This post is part of the AI for Business series
Shedding light on AI from a non-technical business/product/design perspective.
This is post #9 in our AI for Business series.
Wohoo, you’ve made it! 🥳 This is the very last blog in our AI for Business series. Give yourself a pat on the back for making it this far!
In this final week we take a look at the rollout of your new AI strategy and, good news for Product Managers and Business folks, this one is going to feel a lot more familiar 😎
Research Phase
The genesis of any AI-powered digital transformation is the research phase; our blog series focuses on this. To recap, after this series you should be in a position to write a report or create an analysis which can answer questions like:
- What is the problem we are trying to solve?
- Are we diagnosing the problem correctly?
- Are we focusing on the right problem?
- How do we define success?
- How do we measure success? What are our KPIs?
- Where does AI fit with our existing processes?
- Can we use cognitive Robotic Process Automation (RPA) to automate any steps in our process?
- Can we use cognitive insights to enhance an existing process?
- Can we use cognitive engagement to improve a process?
- Should we be improving existing processes or is there an opportunity to make a bold move and reinvent how we do things?
- Are we sure these changes help us?
- Do they drive us towards our definition of success?
- Do they add value to customers?
- What data do we need to take action?
- Do we have this data?
- If not, is it worth trying to gather this data?
- Does our organisational structure and culture help or hinder AI implementation?
- Is AI incorporation aligned with our company vision?
- Do we need to improve communication, transparency and training for our team?
- Is our priority income growth (preferred) rather than expense reduction?
- Where is our organisation on the Gartner AI Maturity Model?
- What are the implications for:
- Customers?
- Employees?
- Society?
- Are there legal or ethical considerations?
These explorations begin, but are not exhaustively explored, in this blog series. Challenge your team by asking these types of questions before proceeding with your AI implementation. It will help you identify whether AI is appropriate / worth the investment for your organisation at an early stage before incurring too many costs, or if a pivot is required.
Cost - Benefit Analysis
Okay, you now have some good ideas about how you think AI can help your company. Your next piece of investigation will be to figure out whether your proposed change is affordable.
You’ll need to conduct some followup research on the costs involved with your proposed changes. This portion of research aims to answer questions like:
- Can we afford this change?
- Similarly, can we afford not to make this change?
- Is it possible to estimate development costs at each stage?
- Are there existing tools we can leverage?
- Can these tools reduce our implementation or operational costs?
- Can we treat AI as a general tool rather than getting tied to a specific provider?
Research around existing tools should be fairly straightforward. Estimating development costs from the outset is much more tricky. Why not check out this thoughtbot article on the subject to help you along?
Rollout Phase
If the research phase and cost-benefit analysis demonstrate that this is a viable project, it’s time for an iterative rollout process. Think pilot studies, Minimum Viable Products (MVP) and the like. We are back on familiar turf 😎
No matter what label you use, you should take an iterative approach to your rollout. Beginning with a pilot study or MVP allows you to validate your assumptions prior to a broader, phased rollout. We have lots of thoughtbot blogs on this subject so we won’t labour the point here.
In short however, each part of the project should be broken down into smaller, more manageable phases. After a small scale test release, each phase should be assessed against our goals and KPIs. If it demonstrates value and gets us closer to our goal, it can be rolled out further. If not, it must be iterated on before broader rollout. This approach reduces risk and costs and improves your odds of achieving Product - Market fit.
thoughtburgers 🍔
Let’s revisit our delectable, AI-powered and entirely fictitious side-business, thoughtburgers (home to the tastiest and most thought-provoking burgers!).
To date in this series, we have looked at how thoughtburgers is planning to use AI to speed up deliveries to millions of hungry customers during rush hours. We looked at how automation, cognitive insights and cognitive engagement could be applied to make our current delivery processes better. This covered everything from automated and improved delivery driver scheduling and route optimisation to more effective triaging of delivery orders between different restaurant branches.
So, our rollout of this strategy could look something like this:
- Our first step is to use Robotic Process Automation (RPA) to automate Steps 1, 2 and 3 (the delivery manager scheduling drivers) in one particular testing branch.
- If this proves successful in the testing branch, as defined by our success KPIs, we begin to roll it out to other branches in London. If the change continues to show promise in other London branches, we roll it out to the thoughtburger branches in other cities like Brighton, Manchester and Newcastle and maybe even to other geographical regions like Ireland or mainland Europe.
- While this RPA rollout is underway in other branches, our test branch begins upgrading the system to Cognitive RPA, leveraging AI to allow the system to learn over time.
- Again, if this change delivers value as defined by our KPIs, we begin a phased rollout to other branches.
- While this rollout is occurring, the test branch teams up with other London-based branches to start experimenting with using cognitive insights to improve order triaging, and so on.
A phased approach like this will allow thoughtburgers to gradually move towards a more mature form of AI. It will also allow our team to understand the changes, become accustomed to them and to provide feedback or raise concerns as they encounter them.
Continuously iterate
The final phase in any good implementation of an AI strategy is continuous iteration. You can’t simply create these systems once and leave them be. It is imperative that as an organisation you continue to:
- Test, moderate and update systems
- Engage, retrain and communicate with employees
- Invest in staying on top of regulatory requirements
- Monitor success metrics
- Provide value to your customers.
Bottom line
A phased rollout is the best way to implement your AI strategy. Start with a detailed research phase before looking at the costs and tools involved. When the time comes to start development, build each phase of your strategy out incrementally and measure each phase against your definition of success and your metrics. Finally, never fall into the trap of thinking your work here is done; for AI systems to be effective they need constant moderation, iteration and investment.
Following this implementation strategy will give you the best possible chance of digitally transforming your organisation with AI.
💡 We hope you have enjoyed this series on AI for Business. If you’re ready to start using AI to transform your business, thoughtbot would love to work with you. Let’s talk about making your AI initiative a success!
This blog post is part of a series based on the course Artificial Intelligence (AI) for Business run by University College Dublin (UCD). I took this course and found it so helpful that I’m recapping my top insights. thoughbot has no affiliation with UCD.
