AI Consultant: What a professional can do for you

As a company searching for AI consulting or an AI consultant, you are faced with new hypes on a daily basis, talented salespeople, snake oil, and real (generative) AI breakthroughs that are actually commercially usable.

The problem: What to do? Buy a set of niche AI tools and weld them together, and change the processes of your entire organization to fit a set of tools that might be obsolote in a week or two? Or wait until the market is bit riper?

The solution: Start with external help – i.e. a freelance AI consultant – on a blank sheet of paper, first in the context of an AI workshop. Before a larger rollout, you should determine which tasks in the company a) are time-consuming and b) can be (partially) automated with AI.

This is where the problem starts, because who has a good overview of their own daily tasks – let alone the tasks of the teams? To be honest: Rather no one. So it’s time for analysis.

It is particularly important to work with user stories: What process optimization do I want to achieve? Only then you are able to select AI systems that will really help you, and you still need to you evaluate them in the next step.

After the workshop, you can start building simple prototypes for everyday use, and a monitored, diverse test team should test them really hard.

These prototypes can be, for instance, prompts, GPTs, ready-made purchased solutions or something self-made. And then you can see how it goes.

Content

• AI Consulting with a Focus on Marketing, Sales, Office Automation and Organization
• My Approach as an AI Consultant
• Open Artificial Intelligence, or closed?
• AI consultant for AI evaluation
• What is AI consulting?
  • What is the scope of AI consulting?
  • AI Strategy Consulting
  • What does an AI consultant do?
  • AI consulting can also be IT consulting
  • AI solutions in Enterprise: 4 Major Applications

AI Consulting with a Focus on Marketing, Sales, Office Automation and Organization

As an AI consultant with the background of a creative marketing / advertising / content / copy freelancer, I support you in the context of AI consulting to introduce marketing-oriented AI tools and AI-supported processes in the company.

This starts with an inventory of the requirements of your teams, but also with the capabilities (and expectations) of the individual team members.

This is followed by an evaluation of current processes and possible Gen AI tools.

In a pilot program, the first key users are trained and shall use the new software. Then we have to see if the solution delivers in terms of saved time or better outcome.

Actually, this is AI business consulting, and with sense and reason.

Example: Adobe Express or Adobe GenStudio are, sort of, of “advertising agencies as a desktop app”, but to get great results with it, you should be an art director / media designer / content creator. The same applies to Canva, Figma, or whatever else is on the market.

For project managers in marketing departments, these tools might be unusable. In reality, they may only need a tool to write briefings faster, and in great quality. But that can go wrong. I’ve seen (and had to deal with) AI generated briefings (for content) that were unusable and moronic. The marketing exec saved time on this, but the time invested was basically wasted. Key take-away: If you don’t work, it won’t work.

And maybe you just need some support to create new language versions of your marketing materials, to enter new markets. This can be done with AI, of course, but that also works with external service providers, for example global transcreation agencies or translators.

It’s always better when a native speaker has a look at your foreign language marketing copy. You see it when you read this article: It’s AI translated (Anthropic Claude 3 Opus, which is not bad), with a custom prompt and custom settings, and heavily fine-tuned – but still, it lacks the oomph a native speaker copy writer would manage to create.

My Approach as an AI Consultant

As always, the first step should be taken first. That’s true for AI consulting as well, and the first question should therefore be: “What do we actually want to achieve?”

This works rather good via user stories, i.e. small wish-lists from the perspective of the internal (marketing) team and that of the end customer / target groups.

Creating user stories is good task for a workshop.

In the next step, you match the wish-list with AI offerings on the market. Then you test the AI tools with key users. These should not only be the nerdy high-flyers, but also traditionalists.

Then the test results should be evaluated before the – possible – big rollout.

You should plan for a) training for the teams and b) perpetuations and reviews.

Maybe the tools will be outdated again in 3 months, so you can’t sit back and relax.

What is the big advantage of this consulting approach? You avoid bias.

Example: Someone in the team says “Tool XY is great, I love it, we’re introducing it”; same is true for AI tool offerings from consultancies, simply selling what they have on stock. Unfortunately, the software might be botched and only delivers good results for a special use case, and only if you have a knack for it.

For a large rollout, other use cases may be important, and operation must be simple.

Likewise, especially when it comes to AI image creation, copyright is an underestimated issue.

Open Artificial Intelligence, or closed?

Another aspect is closed source / open source when it comes to artificial intelligence. The elephant in the LLM room, Open AI, is closed source, as well as e.g., Anthropic. Closed source means that it’s a black box. The problem is that black boxes might be changed without notice, and your product / solution based on that black box might not work any more as expected.

Open Artificial Intelligence can be found e.g., with some LLMs of Mistral. Chinese LLMs are partly open source, but you are not allowed to use them commercially. Key take-away: Read legal stuff.

AI consultant for AI evaluation

It may also turn out in the context of AI consulting that generative AI is not the solution. Perhaps automations – machine learning – are better suited to relieve teams.

To answer customer (service) emails, a popular issue, there are many solutions: Do you use a CRM tool? Do the emails run in Outlook? Should an automated analysis of the customer’s (bad) mood be done first (this can be done with the Microsoft Power Platform, among others)? Or is a solution based on copy & paste sufficient? Or do you feel like building a small app with Python to feed GenAI (possible with Anthropic, among others)? Questions upon questions!

Since as a creative director I can quickly recognize whether creative results are good or suboptimal, I can support you in selecting tools that really deliver good results.

Get a quote for AI consulting

Plan for some four to five figures costs.

What is AI consulting?

AI consulting is a service at the interface between management consulting, IT consulting, organizational development and creative consulting. In the context of an AI consulting mandate, software solutions from the areas of process automation, machine learning and artificial intelligence are selected by the AI consultant and implemented in the company. AI, i.e. artificial intelligence, is often misused as a buzzword – to sell AI products. But it is more important what goal you want to achieve by means of software. An example for AI in retail: A consumer goods company has data from its own online store, from the Amazon store and from orders from (brick-and-mortar) retailers. Machine learning AI tools can help with comprehensive data analysis and support forecasting. On top of that, Generative AI is then used to “talk” to the numbers. AI consulting can include the selection of tools and technical implementation here, but also creative support for correct prompting.

And still, one question remains: What will you do with your new insights?

What is the scope of AI consulting?

The scope of AI consulting can be minimal, for example in the context of AI workshops. The insights gained in the workshop are then independently transferred into AI solutions by the company’s workforce. Further on, additional AI training is then helpful to foster the use of the new tools in the workforce.

AI Strategy Consulting

Larger mandates are more like AI strategy consulting, where it may turn out that one or the other department no longer needs to be as large in the future – or that it is better to sell a complete business unit.

Possible reasons:

• Document processing: AI can do this, including – for example, in accounting – matching contract contents, orders, delivery notes and invoices.
• Customer support: AI can already handle (simple) call center tasks quite well. Let’s imagine the next level, namely retail. The DIY store app (or the physical bot) that you can ask for article XY and then get the shelf name? Not bad. It will, of course, fail, because 80% of the shelf assignments are wrong.
• Field service: Field service management will become more predictable, allowing a) fewer people to be needed and b) higher margins to be possible. AI strategy consulting can show that the business unit can / must be restructured.

What does an AI consultant do?

An AI consultant can be a management consultant, an IT consultant – or even a creative freelancer. In the context of an AI project, he or she can help screen the AI market, create a roadmap, help select tools, or implement (cloud) tools. Training on the new software solutions is also part of an AI consultant’s service spectrum.

AI consulting can also be IT consulting

Larger AI consulting mandates take place in the area of IT consulting: To upgrade an ERP or CRM software with AI tools, for instance, coding is definitely required. The same applies to e.g., setting up a customer service chatbot that is supposed to access catalog data.

A promising in-between thingy are no code / low code applications such as Microsoft Power Apps or special chatbots such as Microsoft Copilot for Finance. This makes business intelligence usable with natural language thanks to generative AI in the background.

In the Salesforce world, this tends to work via the embedding of Tableau (which belongs to Salesforce) and Slack. The goal of the big tech companies is, of course, that ultimately every employee in your company needs a cloud subscription for a few hundred moneys per month.

AI solutions in Enterprise: 4 Major Applications

• AI for Machines: “Industrial Artificial Intelligence” and “AI in Manufacturing” can be seen as a the next big thing after the Internet of Things / IoT. The promise of the “Smart Factory” could become reality with this. The intelligent factory previously consisted of, for example, customer orders running directly into production via CAD / CAM. The even smarter AI factory of the future then no longer needs specific instructions at all, but manages with high-level orders, such as “Here is the photo of my workshop, build me a fully equipped control cabinet”. An example from an end user perspective would be, for example, a car that detects maintenance needs (or defects) itself through sensors and, based on this, considers a workshop appointment necessary, arranges this to fit into the user’s life, and orders the required parts in the workhop’s ERP system. At the same time, it is checked in the talent management software whether the workshop team is trained for the specific problem. If not, the team gets an online course put into the shift schedule. Recommended reading: It’s worth taking a look at the Industrial Artificial Intelligence Journal from Springer.
• AI for Automated Administration: Automate repetetive, boring tasks? Document management? Create shift schedules? Individualize CRM? There are numerous offerings. The next big thing will most likely be inbound and outbound call centers, where you chat fluently with ChatGPT voice chatbots. I’ve already seen an integration via Retell AI in HubSpot. Purpose: appointment scheduling and (self) service, i.e. first and possibly second level.
• No code / low code as AI simplifiers: With solutions like the Microsoft Power Platform, you can already build small software solutions for business with little to no programming. Document handling and photo tagging are examples here. The market needs to be watched. In addition, there are process linking services like Make and Zapier, which are the keys to universal modular systems.
• AI for Users: Research topics? Improve marketing? Review and summarize applications? Create presentations? Offer catalog chatbots for customers? Generative AI helps with this. The diversity is overwhelming. And only slowly the fog is lifting with regard to copyright and ethics: Some “text generators” like Chat GPT were trained with, among other things, millions of YouTube videos, without asking the creators, which means that the generated texts could be full of copyrighted material. This cannot be checked.
  Wikipedia texts are also not suitable as “royalty-free” training data, as they are published under CC BY-SA 4.0 – this means that the author must be named by.
  In the US legal system, things look different, as there are significantly softer regulations due to “fair use” rules, which do not exist in the e.g., EU. However, I assume that the industry of plagiarism scanner providers will see a market here and develop corresponding offerings.
  On the AI provider side, the large language models (LLMs) will then be trained to defend against this, either by not copying & pasting from copyrighted data, but cleverly rephrasing, or by neatly quoting, in a scientific manner (Copilot already does this to some extent).
  Above all, however, the “teaching material” for the AIs will be converted to licensed material, for example via (expensive) cooperations with (newspaper) publishers. And it’s similar with AI image generators; only slowly is the industry coming to the realization that training data must be licensed beforehand – Adobe, Getty / iStock, Bria and Apple are already doing this.