Extend unveils open-source AI toolkit for smarter finance
Extend has released an open-source AI toolkit aimed at enhancing how businesses manage and analyse financial data.
The toolkit supports multiple frameworks, including Anthropic's Model Context Protocol (MCP), OpenAI, native integration with LangChain, and compatibility with CrewAI to facilitate complex multi-agent workflows. The company states that this versatility allows businesses to incorporate Extend's API seamlessly into their existing AI-driven systems, enabling more advanced spend analysis and automated finance processes.
The toolkit is designed to offer flexibility to businesses, allowing them to interact with Extend while continuing to use their preferred banks or credit cards. Its intention is to help organisations adopt AI solutions tailored to their needs, supporting functions such as intelligent financial queries, custom reporting, and workflow automation.
Jonathan Bailey, Extend's Chief Technology Officer, commented on the motivation behind the toolkit: "When I started to explore the multitude of use cases for AI in our industry, I zeroed in on the power of 'agentic frameworks', and realised we could enable tools like Claude to interact directly with Extend via our APIs and immediately unlock extensive AI functionality for our customers."
Through the integration of these frameworks, users are able to query financial data using natural language input, conduct advanced analytics, and generate custom reports. Automation powered by AI agents can manage tasks such as expense categorisation and budget tracking. Businesses will also be able to analyse spending patterns, identify cost-saving opportunities, and obtain greater insights into areas such as cash flow, team spending, and overall budget allocations.
Andrew Jamison, Extend's Chief Executive Officer and co-founder explained the broader company strategy: "At Extend, we believe in empowering businesses to do more with what they already have - whether that's credit lines, banking relationships, or software investments. With this AI toolkit, we're taking that mission to the next level, giving our customers the tools they need to make smarter, faster, and more informed decisions."
Extend indicated that development efforts will continue to focus on expanding AI automation features within its platform, in response to increasing demand from companies seeking more streamlined financial management solutions.
Extend is a modern spend and expense management platform that helps businesses gain control over spending - without changing their existing bank or credit card programs. Thousands of companies use Extend to create and manage virtual cards, streamline payment workflows, and get real-time visibility into team and vendor spend. According to the company, Extend powers billions of dollars in transactions while partnering with the financial institutions businesses already trust.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Techday NZ
19 hours ago
- Techday NZ
Straker partners with n8n to automate quality AI translations
Straker has expanded its range of AI-powered translation tools through an integration with n8n, an open-source automation platform used by more than 230,000 active users and over 3,000 enterprise clients worldwide. The newly released Straker Verify node is now available within the n8n environment, introducing key capabilities for organisations aiming to automate content translation workflows. The integration enables users to trigger translations automatically when new content is created, receive real-time quality assessments, and direct specific content to human linguists at Straker for further review if required. Straker's approach targets businesses facing the challenge of maintaining translation quality and compliance as they accelerate their use of AI in content creation and localisation efforts. This is particularly relevant for highly regulated sectors, including finance, healthcare, and legal services, where translation accuracy can have compliance implications. Organisations in these sectors benefit from automation by being able to quickly scale global content delivery while ensuring linguistic accuracy and meeting regulatory obligations. The integration automates the decision-making process by assigning real-time quality scores to translations and routing content that scores below a certain threshold or is considered high priority for expert human review. Grant Straker, Chief Executive Officer and Co-Founder of Straker, said, "As organisations race to adopt generative AI workflows, translation accuracy and oversight cannot become casualties of speed. This integration with n8n ensures businesses can move fast without compromising on the quality their global audiences demand. It is yet another example of Straker's strategy of leveraging third party platforms to cost effectively broaden the reach of our sales efforts and grow Straker's high margin recurring revenue." The integration is consistent with Straker's ongoing strategy to embed its language solutions within widely used workplace platforms. Previously, Straker developed an application for Slack, which allows users to order translations and monitor progress from within the collaboration platform. The company has also announced intentions to offer a similar solution for Microsoft Teams. The n8n automation platform, headquartered in Berlin, is increasingly used as a foundation for building sophisticated AI-enabled business process flows. It offers flexibility through its open-source nature, allowing businesses to deploy the platform on their own infrastructure and maintain control over their data and AI models. According to the announcement, n8n's deep integration with AI models and tools like LangChain enables technical teams to construct workflows that automate research, generation, refinement, and publication of content across channels. Deploying Straker's Verify node within the n8n ecosystem reflects both companies' focus on supporting complex, AI-driven content operations while ensuring workflow flexibility and regulatory compliance. The newly launched integration is accessible to n8n users who register for Verify AI services. This integration with n8n builds on Straker's history of technical development in AI-driven language services, combining automated translation mechanisms with human linguistic expertise where necessary. Both companies highlight the critical importance of balancing speed and accuracy as the adoption of AI in business workflows accelerates globally.
Techday NZ
a day ago
- Techday NZ
Expert warns of ‘AI wrappers' masquerading as SaaS products
A growing concern among tech-savvy users and developers is the rise of HTML prompt wrappers, simple web interfaces layered over public large language models (LLMs) - being marketed and priced as fully-fledged SaaS products. Wrappers often mimic polished SaaS UX, tricking teams into thinking they are buying secure, scalable, enterprise-grade tools. The stark reality is, wrapping some HTML around an AI API isn't a SaaS, it's a pair of AI handcuffs. HTML Prompt Wrapper Explained At its core, a prompt wrapper is a basic HTML (or low-code) interface that collects user input, sends it to an underlying API (like OpenAI, Anthropic, or Cohere), and returns the response—often with little to no added value in between. These wrappers simply use the open APIs of LLM providers, often built in a few hours using standard web frameworks and do not add proprietary models or deep integration. The final product often lacks original functionality, security architecture, or genuine backend services. Yet, they're sometimes marketed as sophisticated AI tools with monthly subscription fees rivaling those of actual SaaS platforms with meaningful IP and infrastructure. In fact, some charge $20–$100+ per month for features that are easily replicable with a free ChatGPT account and a template prompt. Worryingly, many wrappers don't disclose where or how your data is being processed or stored. And unlike reputable SaaS platforms, these tools may lack GDPR or ISO compliance, especially if built by solo founders or hobbyists. Red flags to look out for: No clear documentation No security/compliance details Fancy UI, no visible backend Vague claims of "proprietary AI" Building a real product means solving an actual consumer need A real product demands brilliant UX and UI. More importantly, it must provide a superior experience compared to simply prompting ChatGPT, Grok, or another LLM - especially with configurable GPTs and AI Agents on the rise. If you already have a truly useful product, service, or process that real teams or customers rely on, AI can be a powerful tool to automate and enhance the experience. However, if your idea for a "product" is just an HTML gateway to AI, pause and take the following steps before moving forward: Take your idea and prompt your preferred LLM to see how quickly and easily you can generate the desired output. Map out your proposed user flow, even on paper, and count the steps you're adding. Ask yourself: Have I actually made my imagined customer's life easier? If the answer to step 3 is no, you have a fundamental amount of spin or PR will fix it. All that glitters is not gold In an era where AI is both accessible and powerful, it's easy to be dazzled by shiny interfaces. But as a business owner, marketer, or tech buyer, you need to scratch the sparkly surface to determine if what you're buying is the 'real deal.' It's also important to note that not all wrappers are bad. Some startups begin as wrappers but evolve into full-stack platforms by building layered proprietary logic, training domain-specific models and offering native integrations with CRMs, ERPs, and databases. The key is transparency and real value - the tool should be able to save you time, secure your data, and offer consistent utility.
Techday NZ
5 days ago
- Techday NZ
Kurrent unveils open-source MCP Server for AI-driven databases
Kurrent has released its open-source MCP Server for KurrentDB, enabling developers to interact with data in the KurrentDB database using natural language and AI agents rather than traditional coding methods. The Kurrent MCP Server offers new functionalities, allowing developers not only to query data but also to create, test, and debug projections directly through conversational commands. This feature is not available in other MCP server implementations, establishing a novel approach to database interaction by integrating AI-driven workflows into the database layer. Central to this release is the introduction of a self-correcting engine, which assists in automatically identifying and fixing logic errors during the prototyping phase. This reduces the need for manual debugging loops, streamlining the development process significantly for users building or modifying projections. The software is fully open-source and released under the MIT license, with documentation and a development roadmap available on GitHub. This permits both enterprise users and open-source contributors to adopt, customise, and improve the KurrentDB MCP Server without licensing restrictions. Kurrent MCP Server supports natural language prompts for tasks such as reading streams, listing streams within the database, building and updating projections, writing events to streams, and retrieving projection status for debugging. These capabilities aim to make the visual and analytical exploration of data more accessible and conversational for users with varying levels of technical expertise. The MCP Server is compatible with a broad range of frontier AI models, such as Claude, GPT-4, and Gemini. It can be integrated with popular IDEs and agent frameworks, including Cursor and Windsurf. This compatibility enables developers to leverage their preferred tools while reducing friction points typically associated with traditional database interactions. Addressing the new approach, Kirk Dunn, CEO of Kurrent, said, "Our new MCP Server makes it possible to use the main features of the KurrentDB database, like reading and writing events to streams and using projections, in a way that's as simple as having a conversation. The system's ability to test and fix itself reduces the need for debugging and increases reliability. Copilots and AI assistants become productive database partners rather than just code generators, seamlessly interfacing with KurrentDB." The server's key functions are designed to reduce development times for database tasks, enabling a focus on higher-value project work. Eight core capabilities are available, including Read_stream, List_streams, Build_projection, Create_projection, Update_projection, Test_projection, Write_events_to_stream, and Get_projections_status. Each of these responds directly to natural language instructions provided by the developer or AI agent. Kurrent has highlighted opportunities for the open source community to participate in the MCP Server's ongoing development. Developers can contribute code, report or tackle issues, and suggest new features through the project's GitHub repository and discussion forums. Comprehensive educational resources and installation guides are intended to help developers quickly integrate the MCP Server with KurrentDB for various use cases. Lokhesh Ujhoodha, Lead Architect at Kurrent, commented, "Before, database interactions required developers to master complex query languages, understand intricate data structures, and spend significant time debugging projections and data flows. Now, everything agentic can interface with KurrentDB through this MCP Server. We're not just connecting to today's AI tools, but we're positioning for a future where AI agents autonomously manage data workflows, make analytical decisions and create business insights with minimal human intervention." Kurrent emphasises that its MCP Server aims to remove barriers historically associated with database development by supporting conversational, agent-driven workflows. This aligns with broader trends towards AI-native infrastructure in enterprise environments, where human and algorithmic agents increasingly collaborate to deliver data-driven business outcomes.
