AI Strategy

Some common ways to use AI

For developing an AI strategy always remember some common ways to use

  • Developing more intelligent products
  • Developing more intelligent services
  • Making business processes smarter
  • Automating repetitive business tasks
  • Automating manufacturing processes 

Startup the AI strategy

  • Leverage AI to make a plus specific to your company
  • A design strategy that aligns with a virtuous cycle of AI
  • Blue River – precision agriculture
  •  AI must be specialized or verticalized to your industry sector
  • Don’t compete with giants Creating a strategy
  • Strategic data acquisition
  • Unified data warehouse – Pull data into a single repository, the software can
  • Create network effect and platform advantages Uber, Careem, Facebook
    Low-cost strategy
  • Low-cost strategy
  • High-value strategy

 Develop internal and external communications

  • Investor relations 
  • Government relations 
  • Consumer / user education 
  • Talent / recruitment 
  • Internal communication

 AI teams

  • AI teams may have hundreds of engineers.
  • A small team can have four or five members. 
  • Example role: Software Engineer (Execute joke, set timer)
  • Machine learning Engineer
  • Machine learning researcher (Extend state-of-the-art, applied Machine learning scientist in-between  machine learning researcher and machine learning Engineer)
  • Data Scientist (Provide insights, make a presentation to team/executives)
  • Data Engineer (Organize data, data is saved in a cost-effective way, we have a lot of data, scalability is important)
  • AI Product Manager ( Help decide what to build; what’s feasible and valuable).

Always Keep in Mind

  • Data privacy will be a key consideration during adopting the AI strategy.
  • Make sure that AI is free of bias and discrimination.
  • AI should be used for the benefit of a company, customers, and its employees.
  • Never adopt the adverse uses of AI.

 Common PitfallsDon’t

  • Expect AI to solve everything
  • Hire 2-3 ML engineers and count solely on them to return up with use cases.
  • Expect the AI project to figure the primary time.
  • Expect the traditional planning process to apply without changes.
  • Think you would like superstar AI engineers before you’ll do anything.


  • Be realistic about what AI can and can not do given limitations of technology, data, and engineering resources.
  • Pair engineering talent with business talent and work cross-functionally to seek out feasible and valid.
  • Plan for AI development to be an iterative process with multiple attempts needed to succeed.
  • Work with the AI team to determine timeline estimates, milestones, KPIs, etc. 

Take your first step

  • Get friends to learn about AI
  • Start brainstorming projects
  • Hire a couple of machine learning/data science people to assist.
  • Hire/appoint an AI leader (Vice President AI, Chief AI Officer)
  • Discuss with CEO about possibilities of AI transformation

Get started with a small team

  • One software Engineer
  • One machine learning/ Data Scientist
  • Nobody but yourself

Building AI projects:

  • How to choose an AI project?
  • Just keep in mind the AI knowledge and domain knowledge.
  • Brainstorming frameworks means group discussion to produce ideas or solve problems.
  • Automate tasks rather than job
  • Main drivers of business value
  • What are the pain points in your business

Is it always necessary to have big data?

  • Having more data is good
  • With small data sets, we can make progress
  • 10,100 or 1000 data points can be a good start

Technical diligence

  • Can AI team/system meet desired performance
  • How much data is needed
  • Engineering timeline

Business diligence

  • Lower cost
  • Increase revenue
  • Launch new product or business

Build Vs Buy

  • Machine learning projects can be housed or outsourced.
  • Data Science projects are generally in-house.
  • Buy industry standard, only build specialized products.

Working with an AI team

  • Specify your acceptance criteria (Goal: detect defects with 95% accuracy)
  • Training, validation, and test data set
  • Don’t expect 100%  accuracy (Limitations of machine learning, insufficient data, mislabeled data, ambiguous labels

AI technical Tools

  • Machine learning frameworks for writing software ( Tensor flow, Pytorch, Keras, MX Net, CN TK Caffe, Paddle Paddle, SciKit-learn, R, Weka)
  • Research publications (Arxiv)
  • Open Source Repositories (Git hub)