How AI Agents Work: Inside Agentic Inference
Agentic workflows fundamentally change AI compute workloads, requiring new benchmarking methods that measure throughput per megawatt, interactivity, and end-to-end latency.
Read articleTopic
AI Agents
All digests tagged AI Agents
The Cost of Accessibility in AI — Jemima Abu, CAIS
Accessibility must be a core requirement from the start. AI agents are powerful, but they require mandatory skills and guidance to prevent inheriting biases and generating verbose, non-semantic code.
Read article
AI Engineer NYC Livestream | Bridgewater, Anthropic, Two Sigma, Arize, Coatue, Nous, Gen. Intuition
The future of AI engineering involves building robust, governed agent frameworks. Key architectural components include a centralized data vault, a five-layer agent lifecycle, and continuous evaluation loops to manage non-deterministic behavior.
Read article
Automating document work with Long-Horizon AI Agents at LlamaIndex
The future of enterprise AI lies in long-running agents that automate document workflows. We explore the advanced document context layer, covering specialized OCR, agent harnesses, and optimizing models for cost and accuracy.
Read article
How to kill your SaaS: what we learned replacing a $43K/yr tool with AI agents
Can AI agents kill expensive SaaS? We tested it by cloning a $43K/yr event management tool. Discover what mixed models, guardrails, and human domain expertise are required to build (and own) the next generation of business software.
Read article
Demand-Driven Context: What 300 Incidents Reveal About Tribal Knowledge - Raj Navakoti
DDC is a methodology that uses AI agents to fail on real production incidents, systematically identifying and curating the minimum necessary knowledge context. This approach treats incident data as knowledge tests, moving beyond static runbooks to a 'living context.'
Read article
I Believe Everything My Agents Tell Me: Three Levels of Review — Corey Gallon, Rexmore
Corey Gallon details a three-level agent review method to combat AI failures. Learn how to implement artifact fingerprinting, independent source checks, and adversarial review for reliable agent outputs.
Read article
What Is PewDiePie Building? | For Beginners
Learn how to build local AI agents using Odysseus and llama.cpp. We break down agent harnesses, tool calling, and fine-tuning Qwen 3.5 9B on agent traces for maximum privacy and control.
Read article
AAIF Reading Group: Measuring Agent Policy Compliance
The reading group explores agent policy compliance, noting that LLMs struggle to follow complex SOPs. Solutions involve implementing deterministic and model-based guards outside the LLM to ensure reliable policy adherence.
Read article
Bridging the Gap Between AI Builders and Domain Experts
The future of AI agents lies in specialization. We must move from 'god models' to thousands of hyper-scoped, domain-specific agents, requiring standardized runtimes and a microservices mindset.
Read article
Your Agent Is a Lethal Trifecta (And What to Do About It) — Kim Maida, Keycard
Securing AI agents requires moving beyond standing credentials. By implementing OAuth token exchange and using a Security Token Service (STS), organizations can enforce task-scoped, ephemeral access, mitigating the risk of the 'lethal trifecta.'
Read article
We Let Our Financial Planners Write the Agents — Matthew Ragone, Range
Financial planners can build AI agents without writing code by capturing institutional judgment. The process involves five steps: Pick, Teach, Pack, Test, and Ship.
Read article
5 Concepts to Know for Coding with AI Agents
AI is changing software development from manual coding to architectural orchestration. Learn the five concepts—Spec-Driven Development, Context Engineering, and Loop Engineering—needed to master AI agents and build robust systems.
Read article
Build a Research Agent with Claude Managed Agents (Part 2 of 4) — Anthropic
Learn to build a 'research desk' app using Claude Managed Agents. We cover multi-agent coordination, Outcomes for self-supervision, and secure credential injection via Vaults.
Read article
The Next Trillion Internet Users Still Don’t Have an Identity — Adi Singh, AgentMail
Adi Singh argues that agent identity will be accumulated on existing human surfaces like email, not issued by a new spec. He details how existing protocols and surfaces are critical for agent adoption, using the Telnyx case study to illustrate the power of API-driven integration.
Read article
Sandboxes Aren't Optional: 4 Ways to Isolate Coding Agents — Robert Brennan, OpenHands
Robert Brennan details four strategies for sandboxing AI coding agents: Docker, GitHub Actions, VMs, and Kubernetes. He argues that Kubernetes offers the best isolation and scale for autonomous agents.
Read article
Hiring Employable Agents + Startup Battlefield Finals — Howie Liu, HyperAgent
The future of AI products lies in 'always-on' agents that orchestrate themselves. We explore the shift from completions to multi-agent systems, demonstrating how agents can run complex, end-to-end businesses.
Read article
Hill-Climbing Skills: Improve Agents Without Changing the Model — Shubhankar Srivastava, Browserbase
Discover how to improve browser agents without changing model weights using the AutoBrowse loop. This system pairs a Doer agent with a Teacher agent to generate reusable 'skill' files, dramatically increasing reliability and reducing operational costs on complex websites.
Read article
Burn Your Flags: Interactive CLIs for Agents — Mark Lummus & Navinkumar Patil, PayPal
The presentation details how to adapt interactive CLIs for AI agents, moving beyond simple flags. Techniques include using tmux sessions, IPC side channels for structured JSON, and context injection to manage complex, contextual workflows.
Read article
Build with Perception Agents — Emile Baizel & Shruti Arora, Amazon AGI Lab
Perception Agents enable AI to understand visual screens, moving beyond APIs. The workflow uses Visual Annotation (deterministic JSON output) and Visual Verification (Gherkin files) to close the design feedback loop, accelerating development.
Read article