common knowledge in AI focused business discussions, many organizations still struggle to get this part of their strategy right. ‘Foundational Data’ requires time, budget, and expertise. In this blog, we will share how Evalueserve has been working with Professional Services clients to help them get it right for Knowledge Management projects, backed by AI.
Professional Services Firms – Using AI for Knowledge Management (KM)
Professional services firms continue to emphasize reusing existing knowledge from documents and the minds of workers. While this is a ‘legacy’ use case, solving this KM problem is still at the heart of many digital projects. Advances in Natural Language Processing (NLP) and Machine Learning (ML) are leading to research in how KM can be done better using AI.
Evalueserve has worked with clients for the past two decades to run extensive KM programs. Our programs are augmented by Digital Platforms and right amount of human intervention. They are designed and customized to deliver the desired value to the end user. These KM programs consist of three intertwined pillars – Knowledge Process Re-Engineering, Digital Solution Deployment, and Change Management.
Publishwise is Evalueserve’s AI-powered platform that lets users swiftly lookup past content using Modern Search and recommends content faster using NLP.
The key business value delivered to end users in this case are: 1) Better access to organizational knowledge 2) Robust process for managing knowledge assets, and 3) Intuitive and easy access and searchability of knowledge assets. It applies AI to recommend relevant past content whenever the user is looking for information.
The ‘Foundational Data’ for Publishwise is a tagged repository of past content, comprising past proposals, pitchbooks, case studies, frameworks, CV libraries and a business taxonomy. Before implementing Publishwise, we conduct a ‘Discovery Phase’ focusing on establishing and executing the strategy for Foundational Data.
Publishwise has been successful even with clients where the Foundational Data was not in place when we started. Evalueserve’s KM experts work with clients to address it over a 4-6 week ‘Discovery Phase’. Only after the Foundational Data strategy is in place, we plan for the implementation of Publishwise. Below table consists of the typical gaps in Foundational Data and how Evalueserve helps fill them:
|Disparate Sources of Information
|Generally, documents are not saved consistently. They are saved in cloud storages such as SharePoint & Google Drive, local file servers, Intranets, etc.
|· Identify 3-5 stakeholders from business and IT teams to understand the as-is state
· In most of the cases, an existing initiative to save documents needs to be re-energized
· Creation of a ‘to-be’ state in consultation with client IT team, aligning with ongoing, and future tech projects
|Either an organization wide taxonomy doesn’t exist or is not followed consistently
|· Evalueserve KM experts review the existing taxonomy with business stakeholders, update it basis the feedback from a variety of users and their own knowledge
· If taxonomy doesn’t exist, Evalueserve’s KM experts suggest an initial draft basis the inputs from business stakeholders and objectives of the KM program
|Lack of tagged past documents
|· Evalueserve KM experts will tag past documents as per the agreed business taxonomy
· Once there is sufficient tagged data from the past, AI is used to auto-tag new documents with 90-95% accuracy
|Ongoing content management
|Even if there is a taxonomy and repository, lack of process and ownership for the ongoing content management
|· This is fulfilled by an ongoing content stewardship service by Evalueserve
· The KM analysts are custodians of taxonomy and make changes as per the feedback from client’s business stakeholders
Multiplier Effect Across KM Strategy
Publishwise helps end users achieve business objectives by ensuring the effective reuse of past documents and content. Indirect benefits of Foundational Data preparation also extend to our clients as well. Most of the technology leaders are tasked with knowledge management problems and in many cases, identifying a starting point is the challenge. While we work towards implementing Publishwise, KM of one type of documents, such as Proposals often becomes a ‘Proof of Concept’ which can be replicated across other data assets. A Publishwise implementation:
- Helps create and establish common KM practices among business users, starting with a small set of users who manage content and documents
- Establishes scalable taxonomies and tagging process which can be extended for multiple document types
- Increases the value derived from existing systems such as CRM, SharePoint, etc.
Contact us for a demo and understanding how Evalueserve can help you achieve your knowledge management goals via a tailored program.
Associate Director – Products, Professional Services
LinkedIn Profile – Shobhit Saxena | LinkedIn