Generative AI Adoption Framework Slides

Error Rate vs Outcome Quality

Dataknobs approach for GenAI adoption

Quick adoption : Use cases where good quality output can be generated from out of box LLM.

Adopt with care : Use cases where there are some issues, but risk of error is low can be adopted quickly.

Task sepcific implmentation: These are use cases where out of box LLM does not work. You may need to implement RAG

Domain specific Finetuning: These are use cases where out of box LLM output is unacceptable. You may have to build your wn LLM or finetune existing one.

Build Data Product: These are use cases where you get your data, build higher level signals, train on your data and build complete Data Product for competetive advantages.

Dataknobs built tool set to automate this. In 30 min simple scenario can be set up for quick adoption. RAG specific implementation can be done in a day.

Multi Disclipinary GenAI Task force for Adoption

How Dataknobs review impact and risks of GenAI

Set up multi disciplinary Gen AI task force to evaluate impact e.g. IT should determine right implementation and operations architecture, cyber security should focus on evaluating security risk impact, finance should review cost and ROI, data governance team should review how new data/content generated will be manged and integrated with existing process.

GenAI Adoption in Enterprise

Dataknobs Adoption Framework - Enterprise and Startup

We have define 2 frameworks - 1. Enterprises 2. Product companies. Enterprise want to get ROI and Product companies want to build defensible products. Using these framework companies can determine which area they should build gen AI capabilities and how to measure impact and benefit.

These framework help you determine error-rate, tolerance for eror, customization and effort involved, risk involved. Combining these one can make a plan for Gen AI adoption

We have build framework to evaluate existing, evolving and new risks.

We have costing framework to evaluate cost.

In addition we have set up best practices for Gen AI implementation, evaluation, risk mitigation.

Using above we are able to apply GenAI in marketing, Edtech scenarios. We have also build capabilities in Finance AI Assistant. We have also built AI asssitant for dietitian and other helpful scenarios in which personalized output is game changer.

GenAI Adoption Stages in Enterprise

Dataknobs Adoption Framework - POC, Well Governed, Strategic and Transformational

After plan, enterprise can implement. For implementation one select what type of initative to launch and accrdingly set acceptance criteris

POC is to prove feasibility, Tactical use cases for cost saving, Strategic implementation are for long term advantages and transformational use cases are for new businesses.

Products and Transformational Use Cases

GenAI for building new Products

Enterprises want to build new products for competitive advanatges

.

Startup and Product companies want to build new genAI based products for revenue.

At Dataknobs - we define framework how companies can evaluate and identify opportunities for data productization

Why Kontrols matters

Control GenAI and AI output

GenAI create new trajectories of data. It may produce unwanted output.

Apply controls to check facts and avoid producing incorrect answers.

Apply controls to produce output that is natural.

Produce responses as per law and governace policies.

Why Knobs matter

Knobs are levers using which you manage output

See Drivetrain appproach for building data product, AI product. It has 4 steps and levers are key to success. Knobs are abstract mechanism on input that you can control.