Compliance checking is very important - it makes sure things are right and delivered on message.

It tends to be difficult, it can be tedious and it is easy to make mistakes.

Most organisations tend to have numerous roles involved, each with many team members. This means co-ordinating the compliance process is also part of the value of automating a compliance procedure.

To bring this "how to" guide to life. Let's use an example of an automated labels compliance tool that automates checking of label artwork against a product specification sheet and label rules for a particular market and product category. Let's assume the current process uses spreadsheets for rules checking, PDF files of labels and approval chains that are completed using email.

Key benefits of automation:

1. Consistency of results - manual checking can produce different results depending on who does it, when and under what time pressure. Machine checking generates always generates the same output.
2. Quality of checking - machines are meticulous, they do not forget and do not accidently miss things. The quality of checking is much higher for automated processes. However, they do rely on good quality inputs.
3. Using the right checklist - automation ensures that the right checklists are used. Manually checking could mean rule checks are missed or extra irrelevant rules applied.
4. Time savings on checking - machines can run multiple rule checks and generate automated reports in seconds. Manually checking can easily take 15-30+ minutes.
5. Time savings on process approval: co-ordinating and following-up with people involved in the process is much more efficient if managed in a co-ordinated tool rather than relying on email.
6. Reduction in risk - lower errors and consistent outputs mean less risk of "bad" approvals.
7. Reduction in wasted downstream spend - "bad" approvals mean wasted print runs, reputational damage and in some cases lead to litigation.
8. Better environmental impact - less printing, less activity time per document and less downstream wasted activities.
9. Allows staff to focus on verification of more standard approvals and focus expertise on reviewing "edge cases" or more difficult / nuanced approval decisions.

In summary, the key drivers are quality increases, cost savings, risk mitigation and better employee engagement.

We follow these steps to quantify the benefits:

• Select areas of the business with the largest volumes of approvals first.
• Focus on approvals that have the highest number of iterations (error detections and reworking).
• Quantify how long the checking currently takes per approval and per iteration and compare to the new way of working. This should generate savings of around 70%.
• Quantify how many "bad" approvals got through the process and what the cost of remedy was. The new way of working should aim to reduce this by 50%+
• Quantify how long an approval chain (reviewer to approver 1 to approver 2, etc.) takes and how the time can be reduced by linking automated approvals to automated process approval. The new way of working should speed up the process by 30-50%. This means products can be sold earlier. This impact of this should be quantified.

So that covers the benefit line of the business case. The costs should include solution development costs, hosting, support, maintainance and rule review and retraining costs. When these are mapped out on a monthly basis over a 3 year period, and if there are sufficient approval volumes - it will produce a compelling business case.

Done in the right way, moving to an automated compliance processes will help increase checking quality, reduce costs and reduce risks. It will also generate a strong financial return and should be near the top of the development "to do" list.

The post Business cases for automated compliance appeared first on Algospark.

Label compliance isn't just about checking boxes; it's a multi-dimensional process that involves accurate claims and ingredients, (in the case of food and drinks) nutritional information, avoiding confusing statements, regulatory rules, brand policies and guidelines.

The risks associated with incorrect labeling span health risks, legal consequences, financial impact and brand damage.

Given the complexities and high stakes, we encourage a migration to automated label compliance systems. This means a move away from manually compiling relevant requirements and then manually going through checklists. An automated approach means that compliance professionals can focus on verifying, ie confirming that automated checks have been performed correctly. This is made even easier if automated label checking tools can highlight areas on the label itself that have been checked and confirm that it is compliant or needs to be fixed.

We find that time saving from automated labels checking typically comes from:

1. Correctly complied check list - have the right specifications automatically complied based on the market the product is to be sold into and the details of particular product specifications.
2. Generating a list of non-compliance items – an automated review of each rule with a pass or fail outcome and details of why it failed, eg wheat missing from the ingredients list, incorrect volume amount, missing recycle label, etc.
3. Identifying the specific area on the label – being able to quickly see why the rule has failed (and where) or if it has passed, where is it on the label.
4. Automatic remedy comment generation - so that the comment can be easily relayed back to the label artwork producer to remedy the error. This could be a paragraph of all errors or detail on the specific error and where it appears (or does not appear where it should).

So how is automated checking more consistent and detail focused than human checking? Using advances in applied AI and document review there are several techniques that can be used to drive significant increases in review consistency and time savings in manual label review.

AI-powered computer vision can analyze label artwork to ensure that text size, positioning, and overall design adhere to both regulatory and brand guidelines. It can detect errors that might be missed by the human eye. Natural language processing techniques can review the language used in claims, ingredient lists, and nutritional information to ensure it meets regulatory standards. Large language models can cross-reference label contents with regulatory requirements across different countries. Other applied AI techniques can understand the context in which information is presented, ensuring that claims are not misleading and that essential warnings are prominently displayed.

In summary, using applied AI in label compliance processes can achieve significantly better standards of accuracy and efficiency. Automated labels compliance helps better safeguard consumers and improve the efficacy and efficiency of the label compliance process for organisations.

Algospark is an applied AI solutions provider that has built numerous compliance solutions. Read more and see our label checker demonstrator here: https://algospark.com/frameworks/label-checker/

The post Product label compliance using applied AI appeared first on Algospark.

Clear scope

• There is wide scope for using AI (innovation and process efficiency)
• Use clarity in scope and purpose (why, what and how)

Planning

• Set modest early goals initially focused on a subset of products for part of the business
• Involve service “end users” early and iterate deliverables frequently
• Use phasing (early prototype / static example / dynamic solution)

AI has to be business area led - not technology led and needs to be supported by a proficient data and IT team. Don't start in house if you do not have a solid AI team. It is best to use applied AI specialists to work on early ideas and develop an applied roadmap. Remember, early AI is experimentation, so keep proof of concept projects focused and short. Once early prototypes demonstrate value, transition to in-house capabilities as the projects grow and solutions can then be rolled out across the business.

Develop applied AI solutions as a standalone capability - avoid developing an integrated system capability from the outset. In other words, develop AI in a sandbox and where possible, use cloud systems that can easily be set-up and scaled or turned off. Do not let technical jargon or existing data and technology get in the way.

Team

• Good high level design skills are key: what is it, why is it valuable, how can minimum viable product be developed
• Keep teams small, but ensure a strong mix of skills spanning business understanding, AI and application development
• Keep learning! Things change quickly.

Engage experts - it is unlikely you will have the upfront design and implementation skills. Make sure that your delivery team includes skills spanning :

• Business understanding: business process understanding, analysis skills (where are opportunities) and high-level business case (finance) understanding
• Data analysis and engineering: understanding and impact of data stores (SQL/JSON), data pipelines (ETL) and dashboards (PowerBI / Tableau)
• Algorithms, models and applied maths: know the best approaches of model selection, considerations and testing. Knowledge of Python (possibly also R).
• Cloud management: set-up of data storage, compute, networking, administration and security.
• Application development: putting models into use (Python and front-end web skills)
• Project management: agile planning, managing, monitoring and reporting

We hope this helps. Please reach out to us at Algospark if you would like to discuss how you can get the most out of applied AI.

The post Applied AI - key considerations and lessons learned appeared first on Algospark.

Anyone can built a dashboard, then add another and another. This is not getting the most out of your data. This is simply presenting the past using several lenses and perspectives.

Likewise, anyone can write a quick note or insight using Chat GPT, Claude or any other Large Language Model (LLM) . This is very useful and saves lots of time, but is essentially copying what others have done and presented before. The real value of text understanding in hospitality is in document synthesis which can deliver much more value than auto-generating reports.

AI in hospitality in not about lots of dashboards and using Chat GPT to quickly generate marketing messages.

True applied AI in hospitality is about better informed decision making for the right people at the right time. It is also about helping reduce administrative processes and letting staff focus on service.

The source of all success in applied AI is business transformation, ie allowing an organisation work better using a better mix of staff, tasks, data and technology. When all working well together, processes are easier, quicker and outcomes are higher quality. This means staff are more focused on the important things and customers get better service. This has knock-on effects with better HR, more efficient supply chains and less need for large teams of analysts.

AI models excel at handling complex models that are fast changing. Linking models across domains such as sales forecasting, ordering, inventory, production, staffing and supplier management means that good decisions can be shared across domains to make the whole organisation even more efficient.

Existing hospitality systems are not typically linked across domains and do not typically have AI intelligence layers that determine what the best action to perform at any given time. Applied AI means identifying the areas of a business with most opportunity to improve and then building a prioritised plan to roll out the technology and ways of working to get the most out a "joined up" plan. In technical terms, using a "systems approach" to problem solving and then building the technology in an "agile iterative" way alongside the teams using it will mean that "change management issues" are minimised and the organisation will be on the "best path to value".

So what does this mean...

Here are a few things to consider in the applied AI journey in hospitality:

1. Co-ordinated product sales forecasting across ordering, staffing and production. This means you serve fresher food, waste less and have the right people working at the right time. Read more about this here.
2. Staff support systems such as training requirements, staff selection and retention.
3. Supplier management
4. Customer understanding, marketing campaigns and loyalty schemes.
5. Data collection is at the heart of great AI systems. This will be be from databases from existing tools. More advanced systems are able to use raw documents (including emails) and video to capture new sources of valuable data for proactive decision making. Read more about extracting meaning from emails, surveys and documents here. Read more about using cameras for stock takes, production activities and customer activity here.
6. Decision support framework for new site investment. Predict sales, store characteristics and trading patterns for locations throughout the UK. More about this here.

Summary benefits from applied AI in hospitality:

• No more spreadsheets and disparate systems
• Data driven marketing and loyalty scheme management
• Better staff engagement
• Increased staff productivity - right roles at the right time
• Increased product availability and optimised waste levels
• Better inventory and supplier management
• Improved product freshness - right products at the time the time
• Higher customer satisfaction- faster service with fresher products
• Improved profitability

Algospark are experts in applied AI in the hospitality sector. Get in touch if you would like to learn more and get started on your applied AI journey.

The post Delivering applied AI for the hospitality sector appeared first on Algospark.

Do you work in Marketing or Legal and would like to improve and speed up your compliance processes? We would like you to join us and take advantage of our Compliance Copilot beta launch. It is an easy use yet sophisticated tool that automatically checks all ASA rules using leading edge AI techniques. The tool provides decision consistency and saves significant time and money. We believe it is a significant asset for teams wanting to improve the quality of compliance and expedite compliance processes.

Algospark has been developing compliance solutions for marketing, labels and promotional materials for several years. We are an AI specialist that has worked with numerous organisations to improve their compliance procedures. We are now giving you the opportunity to use and feedback on our Compliance Copilot tool.

To learn more: https://algospark.com/frameworks/compliance-copilot-uk/

To go straight to registration and get started: https://copilot.algospark.com/

Want to discuss this further or adapt it to suit your particular requirements? https://algospark.com/contact/

We look forward to helping you with your compliance processes and hearing your feedback!

The post Free leading edge AI marketing compliance tool! appeared first on Algospark.

Lots of questions - but first one is what really matters. The other problems are solved by working with people that have built value propositions, business cases, project plans, data maps, AI designs and then developed and delivered successful applied AI solutions.

AI is in the news every day. It is a popular topic and means a lot of different things to different people.

It is easy to get caught up in hype.

• Company X is doing Y with AI and we are still using Excel.
• My job will be done by robots in X years.
• We can't use AI because it is dangerous and re-enforces bias (or anti-bias).
• We will all be slaves to a super-intelligence.

The reality is that huge computing power and data availability have unleashed the power to understand data. Not just data itself, but meta-data and relationships between data (graphs). This means that:

• Information can be retrieved, organised and presented much more effectively (eg ChatGPT and the Large Language Models (LLM's) behind them).
• Forecasts and predictions are much more accurate as there are more data points to inform decisions.
• Outlier detection is much easier as errant data points and their dependencies and impact can be calculated much faster, much easier and be presented in real-time using interactive dashboards.
• Classification and identification of anomalies in data tables, images and text is also much easier.
• Processes can be optimised much easier and people focused on "edge cases", (ie difficult decisions) rather than mundane checking.
• Analysts and decision makers spend less time to decision and make better decisions using data supported decision frameworks.

So where to start? Applied AI is really routed in business transformation. If an organisation or even a process is not willing to change it's ways of working then even the most value adding applied AI will not succeed. The most important thing is to start with an "innovation outlook" and be prepared to map out a brave "future state" or new way of working. Determine how you currently do things and then look for a subset of the challenge to prove out how applied AI and new ways of working will deliver for you. There is no need to change IT systems or processes at the outset as all development and piloting can be done in a "sandbox" or development environment. The design of an applied AI project, the business case and an early working model is the most important piece. Or put another way, What I am trying to do, why (early ideas of the value opportunity) and how can it be done in the shortest possible time to deliver credibility and acceptance that this can make a positive impact. You should discuss this with applied AI experts. A small upfront investment will save you significant time and money down the path.

We often hear - can I buy AI off the shelf? The short answer is no. And the real question should be - what problem am I trying to solve? Picking a technology and then thinking about a problem to solve is the wrong way around. I have heard the argument that - some products have lots of AI, so you can technically buy it off the shelf. Again, you are buying a product that is enabled by AI, you are not buying the AI per se.

Should you care about predictive analytics vs generative AI? Not really. They are used for different things. Predictive analytics is to help you make better forecasts / classifications and therefore decisions that are optimised for these outcomes. Generative AI takes a probabilistic word, sentence and paragraph structure approach to answering a question based on training huge power hungry algorithms on millions of documents. This means that it is able to answer questions very well based on what it has seen and adaptations to responses based on what it is told is a good solution. They are both very, very useful and are invaluable in applied AI along with other models and approaches.

Is your IT or data team spending the budget in the most effective way? If they:

• Have experience in applied AI and have already delivered successful solutions
• Have a business case and can demonstrate return on investment
• Use development sandboxes with an agile iterative approach and "deliver minimum viable product"
• Have a good knowledge of changes in application development, innovation in data engineering / processes and innovation in data science
• Work closely with "product owners" and "change leaders" in the business

Then the answer is probably yes. Unfortunately many organisations do not benefit from this experience and approach in-house.

So, ready to leap ahead with applied AI. Speak with us at Algospark.

The post Leap ahead with applied AI appeared first on Algospark.

Benefits from automated advertising compliance:

1. Check the right rules every time: product specific, code of conduct and regulation
2. Consistent decision making: get the same output irrespective of who has reviewed, under varying deadline pressure, from where and within which context
3. Time saving: 60-70% compliance checking reduction
4. Enabling compliance checks across the wider team: make preliminary compliance checks earlier and across Legal, Marketing, Content Creation & Operations.
   1. Catch bad early
   1. Reduces risk
   1. Reduces re-work (time and cost)
   1. Speed up decision making
5. Better content analytics: what is failing, why and how often.

Get in touch to discuss how we automate advertising compliance.

darren@algospark.com

More: https://algospark.com/frameworks/compliance-solutions

The post Drive benefits from Automated Advertising Compliance appeared first on Algospark.

Marketing content is being created at faster pace and regulatory rules are becoming more stringent. Successfully complying with a code of conduct, brand rules and regulatory rules takes more time from marketing and legal teams. In addition, there are different considerations for mapping relevant rules across different products, channels and markets.

Key challenges:

• Volume of work
• Complexity of rules
• Consistency of decision making
• Checking across multiple sources of rules
• Scope for error and exposure to risk

Using automation for a first pass validation can save up to 70% of review time. It allows legal specialists to focus on validation rather than the more tedious elements of starting from a blank piece of paper, pulling together compliance frameworks and cross-referencing rules manually, one by one.

Forward thinking marketing and legal teams are focused on reducing the amount of time taken for marketing compliance, and empowering the wider team across legal, marketing and agencies to be compliant and reduce risk. Or put another way, they are moving towards a compliance first way of working and moving away from compliance as a last step bottleneck. Automating the compliance process delivers both of these benefits.

The rise of Large Language Models (LLM’s) and the popularity of ChatGPT has underlined the power of using generative AI. However this is one part of the approach to automated marketing compliance. There are numerous considerations in the design of the solution and the techniques used to successfully deliver automated marketing compliance with consistently high accuracy rates. There are also a wide range of models and approaches to get the best results. These span image style models, object detection models, word detection models, word meaning / understanding and sentiment models, large language models and word embedding approaches for context understanding and translation.

Effective automated compliance solutions for large organizations are not sold as products. We work with innovative organisations to design and deliver automated marketing compliance solutions. Our solutions span advertising images and copy, automated label compliance checking and document compliance using our Compliance Engine.

Get in touch to discuss how we can help your team benefit from our experience of delivering high quality automated compliance tools.

The post <strong>Enabling Automated Compliance for Marketing</strong> appeared first on Algospark.

A lot has changed in AI. Recent advances and approaches mean that what was leading edge approaches 1 or 2 years ago have now been superseded. This also means that data science skills and requirements have also shifted dramatically. For example, no-one builds a chat bot from scratch or builds object detectors from scratch nor checks text compliance using LSTM deep learning networks. In summary, AI is eating AI. Recent developments mean that applied AI and how to get the most from it have evolved. The balance of skills has shifted to design, implementation and iteration rather than model build and hyper-parameter tuning. This means you should not hire a “data science team” and expect them to prioritise and solve your business problems. What you need is applied AI specialists. This will get you fast tracked in the right direction and give you a roadmap for how to build and evolve “smart applications” driven by AI. It will also help you determine what should be done in house and how you should use external specialists.

There are typically three ways of accelerating time to value from applied AI. The right approach depends on where you are in your data journey and experience with AI.

If you are just starting out, for example, evolving from silo-ed Excel based reporting, then you should use a fast track transformation approach:

1. Determine your key drivers and key performance indicators
2. Aggregate and link your data (drill-through reporting)
3. Forecast what is going to happen - predictive analytics
4. Map what to do – prescriptive analytics
5. Determine a prioritised road map for AI development

If you know where you should focus first, then the second approach is to add intelligence to your existing processes and tools. For example, a data driven ordering tool that links evolving sales predictions with order constraints to optimise what and when to order from who.

The third approach is more leading edge and exploratory. This approach defines new automated ways of working and pilots new tools to get the most from leading edge AI. Examples could include automated document summarisation, image compliance checking or service personalisation.

Algospark are applied AI specialists. Our team have experience spanning data science, analytics, data design, data engineering, software development, service design, business analysis, finance and technology. We can help and advise the best way to get started and build solutions that get you and your team fast-tracked in the implementation of applied AI. Get in touch and let us help you deliver for Xmas and well beyond!

The post All I want for Xmas is a data science team... or is it? appeared first on Algospark.

Using linked datasets and predictions can help you quickly identify what is an outlier (or exception), how far it is from an expected value and then how much resolving the issue is worth to the organisation.

Linking datasets to allow drill through reporting dramatically reduces the number of reports and creates the basis for quality data investigation and interrogation. Using applied AI to derive predictions and then prescriptive actions (recommendations) helps define the value of outliers, how to prioritise them and what to do to resolve them.

We have used this approach at Algospark across various industries ranging from airlines to social care to retailers. Unraveling value by identifying outliers and recommending remedies is a key driver behind next generation analytics, ie empowering your staff to get more from your data and systems.

Get in touch with us to help your unravel value from your data and empower your staff with next generation analytics.

info@algospark.com

The post Unraveling value with applied AI appeared first on Algospark.