SolonGate

Security Gateway for AI Execution.

Project Details

Solongate is a fully deterministic security gateway for AI agents
LLMs hallucinate You cannot trust their reasoning. That is why we don't. SolonGate runs directly between the agent and its tools. We intercept every tool call and evaluate the payload against user defined policies in under 5 milliseconds. If the action is dangerous or unauthorized, we terminate it instantly.

We are getting ready to open source SolonGate very soon.

As two 19year old cofounders, we have been building agentic products for 3 years. We discovered this problem when our databases were wiped out by a prompt injection while using Claude Code :) and now we can solve this issue very strictly. As all research supports, trying to make LLMs flawless is an endless and wrong solution path. We solve this problem entirely with deterministic and complex security firewalls. SolonGate can be run locally. With many additional features like data masking you can run your agents under complete protection

Also, you can use SolonGate for free on our website.

solongate.com

Theory of Impact

The true risk from AI is giving unpredictable models direct access to real world infrastructure. You can never 100% align an llm. It will eventually hallucinate, get hijacked by a prompt injection or make a catastrophic logical error

Our theory of impact is practical Stop trying to fix the AIs brain. Handcuff its hands instead!!

Here is a small example

// Agent attempts a destructive tool call:
{
"tool": "db_client",
"action": "execute_sql",
"payload": "DROP DATABASE production_data;"
}

// SolonGate intercepts at the execution layer in <5ms:
{
"solongate_status": "BLOCKED",
"action_taken": "TERMINATED",
"rule_triggered": "prevent_destructive_sql",
"latency": "2.4ms",
"message": "Execution denied. Payload contains unauthorized DROP command."
}

Btw this is just our policy layer. SolonGate actually has like 8 more layers DLP, PII masking, rate limiting, budget controls, audit logging, approval workflows, semantic checks

People

Emirhan Demir

Team Member

Grants Received

Only visible to verified funders, reviewers, and admins.

Email hi@grantmaking.ai to get verified

Funding Asks

grantmaking.ai Launch Round

Applied

Minimum

$15,000

Ideal

$50,000

How the money will be spent

We have to test SolonGate on many different OS environments to make sure our execution layer is bulletproof. This funding goes directly to renting various dedicated servers and VMs for cross OS testing. The 15k minimum covers our baseline infrastructure for this. The 50k ideal amount covers extended testing environments and gives us the runway to ship our open-source tier faster without worrying about immediate revenue.

Extra note We both dropped out of our universities. We are building SolonGate full time

Discussion

Private comments in this thread are only shown to approved funders and grant reviewers. Email hi@grantmaking.ai or to request access.

Grant ReviewerPrivate17h?