Brian Fleming的动态

No-code AI platforms are revolutionizing the way we develop applications, making it possible to build sophisticated tools without writing a line of code. But how customizable are the data models and workflows? In our latest article, we explore just how much flexibility no-code AI app builders offer when it comes to creating custom data models and workflows. From defining data schemas and relationships to building dynamic workflows with conditional logic, discover how no-code AI tools empower you to develop applications that meet your unique business needs. Read the full article here: [Link to the article] Have you tried using no-code AI platforms to build your own custom app? Share your thoughts and experiences below! https://zurl.co/b3rJ

"The existing process before Olio Apps got involved took days. Now, it only takes less than 5 minutes" Did you know Olio Apps has AI expertise?? We needed to help our client query and reconcile customer data in a centralized data lake. They appreciated our no-nonsense approach to scoping and execution and we ensured communication was clear throughout the entire process. Ultimately, we were able to provide a solution that could find matching customer records across 5,000,000 entries in less than 5mins, while the previous solution took days. Read the full review here:

PlanAI v0.2 is here! I've just released a major update with new features and important fixes. This release brings: - Enhanced Logging & Monitoring: Gain better insight into OpenAI prompt usage, e.g. such as cached tokens - Expanded Model Support: Now includes o1-mini and o1-preview models which don't support JSON mode or structured outputs yet. - Interactive User Input: Tasks can now request input from users when needed. It has become increasingly challenging to fetch content from the web. - New Social Media Example App: Automatically suggests post topics based on profile interest queries. - Serper Search Integration: Simplifies frequent search scenarios. Explore the full update at GitHub: https://lnkd.in/g7PxtEVj Let me know what you think. #PlanAI #AI #OpenSource #TaskAutomation #Innovation

?? Unlock the power of enterprise AI with RelationalAI's Snowflake Native App! Bringing AI into enterprises is easier said than done. What if you could access advanced analytic tools right where your data lives? Leveraging Snowpark Container Services, RelationalAI's Snowflake Native App runs as a relational knowledge graph coprocessor, surfacing common knowledge about the business for intelligent decision-making. Snowpark Container Services ensures easy deployment and maintenance, a secured, trusted data governance foundation, and access to Snowflake’s Marketplace. The speed to value, combined with the elasticity and power of RelationalAI’s knowledge graphs, makes it easier for companies like AT&T and Cash App to run deeper analytics faster at scale. ?? Explore the full potential of your data with RelationalAI, accessible through Snowflake Marketplace!

Scaling with Vercel + Supabase ???? Exciting times with the new Vercel and Supabase integration! I've been able to spin up full Postgres instances directly from Vercel's dashboard in under a minute, making it easier than ever to get AI-driven projects off the ground. This streamlined setup even includes billing integration for simplicity. If you're exploring rapid prototyping for your AI apps, this could be a game-changer! ?? Let's talk about how I'm using this to build and iterate faster. More on Vercel + Supabase integration ?? https://lnkd.in/gGNzsZUJ RaleighAI Solutions ?? #Vercel #Supabase #AI #RapidPrototyping #TechIntegration #Serverless #Innovation

Introducing Databricks Apps – a new way to build internal data + AI applications! Build custom data visualizations, AI applications, and self-service analytics directly on Databricks!

???????????????? ???????? ???? Over the past couple of weeks, I've been working on developing multiple small app side projects during weekends, using AI, just for fun (dataf.in, diabeticks.com, voicetribe.in and a few more...) ???? ??????????: - Lovable and StackBlitz Bolt.new (for Frontend and Integrations) - Supabase (for DB) - Supabase (for Auth and Storage, Edge Functions and what not!) - Resend(for Emails) - V0.dev (for UI inspiration or design) - Netlify.com to deploy ???????? ?????? ???? ???????? ?????????? ??????????????????: ? It is very easy to build with LLMs but you need to tie lot of loose ends together. So, knowledge of frontend/backend and SQL helps a lot to untangle the issues ? Knowing SQL is the biggest plus when building the data structure, schemas, auth, storage etc. Simple(!) Query Language is simply very helpful! ? Planning and designing would take 80% of the time as most of the coding will depend on the prompt and how well you structure it to execute. So, having an rough idea of what to build and in what sequence is very crucial. ? AI is intelligent as well as dumb as of now. It will need your precise guidance every step of the way from you else garbage in = garbage out. ? Data Structure, Data Structure and Data Structure - any feature can be built if you know your data, schemas and RLS policies! ? with Supabase you can do a lot: No separate auth, db, Storage, Edge Functions etc. needed. All in one. Saves lot of integration hassles and time and passwords needed to remember! ???? ????????????????: 1. Think of an idea and feature set needed for the app 2. Define requirements and data structures first 3. Build a MVP of most basic features inc. auth 4. Define a nice UI layout and color scheme that would work for the feature set 5. Refine. Refine. Refine. Till the time you are satisfied with the output 6. Spruce it up with UI/UX improvements and a landing page AI accelerates a lot of development but requires solid fundamentals - technical knowledge (not in detail but to make sense if any errors), data modelling, and feature planning. AI tools will take you so far. You need to be clear ???????? you want and ?????? you want it!

CRUD support is coming to RAW! Create, update, and delete data in real-time from your APIs, enabling more dynamic interactions for AI agents, apps and LLMs! Learn more at https://zurl.co/TPdU

And here is what you should do to make your software stack ready for future AI: ** Reduce # of languages and technologies ** Even super-intelligent AIs will struggle to work across 4 languages, 3 database types, 2 message queuing APIs, 2 cloud providers, 4 observability stacks, etc. ** Simple architecture ** Keep the number of network hops your data makes, to get something done, to minimum needed.?Every network hop requires understanding of network design, middleware, DNS mappings, routing rules, header handling etc.?You don't want to fill up your AI's context window with this, since it will leave little room left for the actual job. ** Go mono-repo ** Good luck trying to build a RAG that can bring relevant code spread across 135 GitHub repositories into your context. No. Infinite context won’t solve it. Because context size will impact performance and cost. You want to be smarter about what you feed your AI during inference. ** Have standardised development pipelines ** Future AI will not be able to understand your 17 different deployment methods and undocumented terminal commands. ** Reduce the layers of abstraction and modularity in code ** Smaller functions, layers of abstractions—all the software practices optimised for humans to read and write code—work against AI. AI will better understand and modify code where your request deserialisation, config lookup, logic, caching, DB access, and everything else is in one long function. ** Increase the number of components owned per team ** More teams mean more fragmentation of code, technologies, and additional layers of abstraction. Reduce teams, ensure multiple components are built in a consistent way. ** Move debates and decisions from email to internal confluence/wiki, google-docs ** AI transcription is good. Invest in good microphones and transcribe meetings. Debates and decisions captured in email are in a black hole; since companies are unlikely to ever give AI access to emails. ** Don’t throw away data ** Super-intelligent AI is hungry for all the possible data you can generate. It will find patterns in datasets that you did not know existed.But it can’t do so if you throw away metrics data to save storage costs. ** Stay two steps behind the latest releases ** The models can’t help you if you are on the latest releases of APIs and libraries. Time your upgrades with the latest generation model traning-data cut-off dates. ** Write automated tests ** If AI has to maintain a codebase, it needs to understand what is expected from the codebase. Automated tests are arguably the best way to specify expectations from code. Add tests, even if the tests fail, it’s ok — AI will eventually fix them. Having said that, don’t like AI? No problem. Just do the opposite of each of the above and you are good for a decade ;) #genAI #softwareengineering #gpt5 #technology

As AI continues to evolve, it's reshaping the landscape of composability in both technical and non-technical domains. This insightful chart from chiefmartec.com highlights how various tools and platforms are shifting, offering a clearer understanding of where different solutions stand in terms of technical complexity and their role in services/data. From code libraries and cloud API services to no-code automation tools and spreadsheets, the spectrum demonstrates the diverse range of options available for businesses. Notably, AI is playing a pivotal role in bridging gaps and enhancing functionalities across these platforms. Key Takeaways: ?? Technical Solutions: Code libraries and cloud data warehouses (e.g. Snowflake) remain at the technical end, crucial for developers and data scientists. ?? Service and Automation: Tools like enterprise automation (e.g. Workato) and low-code app builders (e.g. Power Apps) are becoming increasingly sophisticated, allowing more users to leverage powerful technologies without deep technical knowledge. ?? Data Management: Low-code ETL tools (e.g. Fivetran) and BI & analytics platforms (e.g. Looker) are enhancing data accessibility and insights. ?? No-Code Revolution: Platforms like Airtable, Zapier and Wix are empowering non-technical users to create and manage complex applications and websites effortlessly. As we navigate this evolving landscape, it's exciting to see how AI-driven composability is democratizing technology, making powerful tools accessible to a broader audience and driving innovation across industries. #AI #ArtificialIntelligence #Composability #TechInnovation #NoCode #LowCode #DataManagement #Automation #BusinessTechnology