Deployment and security
Your code stays where it is. So do the results.
Sahasa.AI runs inside your own cloud, on-premise, or hosted by us. This page covers what each involves, what leaves your network, how a pilot runs and how we charge.
Self-hosted · what crosses the boundary
Three ways to run it.
Fastest start
Managed
An isolated tenant that we operate. Connect a Git provider and a requirements source, use your AI keys or ours, and run the same day. Generated tests still land in your repository.
For regulated and security-sensitive teams
Your cloud or on-premise
Deployed inside your own cloud or data centre, on AWS, Azure, GCP or your own hardware. Code, requirements and results never leave your network.
When you want the outcome, not another tool
Embedded QE team
Either of the above, plus a small senior team of ours inside your organisation. They run the platform, curate suites, triage findings and build the adapters and quality gates you need.
Inside your boundary
What a self-hosted deployment needs.
Compute
- Runs on your infrastructure
Containers on your own VMs or cluster. We size it with you for the number of repositories you connect.
Data
- Stays in your environment
Results, reports and screenshots are stored inside your network, encrypted with your keys and kept on your retention schedule.
Access
- Read-only by default
A read-only token for your Git provider, with write access only to open pull requests if you want that, and a service account for Jira. Single sign-on to the platform arrives in December.
AI calls
- Your provider, your agreements
Anthropic, OpenAI, Google, or an internal gateway in front of Bedrock or Azure OpenAI. Prompts and responses are logged in your database.
Updates
- Signed images, on your schedule
Pulled from our registry when you choose. The licence file has an expiry date. If it lapses, the platform becomes read-only and your data stays put.
Mar 2027
- SOC 2 readiness
On the roadmap for March 2027. Until then, ask us for the architecture and controls documentation.
Pilot
Four weeks, one service, your baseline.
You pick the service. We deploy, connect and come back with results your team can check. The pilot is paid, and the fee is credited if you continue.
Week 1
Connect
Deploy in your cloud or connect to ours. Read-only access to one repository and its requirements. We record the baseline: coverage, open findings, time to get a red build green.
Week 2
Generate
Suites generated into your repository. Bugs, security and the threat model on every pull request from here on.
Week 3
Run
Suites run in your CI. Live UI checks on staging. A traceability matrix for the requirements you gave us.
Week 4
Decide
We triage a real failed build together, then put the new numbers beside the baseline. You decide.
You bring
- One engineer, about two hours a week
- A repository and a requirements source (Jira is fine)
- A staging URL
- An AI provider key, or permission to use ours
We bring
- Deployment and connection to your tools
- The week-one baseline and the week-four comparison
Commercials
How we charge.
Pricing depends on team size, where the platform runs and how much of the work you want from us, so we quote after a short conversation.
Licence
- Annual, per named seat
Every module included. No per-run or per-parallel charges.
AI usage
- Billed by your provider, to you
Model usage goes on your own account, so the AI cost is visible to you and carries no margin from us.
Onboarding
- Scoped before we start
Connecting your tools and fitting templates and quality gates to your process is a one-off, agreed up front.
Optional
- Vigil and the embedded team
Priced separately, so you only pay for what you use.
Next step
Bring one repository.
We'll walk you through Sahasa.AI on a sample project and talk through how it would fit your stack. If it looks useful, we can scope a pilot together.
- 01A short call. Your stack, your frameworks, and where quality hurts today.
- 02A demo. The platform on a sample project, and how it would connect to your tools.
- 03A pilot, if it fits. Scoped with you around one service.