AI Application Architecture
Patterns for building production AI applications, including orchestration, memory, tool use, state management, APIs, and system design.
Persistent Sandboxes Make Agents Remember, Plan, and Reuse Their Work
Nico Albanese, a Vercel engineer working on the AI SDK, argues that agents become more reliable when they are given a persistent sandboxed computer, not just a runtime and tools. In his workshop, he builds that pattern with AI SDK 6, Vercel’s named sandboxes, a bash tool, and a file-backed memory system, showing how an agent can plan in files, preserve context across sessions, and create reusable scripts without a separate memory layer.
SAP Says ERP Context Will Make AI Agents Reliable for Business
SAP chief executive Christian Klein used Bloomberg Technology to frame the company’s new autonomous enterprise platform as a bet that AI agents need business context more than proprietary models. He argued that SAP’s advantage is its access to ERP data and process knowledge, which can make agents reliable enough to coordinate work across finance, commerce, inventory, procurement and supply chains. Pressed on competition from partners such as AWS, Klein said SAP’s role is to provide the enterprise context layer while working with hyperscalers and data platforms to harmonize data beyond SAP systems.
Fixed Evaluation Suites Go Stale as Agents Optimize Toward Intent
Vincent Koc of Comet ML argues that AI evaluation is being outpaced by the systems it is meant to measure. In a talk on adaptive evaluation for agents, Koc says static benchmarks and handcrafted test sets are poorly suited to applications that change with prompts, tools, production traces, user behavior and even their own harnesses. His proposed direction is to define the intended end state, use traces and telemetry to surface drift and edge cases, and treat evals as a continuously revised system rather than a one-time benchmark.
Cerebras’s Higher IPO Range Tests AI Infrastructure Demand
Alex Wilhelm and Jason Calacanis treat Cerebras’s raised IPO range as a test of how much public investors will pay for future AI inference demand and the quality of contracts with customers such as OpenAI. Ori Goshen makes a parallel case that enterprise AI’s hard problem is no longer choosing one model, but routing work across models, tools and inference strategies for cost, latency and accuracy. Across OpenAI’s deployment spinout, AI21’s orchestration pitch, Magrathea Metals’ brine-based magnesium plan and OpenClaw’s fading momentum, the article frames deployment as a question of incentives, constraints and where the bottleneck actually sits.
Coding Agents Work Best When Products Expose Simple Tools
Matthias Luebken argues that coding agents such as OpenClaw are less mysterious than they appear: they are LLMs calling tools in a loop, made more useful by a runtime, shell, sessions and product hooks. In his Tavon talk, he uses Pi, a minimal coding-agent SDK, to show how that loop can be embedded inside business software, including a sales workflow where RFP emails are routed to customer-specific agent sessions and returned to users as draft replies. His architectural point is that teams should not force agents through opaque systems, but expose data, commands and controls in forms coding agents can use cleanly.
Slack-Native AI Coworkers Turn Memory and Permissions Into Product Risks
Fryderyk Wiatrowski argues that building Viktor as an AI coworker inside Slack is not a matter of scaling a personal assistant to more users. A company-level agent gains value from shared context, shared integrations, and the ability to act where work is discussed, but those same features create harder problems around memory isolation, permissions, fragmented Slack conversations, proactivity, and tone. His case is that an “AI employee” has to be designed less like a chatbot and more like a new hire entering the company’s communication layer.
Long Lake’s $6.3 Billion Amex GBT Deal Tests AI-Led Buyouts
Long Lake Management co-founder and CEO Alexander Taubman argues that AI can change the economics of services businesses when the buyer owns the workflow, not just the software layer. In a conversation with Elad Gil about Long Lake’s announced $6.3bn take-private of American Express Global Business Travel, Taubman presents the firm’s model as acquiring trusted services companies, embedding its Nexus AI platform into day-to-day operations, and using productivity gains to drive growth, customer service and employee retention rather than short-term cost cuts.
Durable Agents Need Context Logs and Execution Snapshots
Eric Allam of Trigger.dev argues that durable agents need more than the replay-based workflow model used for durable transactions. In his talk, he separates agent durability into two problems: the LLM context, which fits naturally as an append-only log, and the execution environment — files, memory, subprocesses and local state — which he says should be preserved through OS-level snapshot and restore. Allam uses Trigger.dev’s Firecracker work to make the case that long-running agents are becoming session-like workloads, not just replayable transactions.
Head-Tail Truncation and Memory Stabilized Arize’s Trace-Analyzing Agent
Sally-Ann DeLucia argues that agent performance depends on context management as an operating discipline, not on larger prompts or simple compression. Drawing on Arize’s work building Alyx, an agent that analyzes trace data from AI systems including its own, she says naive truncation broke follow-up reasoning and LLM summarization gave the model too much control over what mattered. Arize’s more durable pattern was to preserve the head and tail of context, store the middle for retrieval, test long sessions explicitly, and move heavy workloads into sub-agents.
Production AI Features Need Feedback Loops, Not One-Shot Prompts
Mehedi Hassan, a product engineer at Granola, argues that the hard part of shipping AI features is not getting a model to work once in a demo, but making its behavior reliable and inspectable in production. Using Granola’s meeting-notes app as the case, he says web search, chat, and prompt personalization quickly expose costs, context limits, provider instability, and role-specific user expectations that a single prompt cannot absorb. Granola’s response, in his account, was to build feedback loops: internal tracing, broadly usable debugging tools, and faster ways to test product variants before shipping.
Voice AI Still Confuses Natural Speech With Real Conversation
Neil Zeghidour, CEO of Gradium AI and one of the researchers behind the full-duplex voice model Moshi, argues that voice AI’s long-promised “Her” moment is still being confused with better synthetic speech. His case is that cascaded voice agents are useful but structurally too slow and lossy to feel conversational, while speech-to-speech models improve flow but remain limited unless they can listen and speak simultaneously, use tools reliably, understand paralinguistic cues, and run cheaply enough to scale.
ElevenLabs Voice Engine Wraps Existing Chat Agents Without Rebuilding Them
Luke Harries of ElevenLabs argues that the next step for chat agents is not a new orchestration stack but a voice layer around the agents companies have already built. His case for ElevenLabs’ Voice Engine is that teams can keep their existing LLM logic, RAG, tools and business rules, while offloading speech-to-text, text-to-speech, turn-taking and interruption handling to a wrapper. The product is positioned for companies that want voice interfaces across web, phone and meeting channels without rebuilding their chat agents inside a fully managed platform.
Personal AI Lets One Builder Do the Work of Teams
Y Combinator CEO Garry Tan argues that personal AI is reaching a stage comparable to the early personal computer: powerful enough to let one person build software that once required a team, but still brittle enough to demand technical ownership. Drawing on his work with Claude Code, OpenClaw and his GStack workflow, Tan makes the case for heavy token use, Markdown-encoded “skills” and multiple coding agents under one accountable human operator. The larger question, he says, is whether users will control their own AI tools, data and prompts, or work inside opaque systems controlled by others.
Agentic Search Needs Specialized Tools and General-Purpose Escape Hatches
Elastic’s Leonie Monigatti argues that context engineering for LLM agents is largely a search-interface problem: the critical question is how an agent decides what to retrieve from files, databases, memory, the web, and other sources before the model answers. In her workshop, she shows why semantic search, database query tools, shell access, and agent skills each solve different parts of that problem and fail in different ways. Her recommendation is to build retrieval stacks that combine easy specialized tools for common tasks with more general tools for ambiguous or complex ones, then use observed failures to refine the stack.
Agentic AI Is Making Enterprise Software a Control Layer
ServiceNow president, COO and chief product officer Amit Zavery argues that agentic AI will change enterprise software, but not by letting unconstrained agents replace the platforms that run corporate workflows. In a ServiceNow-sponsored interview, Zavery says the hard problem is turning probabilistic AI into reliable action across regulated, multi-system businesses, with the context, permissions, auditability and governance that enterprises require. His case is that companies such as ServiceNow retain leverage if they make AI production-ready, while software vendors that fail to adapt remain exposed.
Production Analytics Finds Agent Failures That Standard Evals Miss
Scott Clark, co-founder and chief executive of Distributional, argues that teams running LLM agents need to look beyond pre-production evals and dashboards of known metrics. His case is that the most consequential failures often emerge only in production, where agents interact with users, tools and changing models in ways teams did not know to test. Clark proposes an observability stack in which telemetry records what happened, monitoring tracks known signals, and analytics clusters trace behavior to surface unknown failure modes that can become new evals, guardrails, prompts or system fixes.
AI Coding Makes Software-Engineering Fundamentals More Important
Matt Pocock, a TypeScript teacher now focused on AI engineering, argues that AI coding has made software-engineering fundamentals more important rather than less. In a conversation with Shawn Wang, Pocock says code generation works best when humans define the architecture, module boundaries and domain language that give agents a coherent system to change. The lesson he draws from Claude Code and other fast-moving tools is that tool-specific knowledge ages quickly, while engineering judgment remains the durable layer.
Production Agents Need Evals and Managed Variables After Deployment
Samuel Colvin of Pydantic argues that production agents need more than observability after deployment: they need evals, traces, and typed configuration that can change prompts, models, and other parameters without a redeploy. Using Pydantic AI, Logfire, managed variables, and GEPA, he shows a workflow for moving from manual prompt tuning toward continuous optimization. His case is practical rather than automatic: GEPA can improve a narrow benchmark, but only if the team has representative data, sound evaluation criteria, and a clear definition of what better means.
Perplexity Frames AI Agents as Metered Digital Labor
Perplexity chief business officer Dmitry Shevelenko argues that AI agents should be judged less as software features than as metered digital labor: tools users will pay for when they perform economically useful work. In a Big Technology Podcast interview, he makes the case that Perplexity’s computer-use agents, workflow packaging, broad permissions and multi-model orchestration are all part of that shift. The unresolved question is whether users and companies will accept the access, trust and usage-based pricing required to make those agents a real business rather than another AI novelty cycle.
OpenAI Splits Audio API Into Translation, Transcription, and Voice-Agent Models
OpenAI is presenting three new API audio models as infrastructure for voice applications that can translate, transcribe, reason and act in real time. Romain Huet’s demonstration centered on GPT-Realtime-Translate, which keeps pace with multilingual speech, and GPT-Realtime-2, a voice-agent model that can follow turn-taking instructions, use business context and call tools while explaining its work. GPT-Realtime-Whisper completes the set as a streaming speech-to-text model for live transcription.
Coding Agents Need Library Source Code, Not Longer Prompts
Michael Arnaldi, of Effectful, argues that coding agents use Effect better when the project gives them the Effect source code, not just better prompts or documentation. In a workshop starting from an empty repository, he demonstrates cloning the Effect repo into the project, having the agent extract local pattern files, and then using strict TypeScript diagnostics, tests, lint rules and persistent instructions to steer the agent toward a working Effect HTTP API.
Production Agents Need Semantic Observability Beyond Offline Evals
Raindrop’s workshop argues that production agents need a different observability model from conventional software monitoring or offline evals. Zubin Kumar, Danny Gollapalli and Ben Hylak make the case that teams should track both explicit telemetry such as tool errors, latency and cost, and implicit signals such as user frustration, refusals, task failure, capability gaps and unusual workarounds. Their framework treats real production behavior as the primary surface for finding regressions, running experiments and catching failures that do not appear as clean exceptions.
Voice Will Be the Primary Interface for AI Agents and Robots
At Sequoia’s AI Ascent 2026, ElevenLabs co-founder and CEO Mati Staniszewski argues that audio was an overlooked frontier in 2022 because the AI field was focused on text and images, leaving room for a smaller company to build quickly and monetize early. His broader case is that as AI intelligence becomes more capable, voice becomes the interface problem: the way people will use agents, robots, services, education and healthcare. Staniszewski says the next hard problems are emotional intelligence, timing, authentication and workflow, not merely making synthetic speech sound human.
Luma Is Rebuilding Video AI Around a Unified Multimodal Transformer
In a Stanford CS153 guest lecture, Luma AI co-founder and chief executive Amit Jain argues that generative video is only a staging point toward “unified intelligence”: models that understand and generate across text, images, video, audio, code and tools in a single work loop. Jain traces Luma’s path from Apple-era LiDAR and 3D capture to internet-scale video, saying the company followed the data but now sees prettier clips as insufficient. The destination, he says, is a multimodal AI factory for professional creative and physical work, where human skills, tool use, feedback and unified transformer architectures produce full campaigns, schematics, productions and eventually robotics workflows.