Fergus C.

There’s such a clear opportunity for IT, engineering, or operations people to help companies go AI-first right now if they both understand the workflows of the organization and what’s possible with AI Agents. AI is going to continue to advance at a rate that far exceeds most organizations’ ability to adopt all of the latest technology when it comes out. This will only continue to increase as breakthrouhs keep coming from AI labs at a faster pace. Like it or not, the long pole in the tent on transforming workflows with AI will be the pace that change management can happen. Asking questions and getting answers back in an AI chat doesn’t require much change management. But adjusting an entire workflow to deploy AI Agents in the background to help do real extensive work fundamentally requires quite a bit more education, hand holding, and technical integration. And doing change management amidst a rapidly evolving space is even more complicated - one day a user will try to automate a workflow with AI and it won’t work, but 6 months later all of a sudden it’s possible. The people in the trenches that can help align what’s possible with AI and the individual workflows in an organization are going to have a huge advantage right now and going forward. The companies that do this early will learn faster and develop much more muscle in the process.

297

41 Comments

AI models are trained on public or generally available data sources. By default, they know everything about anything, *other* than your specific workflows and business. For AI Agents to be effective in the enterprise, they need your enterprise context. That context is sitting in all of the contracts, financial documents, research, marketing assets, meeting notes, conversations, and every other piece of information in the enterprise. By volume, most of this data is unstructured data. The reason we're all in on AI at Box is because the only way AI Agents will be successful is if they have access to the right information, in the right structure, at the right time, in a secure and well-governed way. This is exactly the platform we've been building at Box since our founding, and AI will expand the use and value of this information by orders of magnitude over time. We also can't do this alone. This is why we're hyper focused on building a broad partner ecosystem to connect into any other system where a customer needs access to this information via API directly, MCP, A2A, or otherwise. Incredibly exciting times.

239

20 Comments

One of the biggest breakthroughs building our agents was an insight hiding in plain sight... Agents perform better when given more agency. We're featured in Anthropic's customer stories on how Emergent applied this with Claude, giving our agents full autonomy to build and debug. The result? Production-grade full-stack apps shipped without human intervention. Read the full story: https://lnkd.in/gqPHRpSv #anthropic #claude #vibecoding

167

6 Comments

What does a modern database look like? If we were going to start over with the design of databases, instead of pulling forward the design decisions of the last 40 years, what would we choose to do? It's a huge question, but an important one. In a new blog post, I start to break down some of the changes that have come with modern hardware, including SSDs and the cloud, and how those changes affect database system architecture. Check it out here: https://lnkd.in/gYWX3dzR

503

43 Comments

Adding Document Understanding to Claude Code 📑🤖 Coding agents are awesome but they don’t natively understand docs (they kind of do, but it sucks). This limits their ability to build business applications - think about anything from a financial copilot to contract reviewer to rfp synthesizer to synthesizing PRDs. I wrote a blog to capture core patterns for adding document understanding to Claude Code - applicable to coding agents in general. This both helps Claude Code 1) get enterprise context when building apps, and 2) allows it to build more agentic, robust apps in general. 1️⃣ Add an MCP server that preindexes your file knowledge base 2️⃣ Let Claude Code loose on your filesystem with the right CLI tools for understanding files. 3️⃣ Give Claude Code context on core modules for document understanding and agentic workflows - this last step is important if you want to build “better apps in general” and not just “rule-based apps”. Blog: https://lnkd.in/g4EEFftk The blog offers general principles regardless of whether you’re using our stuff, because we fundamentally think that coding agents + docs leads to a ton of useful applications. Of course we are deeply investing in tooling in this space too (some referenced in the blog), a lot of it dependent on LlamaCloud. Check it out: https://lnkd.in/g9Wpqn7w

213

2 Comments

Gordon Food Service was founded 128 years ago, but it's an early champion of the AI future. Using AI, they've automated processes like product cataloging and invoice reconciliation, dramatically improving speed for both customers and vendors. Here is a conversation with Deji Adedayo, who heads Intelligent Automation at Gordon Food Service about how they made it possible. Link to full conversation in the comments.

102

8 Comments

Just a 5% drop in order cancellations can mean millions back in your pocket. Forrester's latest dual-OMS study found that 63% of the gains from adding a secondary OMS came down to resolving cancelled orders caused by gaps in legacy systems. The truth is that legacy OMS platforms can’t deliver accurate inventory quickly enough. So it shows items as in stock when they’re not. The brands that added a modern OMS on top of their legacy system discovered they were quietly losing money in three big ways: • Selling products that weren’t actually available. • Hiding stock that did exist, just not where the OMS was looking. • Failing to reallocate when one location sold out but another still had inventory. The study's participants found they could solve this by integrating a modern, modular OMS alongside their legacy system. Many of those same companies initially planned to run dual systems indefinitely, but they soon realized just how much better the modern OMS was. They ended up wanting to move everything over to the new platform. With all of that in mind, here’s the question you should be asking yourself: How much revenue might your business be silently losing to preventable cancellations?

25

A.I. Clouds are not all the same. Justin Gage of Amplify Partners shares, in explicit detail, how Modal is just built different, and why it's a mistake to think of them as an "inference provider". https://lnkd.in/gmaYQZtT

55

One problem in AI, particularly in agents, is that products that sorta kinda mostly work never really get mass adoption, even though they get a lot of test adoption. Edge cases matter. The last mile matters.

38

10 Comments

Context + Data Compression How's that for click-bait? Humans are really not very good at this information stuff. We get distracted every 8 seconds, and have memories slightly better than goldfish. Our brains compress everything immediately that are not deemed essential to existence. Especially when things are out of context - they just get wiped. Imagine if you could focus and recall everything you've ever heard on a particular subject - piece together all of those good ideas from the blog, the video, the conversation, the thought in the shower etc. Well, thats where we're going. Pretty soon, our AI will be with us all day. Listening to what we hear, seeing what we see.... Just like the Black Mirror episode. Our memories will be 100% backed up. Does that mean we'll become super-intelligent, just like the AI? I think it probably does. Video - just a new AI tool I'm playing with.

29

9 Comments

Too many companies focus on the capability layer of AI agents (voice, email or document ingestion) without addressing what actually drives value. The real value is in the intelligence layer, the ability to pull in the right context (SOPs, tribal knowledge, API docs) and get smarter every day. We view our role at Pallet as building the most intelligent AI agent for the logistics industry. Nailing the intelligence layer makes every capability from voice to email dramatically more powerful.

61

8 Comments