AI engineering and architecture
Building systems around models rather than on top of them: control planes, deterministic rails, retrieval over volatile corpora, and code intelligence at scale.
Research sweeps
2026-04-24 · deep
Engineering AI Control Plane
Engineering AI control planes for software delivery from July 1, 2025 through April 24, 2026: how teams implement AI across development workflows and CI/CD, choose tools/models/SDKs, govern observability and compliance, manage reliability and provider availability, and handle cognitive debt, dark code, case studies, success stories, and failure modes across team size, company scale, and greenfield versus brownfield systems
Claude Opus 4.8- financial
- frontier
- academic
- +3
2026-06-07 · deep
AI on Deterministic Rails
AI on deterministic rails: how AI and traditional deterministic software are forming a symbiotic stack from January 2025 through June 2026: the enterprise "PoC-opalypse" and the shift from token consumption to durable agentic adoption patterns, AI leveraging software-encoded workflows as guardrails (variance and error control) rather than replacing them, the frontier moving from raw model capability to model orchestration and harness design (Claude Code, OpenCode, Pi), right-sizing with smaller and open-weight models (Llama, Qwen, DeepSeek, Mistral) for cheap routine automation and private inference, and the token-pricing economics behind enterprise sticker-shock over agentic spend versus delivered value
Claude Opus 4.8- financial
- frontier
- academic
- +3
2026-06-01 · standard
Handling Large Volatile Corpora with AI: Caching, Freshness, and Retrieval at Scale
Engineering patterns for large, fast-changing corpora from 2024 to 2026: prompt and prefix caching, the shift from prompt engineering to context engineering, embedding staleness and freshness strategies, multi-strategy retrieval beyond pure vector search, and the inference-cost economics now reshaping infrastructure decisions.
Claude Opus 4.8- frontier
- tech
- academic
- +1
2026-06-03 · deep
Code Intelligence & Code-Graph Indexing for AI Agents
Tools and emerging approaches for code intelligence and code-graph indexing for AI coding agents from June 2025 through early June 2026, spanning local/embedded indexers (CodeGraph/Caveman-style repo maps, tree-sitter, SQLite and embedded graph stores), enterprise-scale code understanding (SCIP, code knowledge graphs, embeddings+retrieval), LSP-to-MCP bridges such as Serena, and the semantic-vs-syntactic-vs-embedding trade-off.
GPT-5.5- tech
- frontier
- academic
- +2
Explainers
- Research Explainer · Hu et al. (2025)
EPIC reuses KV caches across any prefix, by recomputing only a handful of tokens per chunk
Position-Independent Caching lets language models reuse document KV vectors regardless of what comes before them. EPIC's LegoLink algorithm fixes the resulting attention sink with O(kN) work instead of O(N²).
- Research Explainer · Liu et al. (2026)
Claude Code is a thin agent loop wrapped in a thick safety harness
A source-level reading of Anthropic's coding agent finds that about 1.6% of the code is AI decision logic. The other 98.4% is permission gates, context compaction, extensibility plumbing, and recovery.
- Research Explainer · Arunkumar (2026)
AI systems are evolving from text generators to autonomous agents, but the architecture for making them reliable is still being invented
A comprehensive survey proposes a six-dimension taxonomy for LLM-based agents, mapping the shift from simple reasoning loops to hierarchical multi-agent systems with standardized tool connectivity, and catalogues the open failure modes that still block real-world deployment.
- Research Explainer · Alenezi (2026)
AI agents are no longer just answering prompts, they're becoming goal-directed systems with their own control loops
This paper maps the architectural shift from stateless LLM calls to autonomous agent systems with typed tools, hierarchical memory, multi-agent coordination, and governance baked in from the start.
- Research Explainer · Vandeputte (2025)
Stop letting AI agents run everything; make them automate themselves out of the critical path
A Nokia Bell Labs framework argues that reliable GenAI systems should blend traditional software engineering with cognitive AI processing, keeping agents as occasional problem-solvers rather than permanent gatekeepers.
- Research Explainer · Su et al. (2025)
Kubernetes dominates five years of practitioner talks; while planning and coding get almost no attention
An analysis of 5,677 talks from eight major industry conferences (2020–2024) reveals that a tiny cluster of technologies shapes modern software architecture, most tools serve late DevOps stages, and early design phases remain a blind spot.
- Research Explainer · Esposito et al. (2025)
GenAI can help architects sketch systems; but nobody is checking whether the sketches are right
A multivocal literature review of 46 studies finds GenAI is already embedded in early architectural tasks, yet 93% of the work skips formal validation of what the models produce.
- Research Explainer · Lin et al. (2026)
An agent rewrites its own coding harness, and beats the engineers who used to do it by hand
Agentic Harness Engineering turns harness tuning into an automated loop. Ten iterations lift pass@1 on Terminal-Bench 2 from a bash-only 69.7% to 77.0%, past the human-built Codex harness and every self-evolving baseline.
- Research Explainer · Demirer, Musolff & Yang (2026)
AI coding agents triple the code developers write, but shipped software barely budges
A study of more than 100,000 GitHub developers finds that each generation of AI coding tool delivers bigger task-level gains, yet those gains shrink dramatically as they travel down the production chain toward actual releases and end users.
- Research Explainer · Liu (2026)
AI coding assistants fix more code smells than they create, but introduce nearly twice the security issues they resolve
Across 304,362 AI-authored commits from 6,275 GitHub repositories, AI tools are a net positive for surface-level code quality but a net negative for bugs and security vulnerabilities, with 24.2% of all introduced issues persisting indefinitely.
- Research Explainer · Li (2025)
AI coding agents now ship 456,000 pull requests, but their code gets rejected far more often than human work
The first large-scale dataset of autonomous coding agent activity on GitHub reveals that speed and scale are real, but acceptance rates, review dynamics, and code complexity tell a more sobering story about the gap between benchmarks and production.
- Research Explainer · Humberd (2026)
Agency theory was built for human managers, but AI is becoming the agent nobody knows how to supervise
A new framework maps five stages of AI evolution against traditional agency mechanisms, arguing that firms need to scaffold monitoring and incentive systems now, well before AI gains full decision-making autonomy.