Sakshi is the governance layer for AI agents in Indian enterprises. Register every agent, witness every decision on a tamper-evident chain, bound autonomy with human oversight, and generate the evidence your auditors and regulators ask for, all inside your own environment.
The platform
Not another model or another dashboard: the accountable record beneath the agents you already run. Five modules, one tamper-evident spine.
A live inventory of every agent and model: owner, autonomy tier, provenance, cryptographically verified identity. Nothing runs unregistered.
Every decision hashed into a chain anyone can independently verify. PII is tokenized before it is ever stored.
Envelopes route each action to auto, human review, or block, with oversight telemetry and a drilled kill switch.
Generate signed RBI MRM, DPDP, SEBI and IRDAI evidence packs from live evidence, on demand, clause by clause.
Declared-first bias screens (gender, caste category) gated on statistical significance, so they don't cry wolf.
Who it's for
See exactly which obligations your live evidence covers, close the gaps in leverage order, and generate the signed packs your regulator or auditor asks for. Deepest coverage today for RBI MRM, DPDP, SEBI and IRDAI.
Get your readiness scoreIntegrate in an afternoon. A few lines with the Python SDK, or zero code through the gateway, including an MCP server for agents that speak MCP. Both on PyPI today.
Read the docsYour environment, your data. Deploy in your VPC or on-prem; evidence never leaves India; every record is tamper-evident and independently verifiable, even by an auditor who doesn't trust us.
See the architectureWhere it applies
Agent governance is the same everywhere: know every agent, witness every decision, bound autonomy, prove it. BFSI is where our coverage goes deepest today, but the same spine governs agents in any industry.
Banks, NBFCs, cards, lending, and securities. Clause-mapped to RBI MRM, DPDP, SEBI and IRDAI: our deepest coverage, and the arc you can explore in the live demo.
Banking & finance solutionsDiagnostic, claims, and patient-facing agents. DPDP-aligned by design: PII is tokenized at ingest and never stored raw, and every decision stays on the record.
Healthcare solutionsAccountable, auditable automated decisions, with RTI-ready evidence exports and independent verification a citizen or auditor can check for themselves.
Government solutionsTelecom, logistics, technology, and beyond. If AI agents make or shape decisions in your business, Sakshi gives them an accountable record. Governance is the same everywhere.
Enterprise deploymentIntegration
Sakshi never asks you to rewrite an agent, and never sees a raw identifier: values are tokenized at ingest.
Wrap the model and agent calls you already make. Register once, then witness decisions and enforce autonomy envelopes from inside your process.
Put an un-instrumentable or third-party agent behind the interception proxy, or add the Sakshi MCP server to any MCP host. Governance and enforcement without changing a single line of the agent.
Full developer docs & API reference
Backed by open research from Rota Labs: the governance models are Apache-2.0 on Hugging Face.
Regulatory evidence
Each instrument is encoded as data and scored against your live evidence, a readiness score you can act on, and signed packs you can submit.
Model inventory, validation evidence, AI disclosure & human handoff, red-teaming, a drilled kill switch.
Rule 13 SDF due-diligence: DPIA from live evidence, minimization & localization screens, signed audit bundle.
Sole responsibility for AI/ML outputs, including third-party models. Vendor model identity, verified.
FREE-AI read-across skeleton, ready to re-cut the day the working-group report lands.
Deployment & data residency
A governance product you can't trust is a contradiction. So Sakshi is single-tenant and installs in your environment, we never see your evidence, and couldn't hand it over if compelled.
The chain recomputes every hash. An auditor, or a regulator, can verify your evidence without trusting Rotavision at all. Tamper a single field and the chain says exactly where.
Try it with synthetic data
We'll spin up a private, synthetic-data sandbox and email you the link: register an agent, witness a decision, generate an evidence pack. No procurement, no real data, no risk.
