Kontext raises $4 million to police what AI agents do inside businesses · Digg

Kontext raises $4 million to police what AI agents do inside businesses

The Munich startup checks an agent's identity, requested resource and assigned task before allowing an action to run.

Kontext co-founders Jens Ernstberger and Michel Osswald seated in an office.

TLDR

Munich security startup Kontext has raised $4 million for software that enforces action-level policies on AI agents operating inside businesses. Its system sits between an agent and company tools, evaluating the agent's identity, the resource it wants and the task it was assigned before allowing or denying an action. Teams can begin in observe mode, then switch on enforcement with an audit log of decisions. 42CAP led the round, with a16z CSX and High-Tech Gründerfonds participating. Kontext says its product already supports coding agents including Claude Code and Codex.

Featured Source

German AI-security startup Kontext has raised $4 million to expand software designed to stop workplace agents from taking actions that their credentials technically allow but their assigned task does not justify.

The Munich company places a policy layer between an AI agent and the systems it can reach, according to SiliconANGLE's report. Each requested action is evaluated using the agent's identity, the resource involved and the job the agent was actually given.

Kontext's example is a coding agent assigned to fix a bug. Reading the relevant repository fits the task. Sending that code to an outside service or changing unrelated infrastructure does not, even if the same credentials permit those actions.

From observation to enforcement

Teams can first run the product in an observe mode to see how proposed policies would affect agents without blocking work. When enforcement is enabled, Kontext says unauthorized actions are denied before execution and each decision is written to an audit log.

That task-aware layer addresses a gap in conventional identity and access management. Human-oriented systems often decide whether a credential can reach a resource; an autonomous agent may authenticate once and then perform many actions across several systems without a person approving each step.

42CAP led the funding round. Andreessen Horowitz's a16z CSX accelerator and High-Tech Gründerfonds also participated. Kontext plans to use the capital to expand its engineering team. The company says its software currently works with coding agents including Anthropic's Claude Code and OpenAI's Codex; individual developers can use it free, while team plans start at $149 per month.

Those product and pricing details come from the company and have not been independently tested here. The investment is a bet that agent security will need to evaluate purpose and behavior, not just whether a tool call came from an authenticated identity.

Today's Rank

—

Not ranked yet

3 Sources

@SiliconANGLEKontext raises $4M to control what AI agents are allowed to do inside businesses https://ift.tt/4TbaG9n

@TFNBreakingNews🇩🇪 Kontext has raised $4 million in seed funding. https://techfundingnews.com/kontext-4m-seed-ai-agent-security-banking-fintech It was led by 42CAP and @a16zcrypto, a16z CSX, as well as @HTGF also participating. #ai #eutech #vcfunding #innovation #investment #startups #aiagents #banking #security #fintech

@SecurityWeekKontext Security Emerges With $4 Million for AI Agent Runtime Controls https://www.securityweek.com/kontext-security-emerges-with-4-million-for-ai-agent-runtime-controls/

Today's Rank

—

Not ranked yet

Combined views

1.2K

3 Sources, first seen 7h ago

1 likes

TLDR

Featured Source

Kontext raises $4 million to police what AI agents do inside businesses

TLDR

From observation to enforcement

Kontext raises $4 million to police what AI agents do inside businesses

3 Sources

Combined views

From observation to enforcement

Combined views

Sentiment

Summary

TLDR

From observation to enforcement

Sentiment

Summary

3 Sources

Sentiment

Summary