Agentic Futures Radar

A forward-looking map of where agentic engineering and AI are heading. Used in leadership briefings (module 21), engineer sessions, and as a recurring content series. Three horizons: where we are now, 3 months out, 6 months out.

Each signal: what it is, why it matters, who should care. Review monthly; signals move between horizons or fall off.

Last reviewed: June 2026.

Horizon 1. Where we are now

Spec-driven development

Write the specification first, let agents implement against it. Tooling like Microsoft's Spec Kit formalizes the pattern: spec as the contract, code as the output. Shifts engineering effort from writing code to writing intent precisely. For: engineers, engineering leaders. Source: https://developer.microsoft.com/blog/spec-driven-development-spec-kit

Claude wikis and skill files

The file taxonomy around agents is settling: CLAUDE.md for operating rules, AGENTS.md for team definitions, memory and context files, SKILL.md for workflows. Teams that treat these files as maintained assets outperform teams that treat them as setup chores. This is our module 03/06/19 territory becoming industry standard. For: everyone using agents seriously. Source: https://amitray.com/claude-md-vs-agents-md-memory-md-skills-md-context-md-guide-2026/

Context engineering

The successor discipline to prompt engineering: deciding what information reaches the model, when, and in what form. Selection, compression, isolation. The skill that separates working agents from flaky ones. For: engineers, power users. Source: https://www.langchain.com/blog/context-engineering-for-agents

MCP everywhere

Model Context Protocol crossed into mainstream adoption (tens of millions of downloads). It's becoming the USB of agent-to-tool connection. Every internal tool without an MCP server is invisible to agents. For: engineering leaders, platform teams. Source: https://www.digitalapplied.com/blog/mcp-97-million-downloads-model-context-protocol-mainstream

Computer use and browser agents

Agents operating real desktops and browsers as a new automation layer, replacing brittle RPA. Our module 10 territory. The constraint is reliability and permissioning, not capability. For: operations leaders, automation teams. Source: https://fordelstudios.com/research/ai-browser-agents-new-automation-layer-2026

Multi-agent orchestration

Coordination is the new scale frontier: teams of specialized agents with task graphs, mailboxes, and isolated workspaces. Our module 20 covers the mechanics. For: engineers, architects. Source: https://www.codebridge.tech/articles/mastering-multi-agent-orchestration-coordination-is-the-new-scale-frontier

Horizon 2. Three months out

Harness engineering

The emerging job title for what most "AI engineers" actually do: building the environment around the model (tools, knowledge, permissions, observation) rather than the intelligence itself. Expect role definitions and hiring to follow the term. For: engineering leaders, engineers planning careers. Source: https://www.faros.ai/blog/harness-engineering

Interactive agent evals

Static benchmarks are saturating. The frontier is interactive evaluation: agents judged on multi-step tasks in live environments (ARC-AGI-3 style). For buyers this means demo performance stops being proof; task-level evals become procurement criteria. For: leaders buying AI, engineers shipping it. Source: https://buildmind.ai/blog/arc-agi-3-frontier-agent-evals-march-2026/

Agent identity and security

Adoption is outpacing control. Agents need identities, scoped credentials, audit trails, and revocation, same as human employees. The first serious agent-caused incidents will make this a board topic. For: CISOs, leadership, platform teams. Source: https://www.gravitee.io/blog/state-of-ai-agent-security-2026-report-when-adoption-outpaces-control

Small language models

Gemma, Phi, Qwen class models running cheap, local, and private. The pattern: big models for reasoning, small models for high-volume narrow tasks. Changes the cost equation for production agent fleets. For: CTOs, engineers. Source: https://www.digitalapplied.com/blog/small-language-models-business-guide-gemma-phi-qwen

Horizon 3. Six months out

Agentic mesh

Enterprise architecture where many agents discover, trust, and transact with each other across the org. The successor question to "which agent do we buy": how do our agents interoperate. For: enterprise architects, CIOs. Source: https://medium.com/@chathuskadilhan/architecting-the-agentic-mesh-for-the-autonomous-enterprise-32af0593dd86

Memory as a platform

Agent memory graduating from session hack to infrastructure layer: shared, queryable, governed. Whoever owns the memory layer owns the switching costs. For: CTOs, platform teams. Source: https://blog.bymar.co/posts/agent-memory-systems-2026/

Continual learning

Agents that improve from their own deployment experience instead of waiting for the next base model. Trajectory data becomes a strategic asset; today's logs are tomorrow's training signal. For: leadership, data teams. Source: https://www.langchain.com/blog/continual-learning-for-ai-agents

World models

Models that learn how environments behave, not just how text continues. If they mature, agents get planning and physical-world competence that LLMs lack. Watch, don't bet yet. For: leaders tracking the frontier. Source: https://www.technologyreview.com/2026/04/21/1135650/world-models-ai-artificial-intelligence/

Governance as architecture

Compliance moving from policy documents into the runtime: policy-as-code gating what agents can do, with the EU AI Act as forcing function. Governance becomes an engineering deliverable, which is good news for whoever builds it early. For: leadership, legal, platform teams. Source: https://www.digitalapplied.com/blog/ai-agent-governance-policy-compliance-2026

How we use this

1. Module 21 delivers the radar as a briefing for leaders and engineers.
2. Quarterly 1:1 briefings for executive clients use the current radar as the agenda.
3. Content series: one signal, one post.
4. Catalog steering: when a signal moves to "now," it becomes a module candidate. Harness engineering already made that jump (module 20). Agent security and governance are next in line (pipeline modules 22-23 in the catalog).

Maintenance

Monthly review alongside the keep-current loop. For each signal: still accurate, moved horizons, or dead. Add at most two new signals per review so the radar stays a filter, not a feed.