Appian wants to bring Private AI to companies that do not intend to transfer their data to third parties (due to the use of artificial intelligence-based solutions that also evolve thanks to the use of user and company data) or use pre-trained models with data not “adherent” to your business needs
That of one Private AI (Private Artificial Intelligence) is not a new intuition, it has been talked about since around 2017. The basic idea is develop artificial intelligence techniques – and application technological solutions based on them – while preserving the privacy and confidentiality of data. In other words, by leveraging one of the main current concerns of companies in terms of data protection and safeguarding, not only for regulatory compliance needs but also to mitigate risks.
In very simple terms, a Private AI allows organizations to exploit the potential of artificial intelligence (today no one doubts the enormous benefits it can generate) while maintaining control over their data, guaranteeing its protection and privacy.
And it is on this playing field that Appian intends to compete on a competitive level, as widely reiterated by the company’s founders during the recent Appian World 2023the annual convention that brought together customers, partners, analysts and journalists from all over the world in San Diego (California).
Matt Calkins, CEO of the company was very clear in explaining the company’s choice: «When today we talk about solutions and tools based on artificial intelligence, such as ChatGPT and the other OpenAI solutions (now on everyone’s lips), we are referring to Public AI: companies (and people) send their data to these usable environments via web and via cloud in order to have – in exchange – a service (for example, an answer to a question formulated in natural language or a graphic proposal in response to a command prompt). That kind of AI is made “for them” (the providers), not for companies».
Calkins believes this approach won’t work for long, «companies are already looking for alternatives, they want their own AI, they want to be able to train models on their data and deploy the solutions within their enterprise environments» says the number one of the company.
And to those who asked if this “new” vision of the company could be read as a repositioning, the CEO’s seraphic response left no room for misinterpretations: «We are an end-to-end process automation company, everything we do goes in this direction».
Appian Private AI
The Appian World 2023 convention was thus an opportunity to announce some innovations precisely on the front of this approach to Private AI.
First of all, like many other IT providers already, Appian has also decided to integrate ChatGPT into its platform, «addressing privacy and transparency issues surrounding Large Language Models (LLMs), allowing companies to use their data in a private environment», explained Malcolm Ross, Senior Vice President Product Strategy at Appian, «especially for allow companies to design and build automations in an even simpler and faster way, taking advantage of natural language and Generative AI to accelerate development».
Some practical examples of using the natural language based on ChatGPT within the Appian platform range from the identification of patterns and schemes that can be reused in the design phases (always with a view to low-code development), up to the complete development of a application by exploiting the Large Language Models (the algorithms create and suggest, on the basis of simple questions and indications from the developers, all the objects necessary for the construction of a functioning application).
«This by no means means that AI replaces people’s work – Matt Calkins is keen to point out – but who becomes a real partner in corporate teams».
Michael Beckley, CTO of the company, presented the first features of Private AI – defined AI Skills – available to companies, who spoke of «algorithms that Appian has developed for some very specific functional areas. Companies can deploy and apply these algorithms to their data, train them through the Appian AI Skill Designer environment – on Appian Cloud – then use them within their own enterprise context».
These are the first AI Skills presented in San Diego:
1) Email classification: the algorithm that allows to automatically classify and interpret e-mails can be used to automate some communication processes. The algorithm must be trained with your company data (as mentioned, within the Appian AI Skill Designer on the Appian cloud) and then connected to your mail server;
2) Classification of documents: in this case the algorithm is designed to automate the classification of documents based on their content, data and information they contain. Again, once trained through your company data, it can be integrated to support the automation and management of document processes;
3) Extraction of documents: extension of the previous functionality, this algorithm allows to extract information from various types of documents with the main objective of reducing manual data entry.
Faced with the tantalizing features that we have seen in recent months, especially in the field of Generative AI, these applications may seem less exciting, but «companies don’t want something exciting, they want something that works and solves real problems» says Beckley bluntly.
breaking latest news © (Article protected by copyright)