How to Choose a Machine Learning Consulting Firm

9 things to check when you pick a consultancy.

Building an AI team is hard. Many times, it’s best to hire an outside team to build your AI solution.

Still, the downsides to consulting firms are also well-known:

  • They’re famous for making things more complex than they need to be.
  • They can struggle to meet deadlines and stay on budget, or even
  • Fail to deliver on their promise of a working ML solution altogether.

It’s hard to separate a top consulting firm from an upstart when you yourself are just getting started with machine learning. This guide will help you find a consultancy that is worth working with.

In this article, you’ll see

  • 9 things to look for in an ML consultancy,
  • The questions you need to ask yourself before reaching out to firms, and
  • Tips to avoid the common mistakes that lead to failed or overly expensive projects.

9 things to look for in a machine learning consultancy

Taking a look inside the box

1. They make machine learning simple

Your machine learning consultancy should be able to explain how machine learning works in language that you and your team can understand.

What is machine learning? If you use AI today then it’s almost always about building a system that can take data and turn it into some prediction.

Simple examples of machine learning

To be able to make these predictions, Machine learning algorithms find hidden patterns that connect the data to the correct prediction, by looking at lots of training examples (that’s why you need lots of data to make AI work). 

For an excellent visual illustration of how a model trains, finds patterns and then makes predictions check R2D3’s a visual introduction to machine learning.

2. They’re quick on their feet

An AI consulting firm that’s worked on enough projects can tell you whether your idea is promising in minutes, not days or weeks.

In 90% of cases, an experienced agency can give you a “Yes, this could work,” or a confident “No” within ~15 minutes of you explaining your problem, goal and available data.

They can’t guarantee you’ll get meaningful results at that point, but they should be able to tell you which ideas are worth pursuing. You’ll still need to build out a proof of concept to validate that what you want to predict can in fact be predicted accurately with your data – and that should take no more than 8 weeks.

3. They’re not afraid to tell you “no”

Is a machine learning consulting firm ready to work on every idea you give them? 

Or do they guide you to the best solution, even if it’s not the one you expected?

Many machine learning ideas are not a good idea upon closer look. If you don’t have the right data, or there is no clear path to a solution – even for experts – then they should tell you.

Likewise, not every project will require AI.

We regularly have clients come to us for machine learning consulting when a simpler system is likely to get them 90% of the results. In these and other cases we often recommend against using machine learning and point them in the right direction to do it themselves or with a standard software development consultancy.

Not every project is possible. Even fewer are worth pursuing. 

Seek out an ML agency that will tell you which ideas to toss out early and focus on the use cases that will bring you results quickly.

4. They’re transparent about their experience

Don’t let an AI consultancy hide behind NDAs. Privacy is a top priority in the world of data science, but a respected AI firm should still have permission to point you to some specific projects and talk about some real names and numbers.

We have NDAs with all of our clients, but we can still tell you who we worked with, what problem we solved and what the impact was.

These cases should come with names and contact details of the project managers of the companies they worked with.

Choose a trusted AI agency that has strong references and gives you insights into specific projects they’ve built, how they approached those projects, and what results they accomplished. Be sure to ask if those projects are research or POC, or if they are actually deployed for use in their clients’ everyday business.

5. They ask the right questions

Did you just leave a meeting feeling like they asked all the right questions? Did they ask follow-up questions and dig and probe until they got right to the heart of your problem?

To get your project right, your AI consulting firm needs to learn as much as possible about your business and your domain—what you care about, your vocabulary, what your data looks like, and why what you want to predict is meaningful to you.

They need to be able to repeat your goals back to you in a way that sounds right to you.

Over the course of a machine learning project, the consulting team will make critical decisions that will determine the outcome of your project. They’ll also bring you ideas or make suggestions for improvement. They can only do these things effectively if they truly understand your business and your goals.

Your consultancy should be more focused on understanding your problem than selling you their services.

Small miscommunications in the beginning can lead to expensive, hard to fix, mistakes later on.

6. They’re generous with their knowledge

Your machine learning consulting agency should be upfront about everything—like how much effort it will take, which technologies they’ll employ, and what techniques they’ll use to get there.

In our first calls with potential clients we give specific feedback. We assess whether they have the right data and sketch a plan for how they can go from idea to finished solution – including the algorithms, setup, and the budget they will need to get there. 

There are very few “secrets” in machine learning. It’s not a secret that you’ll use convolutional neural nets for automated radiology exams, or that hybrid recommenders work best in recommending content from a frequently updated library (like Netflix or news sites).

The real value of a consultancy is not in their ideas, but in elegantly solving the hundreds of small problems they are sure to run in – on the way to building a solution that solves the right problem.

7. They’re serious about “process”

“We’ll do whatever it takes” isn’t good enough. Ambition should always be backed by experience.

An experienced ML or AI consultancy will have a proven process – fine-tuned over many projects. Otherwise the chances are very slim that they will finish the project within the time and budget you originally planned.

8. They have a technical project manager

Machine learning project manager role

Machine learning projects are different from standard software projects and it takes a project manager (PM) with both communication skills and technical know-how to keep your ML project on schedule.

In a machine learning project, it’s impossible to separate the project management role from the role of the engineering manager, because how your data is selected and combined directly impacts whether your solution produces the results you want.

That’s why one of the people you talk to daily should be the same person who is writing the tasks for the engineering team. And that person must also have the communication skills to give you progress reports that you can understand.

We found that a senior machine learning engineer can take that role. In addition our CEO and CTO is overseeing every project to constantly ensure we are keeping true to our clients' actual goals.

9. They help make your team better

A good consulting firm keeps you in the loop and helps your team grow.

Most likely your project managers and engineers want to learn as much as possible from the project. So pick a consultancy that will work openly with your team rather than building a solution in isolation.

Your consulting firm should clearly document the requirements before the project, regularly push code to a code repository that you can access and give you frequent updates on so that your team can learn, ask questions and – which is crucial – course correct if necessary.

3 questions to ask yourself before choosing a machine learning consultancy

Question marks

Now that you have an idea of what to look for, here are a few questions to ask yourself about what YOU want from your machine learning consulting firm:

1. Do you need an AI strategist, or a builder?

You’ll find two main types of consultancies working in AI: strategists and builders.

Almost all consultancies do both of these things at some level, but very few firms do both well.

Machine learning strategy consulting:

Teams that work on projects that revolve around developing an AI strategy that fits into your corporate vision. Their employees are more likely to have experience as corporate consultants rather than engineers. They tend to focus on developing a comprehensive strategy rather than a working solution. They most likely don’t have the engineering power to execute AI products end-to-end, and they often outsource the development that they offer or work on PoCs with students.

Machine learning building consulting:

A consulting firm that focuses on building means more engineers and fewer career consultants. That’s not to say they don’t value strategy, but their approach is to identify the areas where they can make the biggest difference, and then to do the work.

In our case we are focused on building, and our staff is 90% engineers. So we can produce a proof of concept in weeks and fully-functional AI products in months—much more quickly than a strategy-focused firm.

Nothing builds confidence in the long-term viability of AI like putting a functional AI system in place quickly. Even skeptical board members become convinced once you start tapping lost revenue from your customer data or trimming 80% of analysis time out of a critical workflow.

We’ve seen that the momentum from just one early win with AI can kick-start the adoption of AI-driven strategies company-wide.

2. What do you want to pay for? (results vs. hours)

Most consulting firms want to work by the hour or day because it limits their risk and shifts the risk to you, the client.

If a project takes twice as long as estimated or never reaches a successful conclusion, it’s not their problem. The consultancy gets paid more despite your frustration.

Only consultancies that are confident in their assessment of both your problem and their own skills can offer you the alternative: fixed project pricing.

Project pricing moves your risk to the consultancy. If they don’t deliver the results you agreed upon, then they don’t get paid. And if it takes longer or requires more effort than expected, then the extra cost falls on them, instead of you.

3. Can you get by with SaaS, or do you need a custom machine learning solution?

You have 3 options for building a solution with AI or machine learning:

1. AI SaaS services

Example: Google’s Speech-to-Text API.

These services are good building blocks that you can build into an application. If hundreds or thousands of other businesses have exactly the same AI challenge as you (like turning English audio to text) then it’s likely that there is an API out there that solves this problem better than you can with reasonable effort.

2. Machine Learning as a Service Platforms 

Examples: Cloud platforms like Amazon SageMaker, IBM Watson or Microsoft Azure

These platforms promise to make any machine learning project easier to build but our experience is that they only work for straightforward use cases. And surprisingly few ML applications turn out to be straightforward enough to fit into that model.

3. Build your own

If you want to use data sets that are unique to you to solve a complex problem, then in most cases you need to build your own custom ML or AI solution. 

Fortunately, best practices for planning, building, and scaling applications are readily available. That, combined with open source machine learning libraries and accessible cloud compute power (like AWS EC2) are making building custom ML products easier every day.

Beware of “superior algorithms”

Digital look, blackbox algorithms

A final note of caution: People selling AI or machine learning consulting services based on their “proprietary algorithms” either don’t know what they’re doing, or they’re not being honest about what matters to succeed in a project.

After all, if algorithms are such a big secret, then why would Google and Amazon make so many of their best algorithms public?

Your data, your domain knowledge, and how your project is executed is what makes the real difference.

Here are the 3 most important elements that actually determine machine learning success:

  1. Understanding: internalizing your data, problems, and goals;
  2. Engineering: writing well-documented, easy-to-scale code;
  3. Infrastructure: building a rock-solid infrastructure that can handle your volume of requests.

Pay for expertise, not algorithms.

In summary

Find a machine learning consulting firm that is 100% transparent about their experience, their skill set, and their process. They should speak your language, ask you a ton of questions, and tell you what they’ll do in a way that makes sense to you.

Don’t let them hide behind AI terminology. If you don’t understand how something works, keep asking questions until you do. And if they can’t make it clear for you, find another consulting firm who can.

If you don’t find a machine learning consulting firm that fits your criteria – we are builders who focus on taking complex and large scale AI projects from idea to production – if that’s what you are looking for, let’s talk.

Energy Transmission

Anticipating and Preventing Power Grid Failure

Massive power outages cause chaos for the general public, and they cost utility providers roughly $49 billion a year.

This wouldn’t be much of a problem if massive power outages were rare, but outages affecting more than 50,000 people have increased dramatically in recent years. This means utility companies need to find new ways of anticipating and managing these outages.

These days, smart grids are producing massive amounts of data, which means predicting and managing outages is easier than ever. Unlike traditional power grids, which are one-directional (meaning they only transmit power in one direction), smart grids are two-directional. They can capture data from every possible source in the grid at the same time as they’re providing electricity. They collect and monitor data from sources like smart meters, IoT devices, and power generation stations, providing a clear, real-time look at power usage.

Machine learning can use this data to anticipate and prevent massive power outages in the grid. Machine learning helps identify non-obvious patterns in the data that can be a precursor to grid failure, which helps maintenance teams preempt failure.

Balancing the Grid

Balancing the grid — making sure energy supply matches energy demand — is one of the most important jobs a transmission operator has. But renewable energy sources depend heavily on the weather, making them harder to predict.

Transmission operators spend millions each year fixing planning mistakes that lead to producing too much or too little power. In hybrid systems — which rely on both renewable energy sources and fossil fuels to generate electricity — these mistakes have to be corrected at the last minute by buying more energy or compensating power plants for the excess.

Machine learning is the most accurate method available to forecast the output of renewable energy. Advanced methods, like Long Short-Term Neural Networks (LSTMs), can weigh the many factors involved — wind, temperature, sunlight, and humidity forecasts — and make the best predictions. This saves money for operators and preserves resources for power plants.

Preventing Blackouts and Brownouts With Real-time Monitoring and AI Prediction

Power grids have a lot of obstacles to overcome in providing continuous energy to customers. Weather patterns, usage, internal failure, even wildcard incidents like lightning strikes and interference from wild animals can all affect power delivery.

Machine learning is increasingly being used to help predict potential brownout and blackout conditions. By feeding historical data into the AI and running Monte Carlo simulations to predict potential outcomes, grid operators can use machine learning to identify conditions that could lead to grid failure. And they can act accordingly.

Sensors like phase measurement units (PMU) and smart meters can provide usage information in real-time. When combined with both historical and simulation data, AI can help mitigate potential grid failure, using techniques like grid balancing and demand response optimization. Incidents that would otherwise have affected millions of people can be contained to a smaller area and fixed faster for less money.

Differentiate Power System Disturbances from Cyber Attacks

Cyber attacks are increasingly used to target important infrastructure, like shutting down hospitals with Ransomware attacks (when attackers break into the system and lock legitimate users out until a ransom is paid). With utility grids, a cyber attack can have widespread consequences and affect millions of users.

Detecting these attacks is critical.

Developers are using machine learning to differentiate between a fault (a short-circuit, for example) or a disturbance (such as line maintenance) in the grid and an intelligent cyber attack (like a data injection).

Since deception is a huge component of these attacks, the model needs to be trained to look for suspicious activity – things like malicious code or bots – that get left behind after the deception has occurred.

One such method uses feature extraction with Symbolic Dynamic Filtering (an information theory-based pattern recognition tool) to discover causal interactions between the subsystems, without overburdening computer systems. In testing, it accurately detected 99% of cyber attacks, with a true-positive rate of 98% and a false-positive rate of less than 2%. This low false-positive rate is significant because false alarms are one of the biggest concerns in detecting cyber attacks.

Balance Supply and Demand

Utility providers are looking for ways to better predict power usage while maintaining maintaining energy supply at all times. This becomes critical when renewable power sources (like solar or wind) are introduced into the grid.

Because these renewable power sources rely on elements beyond human control (like the weather), utility providers know they can’t always rely on renewables for continuous production. Knowing precisely when demand levels will peak allows utility providers to connect to secondary power sources (like conventionally generated electricity) to bolster the available resources and ensure constant service provision.

More and more utility providers are turning to machine learning for help. We can feed historical data into machine learning algorithms -- like Support Vector Machines (SVM) -- to accurately forecast energy usage and ensure sufficient levels and constant supply.

Detect Power Grid Faults

Current methods for detecting faults in the grid consume a lot of unnecessary time and resources. This creates a situation where power transmission is interrupted and customers are without electricity while faults are first located, then fixed.  

Machine learning can find faults quickly and more accurately helping you minimize service interruption for your customers.. Support Vector Machines (SVM) are combined with Discrete Wavelet Transformation (DWT) to locate faults in the lines using a traveling wave-based location method.

When we apply  DWT (a form of numerical and functional analysis that captures both frequency and location information) to the transient voltage recorded on the transmission line, we can determine the location of the fault by calculating aerial and ground mode voltage wavelets. So far, this method has detected fault inception angles, fault locations, loading levels, and non-linear high-impedance faults for both aerial and underground transmission lines.

Detect Non-Technical Power Grid Losses

In the energy world, “non-technical losses” means energy theft or fraud from the system.

There are two common types of non-technical losses. The first is when a customer uses more energy than the meter reports. The second involves rogue connections stealing energy from paying customers. To pull off this theft or fraud, bad actors can bypass smart meters completely or insert chips into the system that change how meters track energy use. Meter readers can also be bribed to report lower numbers (though thanks to smart meters, this is increasingly hard to do).

Because these non-technical losses cost $96 billion annually, utility providers are turning to machine learning to combat the problem.

We can help utility providers mine historical customer data to discover irregularities that indicate theft or fraud. These can be things like unusual spikes in usage, differences between reported and actual usage, and even evidence of equipment tampering.

Energy Distribution

Better Predict Energy Demand

Accurately predicting customers’ energy needs is critical for any utility provider. To date, we haven’t found an adequate solution for bulk energy storage, which means energy needs to be transmitted and consumed almost as soon as it’s produced.

We're using machine learning to increase the accuracy of these predictions. Historical energy use data, weather forecasts, and the types of businesses or buildings operating on a given day all play a role in determining how much energy is used.

For example, a hot summer day mid-week means more energy usage because office buildings run air conditioning at a high capacity. Weather forecasts and historical data can help identify those patterns in time to prevent rolling blackouts caused by air conditioners in the summer.

Machine Learning finds complicated patterns in the various influencing factors (such as day, time, predicted wind and solar radiation, major sports events, past demand, mean demand, air temperature, moisture and pressure, wind direction, day of the week, etc.) to explain the development of demand. Because machine learning finds more intricate patterns, its predictions are more accurate. This means energy distributors can increase efficiency and decrease costs when they buy energy – without having to make expensive adjustments.

Energy Generation

Predict Turbine Malfunction

Wind is a great renewable energy source, but wind turbine maintenance is notoriously expensive. It accounts for up to 25% of the cost per kWh. And fixing problems after they occur can be even more expensive.

Machine learning can help you get ahead of this problem. The goal is to reduce maintenance costs by catching problems before the turbine malfunctions. This is particularly important when wind farms are located in hard-to-access places, such as the middle of the ocean, which makes repair costs even higher.

Real-time data gathered with Supervisory Control and Data Acquisition (SCADA) can help identify possible malfunctions in the system far enough in advance to prevent failure.

For example, data from sensors found within the turbines – such as oil, grease, and vibration sensors – have been used to train machine learning models to identify precursors to failure, such as low levels of lubricant.

This method can train machine learning models to predict failures up to 60 days in advance.

Consumption / Retail

Accurately Predict Energy Prices

As personal power generation (using solar or wind power) gets easier and cheaper, consumers and businesses are increasingly producing their own power.

Personal power generation allows people to make, consume, and store their own energy. Depending on where they live, they may even be able to sell surplus power back to the local power utility.

Machine learning can help find the best time to produce, store, or sell this energy. Ideally, energy should be consumed or stored when prices are low and sold back to the grid when prices are high.

By looking at historical data, usage trends, and weather forecasts, machine learning models have made accurate predictions on an hourly basis. People with personal and business energy generation systems can use these predictions to make strategic decisions about whether to use, store, or sell their energy.

For example, Adaptive Neural Fuzzy Inference System (ANFIS) has been used to predict short-term wind patterns for wind power generation. This allows producers to maximize energy production and sell it when energy prices are at their peak.

Reduce Customer Churn

In open energy markets, where customers have a choice of utility providers, understanding which customers are going to churn out can be critical. Churn rates, which is the percentage of customers who stop using your service in a year, can be as high as 25%. Being able to predict churn and stay ahead of it is essential to survival.

Machine learning is helping utility owners predict when a customer is getting ready to churn out. By using techniques such as Cross-industry Standard Process for Data Mining (CRISP-DM), AdaBoost, and Support Vector Machines, as well as historical usage data, utility providers can identify key indicators of whether or not a customer is going to churn. These indicators include things like customer satisfaction, employment status, energy consumption, home ownership or rental status. A change in any of these can indicate a customer is getting ready to terminate their service.

When these indicators are identified far enough in advance, it’s possible to avoid churn by working with customers to solve any problems they’re experiencing.

Energy Trading

Predict Energy Prices

Just like natural gas and oil, wholesale energy is a market commodity. So naturally it's important for traders to be aware of market fluctuations and pricing when it comes to buying and selling energy.

To help make sense of the massive amounts of data used to make trading decisions, traders are increasingly turning to machine learning.

A mix of statistical analysis and machine learning can help commodity traders make better predictions. Classical statistical analysis techniques like time series analysis, Seasonal Autoregressive Integrated Moving Average (SARIMA), and regression models are used to deal with the data. And machine learning makes connections between the various data points.

What’s more, machine learning trains itself to make increasingly accurate predictions using the constant flow of real-time data.

Get Notified of New Articles

Leave your email to get our weekly newsletter.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.