Coding for the Future Agentic World – O’Reilly

Could 8 AI Codecon was an enormous success. We had wonderful audio system and content material. We additionally had over 9,000 reside attendees and one other 12,000 who signed up to have the ability to view the content material in a while the O’Reilly studying platform. (Right here’s a put up with video excerpts and a few of my takeaways.)

So we’re doing it once more. The following AI Codecon is scheduled for September 9. Our focus this time goes to be on agentic coding. Now I do know that Simon Willison and others have derided “agentic” as a advertising and marketing buzzword, and that nobody may even agree on what it means (Simon has collected dozens of competing definitions), however regardless of the time period involves imply to most individuals, the fact is one thing all of us have to come back to grips with. We now have LLMs with specialised system prompts, utilizing instruments, chained collectively in pipelines or operating in parallel, operating in loops, and modifying their environments. That looks like a fairly good start line for a working definition.

Within the September 9 AI Codecon, we’ll be concentrating on 4 crucial frontiers of agentic improvement:

Agentic interfaces: Transferring past chat UX to stylish agent interactions. New paradigms don’t simply require new infrastructure; additionally they allow new interfaces. We’re trying to spotlight improvements in AI interfaces, particularly as agentic AI functions prolong far past easy chat.Software-to-tool workflows: How brokers chain throughout environments to finish advanced duties. As an previous Unix/Linux head, I like the thought of pipelines (and now networks) of small cooperating applications. We are actually reinventing that type of network-enabled method to functions for AI.Background coding brokers: Asynchronous, autonomous code technology in manufacturing. When AI duties begin operating within the background, anticipate both magic or mayhem. We’d favor the previous, and wish to present the reducing fringe of how one can construct safer, extra dependable brokers.MCP and agent protocols: The infrastructure enabling the agentic net. Whereas the primary technology of AI functions have been centralized monoliths, we’re satisfied that the agentic future is certainly one of cooperating AIs, interoperating not solely with functions designed for people but additionally with AI-native endpoints designed to be consumed by AI brokers. MCP is a good begin, however it’s removed from the top of protocols for agent-to-agent communication. (See my put up “Disclosures. I Do Not Suppose That Phrase Means What You Suppose It Means.” for an account of how communication protocols allow participatory markets. I’m tremendous enthusiastic about the way in which that AI is creating new alternatives for builders and entrepreneurs that aren’t capital-intensive, winner-takes-all races just like the preliminary race for chatbot supremacy. These alternatives come from the community protocols that allow cooperating AIs.)

The first convention observe can be organized very like the Could occasion: a curated assortment of hearth chats with senior technical executives, sensible engineers, and entrepreneurs; sensible talks on the brand new instruments, workflows, and hacks which can be shaping the rising self-discipline of agentic AI; demos of how skilled builders are utilizing the brand new instruments to supercharge their productiveness, their progressive functions, and consumer interfaces; and lightning talks that are available from our name for proposals (see under). We’ll even have a set of in-depth tutorials on separate days as a way to go deeper if you need. You’ll be able to enroll right here. The mainstage occasion is free. Tutorials can be found to O’Reilly subscribers and may also be bought à la carte—trial memberships may even get you within the door. 🙂 The separate demo showcase can be sponsored (and thus free).

Name for proposals

Do you’ve got a narrative to share about how you might be utilizing brokers to construct progressive and efficient AI-powered experiences? We wish to hear it. The AI Codecon program can be a mixture of invited talks and your proposals, so we’re asking you to submit your concept for a fast five-minute lightning speak about cutting-edge work. We aren’t searching for high-level discussions; we wish struggle tales, demos of merchandise and workflows, progressive functions, and accounts of recent instruments which have modified how you’re employed. You (collectively) are inventing the long run at a livid price. We’d love to listen to about work that may make folks say “wow!” and rush to study extra. Your objective ought to be to make the rising future occur sooner by sharing what you’ve realized.

After studying your proposal, we might ask you to current it as proposed, modify it, develop it into an extended discuss, be part of a dialogue panel, or seem in our related demo day program.

Subjects we’re considering embody:

UI/UX—How are you or your organization exploring agentic interfaces and shifting past chatbot UX?The way you’re utilizing brokers as we speak—Are you handing off whole duties to brokers? What duties are you handing off, and the way are you integrating the brokers’ work into the SDLC?Your tool-to-tool workflows—How are you chaining brokers throughout environments and companies to finish duties end-to-end?Background coding brokers—What’s rising from extra asynchronous and autonomous code technology, and the place is that this headed?MCP and the way forward for the online—How are agentic protocols unlocking progressive workflows and experiences?Shock us. With the market shifting this quick, chances are you’ll be doing one thing wonderful that doesn’t match this system as we’re envisioning it as we speak however that the AI coding group wants to listen to about. Don’t be afraid to paint outdoors the strains.

We’re additionally nonetheless considering listening to about subjects we explored in our earlier name for proposals:

What has modified about the way you code, what you’re employed on, and the instruments you utilize?Are we working towards a brand new improvement stack? How are your architectures and patterns altering as you progress towards AI-native functions?How is AI altering the make-up and workload of your dev groups? What have you ever achieved to take care of high quality requirements with AI-generated code?What forms of duties are you taking up that had been beforehand too time-consuming to perform?What issues have you ever encountered that you simply want others had instructed you about if you had been beginning out in your journey?What sorts of enjoyable initiatives are you taking up in your free time?

So submit your proposal for a chat by July 18. And when you have a product you’d prefer to demo as a part of our sponsored demo showcase, please tell us at AI-Engineering@oreilly.com.

Thanks!