CPX2019chat

should be the biggest and "best" problem - biggest drives the most value. Best (do you have the data and the skills?) drives the best chance of success.

@Anaplan that said, should be and is are 2 different things!

(edited)

Maybe we need AI to figure out where we need AI!

This is where you have to identify what problems you are trying to solve.

Currently, AI and ML are most easily adopted in areas where large numbers of small decisions are required. Like customer scoring and classification.

@jsa_SF I was thinking about that. a great startup idea :-)

Findings use cases that are inefficient and could use a machine to do more of that work.

The priorities for AI/ML use cases should be based on solving some existing business problems which will help customer to take better and smarter decision irrespective of industries they are operating in.

Also look for low hanging fruit - this is usually an application of RPA. @anaplan

Using ML to assist with high-level strategic decisions is rare, but will become more common, especially via forecasting and scenario analysis.

start with what are the key challenges today in the organization, what need to be improved and automate. from what we saw, if you start with the mission to learn what AI&ML can do, it won't get anywhere. it need to start with the business problem

As with any tech, I would always be cautious not to fall into a 'we've got a shiny new solution, now can we find a problem for it' trap. Let the normal business problems/opportunities bubble to the surface and just know that AI/ML might be an option.

Definitely do not let market hype fuel initiatives

So far most of the value that has been derived has been in robotic process automation (RPA) – automating repetitive tasks, freeing people up for more strategic analysis. That said, there is a skills shortage here as well.@anaplan

Preeminently it’s the ability to quickly adapt to changing environments and make decisions in real and right time.

We see higher level conversations about planning, once the detailed data analysis can be guided by an AI. Forecasts and classification from AI is currently an input that drives better analysis.

@ERP_cindyjutras Totally agree

@DaveMario Yes we see consistently customers feel the need to make decisions faster and take out time lost to data management.

efficiency, accuracy, productivity.

@Anaplan more accurate prediction (using ML) compared to traditional forecast means, better connected planning across different line of business. For ex, more accurate demand forecasting means better inventory optimization, better demand planning, better production planning.

Finance: mostly cash management, including forecasting of cash.
Manufacturing: analyzing data from sensors to predict and prevent machine failure

(edited)

Yes Sales, forecasting, Marketing, HR are core business areas we see. Initially customer experience goals is a key driver

HR in recruiting and predicting fit based on more than the obvious experience

forecasting is very popular in sales, but I'm surprised how few businesses plan to respond to variations in the forecast with the right levers.

We're seeing a variant of this for customer interactions that don't involve a person.

In HR, lots of analytical tools but CHROs crave a planning environment to act on the insights and connect more closely with the business.

Operational areas...particularly where there are dynamic resource allocation challenges.

accurate forecasting is a great starting point, but tying it back to planning decisions to drive course-correction is far more valuable

No matter whichever industries it's, the main goal of intelligent planning should be making smarter decision faster than ever. But in terms of business use case we have seen tremendous value in fields like Finance and Retail when it comes to ML/Stats based forecasting

in almost every industry and domain, there is a great potential for improvement and accuracy by learning from the past. Sales = Accurate Sales forecasting, account segmentation... HR = predict employee attrition, workforce opt. ; Marketing - what is the ideal promo plan

AI is really appealing at a high level, but business leaders are waiting for the tools to help them gain confidence in the machine generated results.

If it is not led by the business leaders it has less chance of success. That said there is still a lot of work to be done in educating business leaders on what is possible@anaplan

It will have to be a mix. A combination of finding the right use cases and the technology to make it happen.

Key lines of business leaders mostly. But staffing for an AI project is a big problem that negatively hinders initiatives

@ERP_cindyjutras Yes

there is more push from executive to explore the value of AI & ML for their organization.The digital transformation push companies to think differently and be more creative

Although modern huge data sizes have really allowed AI to shine, the results are sometimes mysterious and may require human oversight.

One of the big issues organizations face is the daunting task of aggregating their data and cleansing it to even be able to begin leveraging AI and ML. If that’s not ready the AI project or initiative is likely to fail.

AI/ML is not magic. Need to be clear on problem to solve. It needs address problems that can be solved with math on large data. Problems where history directionally informs future. Data quality also comes up.

Management is rightfully hesitant to make final decisions without concrete evidence and human explanations.

As with many new tech innovations, people/change management can be one of the largest challenges. Getting folk to accept decisions made by a potential opaque process is tough.

A skills shortage is at the top of the list. Most projects now need data scientists. The solution is adding skills and/or delivering applications with embedded AI that don't require these skills@anaplan

Adoption. one of the biggest issue by having a solution powered by ML is having users use the ML recommendation. most cases they don't, due to trust.

another one we've been hearing from our customers is "Lack of engagement between the business and technical teams' most of cases data scientist are experimenting ML but without any objective or high level goal to what to improve in planning

Maturity of the technology and skillset. I think most enterprises will take advantage of AI/ML that comes embedded in technology they already use.

Another challenge is that AI/ML is often out of the comfort zone of the technologists within a company. It doesn't look and feel like a 'normal' systems implementation, and needs a whole new set of skills.

And don't forget some of the legal/moral/ethical challenges. For instance, if you are allowing AI/ML to reach conclusions on your customer base (profiling those customers etc) then you better be sure you can defend those conclusions.

First and foremost, quality of available data to implement different algorithms!, lack of right platform/infrastructure and people to implement AI/ML, stakeholders' understanding of value of AI/ML, user adoption etc.

partnering with the business functions to identify the right use cases is a critical piece, and often overlooked when starting with toolsets

It seems like everyone is in the space. I’ve said it feels like AI = code these days. Does the audience feel they have the guidelines to navigate a crowded space?

I would agree a lot of vendors are dipping their toes in the space but not a lot have made it through to production

(edited)

A lot of AI projects I’m seeing are mostly pilots. Smart business leaders start with a vision for AI and an up-front understanding of what it can do for the business. What areas of the business are they trying to impact, whether sales, HR, etc

It seems that finding people with deep expertise in this area would be a challenge.

Companies need staff that can translate the needs of their specific industry to whatever AI tools are currently available. The field is changing quickly, and the best algorithms are often hosted by dedicated AI providers.

Companies should invest in data science tools (and people) that can evolve to leverage new ideas, not get locked into what is available now.

I agree that a lot is about piloting at the moment, and the use of that to work out if a business case can stack up, and what the future implications might be for resource etc

@DaveMario well said. in fact according to many surveys less than 20% of companies pilot with ML go to production.

Orgs are doing lot of pilot projects before they invest in production, and even if pilot goes successful they are hesitant. Mostly they are trying to find value from pilots but without proper investment of infra and people it could hardly provide the expected value...