About a-team Marketing Services

A-Team Insight Blogs

Wand.ai Software Update Seeks to Democratise Use of AI

Subscribe to our newsletter

Software developer Wand.ai has rolled out an enhanced version of its artificial intelligence (AI)-based platform that’s aimed at enabling even the least tech-savvy operator to harness the benefits of machine learning and generative AI (GenAI).

The updated version of the collaborative multi-agent software adds broader agent capabilities and updates the user interface. Wand.ai co-founder Philippe Chambadal said the platform, which is being utilised by insurance and asset management companies, is dedicated to making AI more inclusive.

“The idea is to let knowledge workers who know nothing about data science, nothing about coding and nothing about prompt engineering to leverage both machine learning AI and GenAI immediately just by using natural language,” Chambadal told Data Management Insight.

Empowering AI

Agentic AI is relatively new but is seen as the next step in empowering AI applications to do more than their models permit. Already agents enable interaction with other applications to aid in tasks such as planning and remembering chat-based prompts.

US-based Wand.ai calls its software a “multi-agent cognitive layer” that acts as an intermediary between users and AI models. It’s built on the principle that no AI agent can properly instruct a model to do everything it’s asked. However, if an array of agents specialised in specific topics are tasked with collaborating on queries, then users will receive better outputs. Wand.ai provides the connective tissue between those agents.

“You can see our platform as a brain sitting on top of a bunch of agents and GenAI,” said Chambadal.

More Agents

Via a chatbot architecture, users enter their query into Wand.ai and the software will determine which combination of agents is best suited to provide a solution. The updated version launched last week expands the number agents and widens their expertise, enabling faster responses to more complex queries.

Wand.ai doesn’t build the models, but it lets users target specific data sources without moving the data. While the platform doesn’t gather or store data, it does provide the lineage of the information it uses to produce its outputs.

“This lets users assess the reliability of the source,” Chambadal said.

“It’s not a black box; you have complete transparency into the code that’s created,” he said, adding that this ensures that users can trust the results generated by the AI, knowing exactly where the data came from and how it was processed.

Chambadal co-founded Wand.ai after working in a large European banking software company where he became frustrated with the absence of tools to help lenders make the most of the “gold mine” of data that they own. He was struck by studies that claimed bank AI models were not only being built at a glacial pace but also that they had a success rate of no more than 15 per cent.

Taking Action

Almost four years ago he decided to do something about that and with a business partner set up Wand.ai.

“We had a common vision that if you want to scale AI and really leverage it in the enterprise, you shouldn’t use data scientists – it has to be a self-service, no-code, no-data science platform that knowledge workers can use immediately and solve problems immediately, again, without the need for data scientists,” Chambadal said.

“Data scientists are great but they should be solving real problems like programme trading; if you’re trying to predict churn or next-best product, why use data scientists? We can always produce better and 100-times faster results than any data science team for these mundane problems. It’s the 1 or 2 per cent of difficult problems that data scientists should sort out.”

Among the general tasks that Wand.ai has built agents to manage are data cleansing and management, but its latest iteration includes agents that are focused only on industry-specific use cases, including risk-mitigation for insurers, as well as memo analysis for venture capital and private equity. Banks are also hiring Wand.ai to build deal books, Chambadal explained.

The software can be accessed via APIs, the cloud, virtual private clouds and through on-premises setups. The latter two are most common among banking clients, because “they don’t want the risk of data leakage”. Queries can be input directly or via email.

Higher Speeds

The most apparent benefit of using the software, said Chambadal, is the acceleration of task processing times. Analyses that would normally take days can be crunched in a couple of minutes, he said.

Chambadal conceded that no AI model will get things right all the time and for that reason future versions of the software will bring humans more closely into the process, providing expertise when the models can’t find an answer. It’s a concept that Wand.ai chief executive Rotem Alaluf expands on in interviews. During one, posted to the company’s website, Alaluf said human involvement was crucial to providing guardrails around AI.

“We want to foster collaboration between people and agents,” Chambadal said. “We want to bring that collaboration to a higher level.”

Subscribe to our newsletter

Related content

WEBINAR

Recorded Webinar: AI in Asset Management: Buy-Side Attitudes toward GenAI and LLMs

Since ChatGPT exploded onto the scene in late 2022, financial markets participants have been trying to understand the opportunities and risks posed by artificial intelligence and in particular generative AI (GenAI) and large language models (LLMs). While the full value of the technology continues to become apparent, it’s already clear that AI has enormous potential...

BLOG

FINBOURNE Technology Expands Integration with Taskize to Enhance Post-Trade Operational Workflows

FINBOURNE Technology, the investment data management solutions provider, has expanded its integration with Taskize, the post-trade operational workflow specialists, to improve the management of complex IBOR (Investment Book of Record) and ABOR (Accounting Book of Record) post-trade exceptions. For FINBOURNE clients, the integration with Taskize aims to reduce resolution times, lower human error, and boost...

EVENT

Data Management Summit New York City

Now in its 15th year the Data Management Summit NYC brings together the North American data management community to explore how data strategy is evolving to drive business outcomes and speed to market in changing times.

GUIDE

Regulatory Data Handbook 2024 – Twelfth Edition

Welcome to the twelfth edition of A-Team Group’s Regulatory Data Handbook, a unique and useful guide to capital markets regulation, regulatory change and the data and data management requirements of compliance. The handbook covers regulation in Europe, the UK, US and Asia-Pacific. This edition of the handbook includes a detailed review of acts, plans and...