01
Control plane
Works in the browser
Individuals and teams use one workspace for project intent, crew activity, AI selection, token budgets, approvals, delivery evidence, and operational visibility.
SPENVEST LAB FLAGSHIP PRODUCT
Suroshna
Your AI software company.Meet Suro.
Suro is your AI software companion inside Suroshna: a calm, capable interface for turning an idea—or an unfinished codebase—into working software with a controlled Crew, browser workspace, real local runner, and infrastructure accounts you own.
It is a delivery platform, not an unchecked chatbot. Choose the agent, set the token policy, review the work, approve consequential actions, and keep the code, credentials, cloud, and deployment history portable.
- 10
- Crew roles
- 3
- Token profiles
- 0
- Silent infra changes
Suroshna browserLocal runner connected
Existing repository
Continue where the project stopped
main · local runner · project workspace
Active agent
GitHub agent
Routing
Auto optimise
Token mode
Balanced
Recovery plan
Safest next three steps
- 01 Confirm the current feature boundary and changed files.
- 02 Complete the smallest unfinished implementation slice.
- 03 Run the repository's real validation commands.
Writes require approval
BROWSER EXPERIENCE · LOCAL EXECUTION
One platform across four controlled planes
The browser stays simple for individuals and teams. The local runner does machine-level work. AI providers supply intelligence. Customer accounts remain the system of ownership.
02
Execution plane
Runs on the local machine
A secure companion service reaches the real filesystem, Git worktree, terminal, Docker, package tooling, and approved host capabilities—similar to the browser-plus-local pattern used by modern developer platforms.
03
Intelligence plane
Uses the agent you choose
Connected providers and project agents contribute intelligence through bounded context, project-scoped settings, provider readiness checks, and visible usage policy.
04
Infrastructure plane
Operates in accounts you own
Approved source-control, cloud, DNS, email, deployment, and production actions target the customer's connected accounts—not infrastructure hidden inside SpenVest Lab.
PRODUCT CAPABILITIES
From project state to approved next action
Start new or continue existing
Begin with a product brief or connect a real Git repository. Recover current state, likely unfinished work, and the safest next delivery slice before proposing changes.
Choose and switch AI agents
Use a connected AI provider, GitHub agent, or repository-specific agent. Pin one per project, switch deliberately, or enable policy-based routing.
Optimise token usage
Economy, Balanced, and Quality profiles bound context and response size. Repository fingerprints and caching avoid paying to repeat unchanged analysis.
Ten-role software crew
Product, architecture, engineering, data, QA, security, DevOps, and release responsibilities move through a dependency-controlled workflow.
Approval before consequence
AI output becomes a typed proposal. Policies and explicit approval control writes, commands, provider changes, paid resources, deployments, and releases.
Local execution and ownership
The companion runner reaches approved local tools and connected customer accounts. Code, credentials, repositories, and infrastructure stay under customer control.
CONNECT & CONTINUE
Resume the real project without erasing its history.
Connect a local Git worktree and let Suro recover enough context to continue. Initial inspection is read-only: no reset, checkout, stage, commit, or worktree write is needed to understand where the project stopped.
Credentials and remotes stay bounded
Remote credentials are sanitised before persistence, and secret-like values are excluded from summaries and AI context.
Git state
Canonical path, branch, HEAD, last commit, dirty state, and changed paths.
Project tooling
Detected languages, package managers, scripts, and repository conventions.
Repository agents
Project-specific agents discovered from repository configuration and offered for selection.
Recovery brief
Current state, likely unfinished work, safest next steps, and validation commands.
AGENT CHOICE
Use the AI agent that fits the project.
AI selection is project-scoped. Switch providers without losing the delivery state, use a GitHub agent where it has repository context, or let routing policy choose a ready provider for suitable work.
Connected AI providers
Bring compatible provider access and choose the model and project policy.
GitHub agents
Use existing GitHub authentication and repository-specific agents without storing a GitHub token in Suroshna.
Auto-optimise routing
Prefer a ready provider based on project policy, work type, context need, and visible budget.
Provider credentials remain behind approved local or provider authentication boundaries and are not copied into project prompts.
TOKEN CONTROL
Spend context where it changes the answer.
Suroshna exposes budgets instead of treating AI usage as invisible. Profiles bound context and response size while caching avoids repeating unchanged repository analysis.
Economy
Smallest useful context
Fast triage and narrow project questions
Balanced
Default bounded context
Everyday recovery and implementation planning
Quality
Expanded approved context
Deeper analysis when the decision justifies it
THE SOFTWARE CREW
Ten roles. One accountable workflow.
Each role produces typed artifacts. Work becomes ready only when its dependencies and required approvals are satisfied.
Product Manager
Problem, users, scope, and product decisions
Solution Architect
System design, risks, and trade-offs
Engineering Manager
Dependencies, sequencing, and integration
Frontend Engineer
Accessible browser product experience
Backend Engineer
Application services and boundaries
Database Engineer
Schema, migrations, integrity, and recovery
QA Engineer
Behaviour, regression, and release evidence
Security Engineer
Threats, access, secrets, and data handling
DevOps Engineer
Local runner, environments, and delivery
Release Manager
Release gates and operational readiness
PRODUCT STATUS
A real product, built for anyone creating software.
Suroshna is not a founder-only MVP service. It is designed for individuals, founders, developers, product teams, and organisations starting new software or continuing a real existing project.
Suroshna Local is the current working product. Public account, download, and browser-to-runner onboarding are being released in stages. Suroshna Cloud, Suroshna Managed, and Suroshna Enterprise remain named future product boundaries—not claims of released capability.
We do not label cloud, DNS, paid-resource, source-control, or production mutations as complete until they are exercised against real connected accounts with approval, rollback, and evidence.
Working foundation
Built beneath the product vision
- Native local application foundation with typed boundaries and durable SQLite state
- Canonical, read-only inspection of an existing local Git repository
- Durable recovery briefs, restart resume, and repository-aware terminal context
- Project-scoped AI agent, model, routing, and token-profile selection
- Optional GitHub Copilot CLI and repository custom-agent discovery
- User-controlled host terminal sessions, approval policy, and redacted local activity
OWNERSHIP MODEL
You own the code. You own the cloud. Suro handles the complexity.
Repository
Your Git history, branches, conventions, and source-control organisation.
Infrastructure
Your cloud, DNS, email, deployment, and production resources.
Credentials
Your authentication boundaries, provider permissions, and revocation controls.
Portability
Your delivery evidence and ability to continue with a different tool or team.
Suroshna by SpenVest Lab
Continue the project. Keep the control.
Use Suroshna for a new product, an unfinished repository, or software already in operation. Product access requests are open to individuals and teams.