Skip to content
Discuss a project
  1. Kiz8
  2. Research
  3. Market research

Market research.

Market analysis of developer tools, including AI and data tooling. We track vendors, public product data, founders, funding, and how each segment formed. These reviews map the commercial landscape; they are separate from our scientific AI experiments.

Review 01 · SandboxReady · updated May 2026

Where agents run code.

Compute sandboxes for AI agents — microVMs, containers, and snapshot architectures that let agents run code, browse the web, and persist state without crashing each other or the host.

Open the reviewAll reviews
46Vendors tracked
35Milestones, May 2018 → Jun 2026
9Funding rounds
12Launches & GA releases
Fig. 01 · Milestones per year
YearMilestones
20182
20190
20201
20212
20221
20231
20242
2025 · peak14
202612

From a single milestone per year to one per week. 2026 counts through June.

In preparation
Queued · 02
Agent frameworksHow agents are built.

Frameworks for building autonomous agents — planning, tool use, multi-agent orchestration.

Queued · 03
Foundation modelsWhat agents think with.

Frontier LLMs and their host APIs — context windows, tool use, multimodal, pricing per token.

Queued · 04
LLM observabilityWhy agents fail.

Tracing, evaluation, prompt diffing, cost monitoring — the layer that turns agent runs into debuggable artifacts.

Queued · 05
Evaluation frameworksDid the agent do the job.

Tooling for grading model output — task-specific, golden-set, LLM-as-judge, regression suites.

Queued · 06
RAG & retrievalWhat agents remember.

Vector stores, embedding pipelines, and retrieval frameworks that ground agents in private data.

Queued · 07
Browser automationHow agents click.

Browser sandboxes, agentic web drivers, scraping toolkits — the surface for agents to act on the open web.

Saw something worth a deep read? Send us tips, technologies to profile, or a market dataset you wish someone would publish.

Send a tip ›