Hello, Osyrin
An introduction to what Osyrin is, why it exists, and what it's building toward.
#announcement#osyrin
Osyrin
The trust layer for agentic applications.
Osyrin is a platform for building applications where agents, retrieval, integrations, and people operate under the same security model—and where every decision is auditable, governed, and yours to trust.
What Osyrin makes possible
Two ways the same platform shows up.
Sarah, Tuesday morning.
Sarah leads enterprise sales at Acme. She opens Osyrin while her coffee is still hot. Before she's clicked anything, the morning Pulse tells her what's happened on her opportunities overnight—three risks her team flagged, two stage changes, and one anomaly the agents noticed on their own: a customer that's gone quiet for eight days when they usually reply within two.
She taps into the quiet account and gets a deeper Pulse on that one opportunity. She asks the chat a question—"what's our history with their procurement team on past renewals?"—and the agent answers from the entity graph, pulling in references from two previous deals and a Slack thread someone linked in last month.
She drafts a few instructions back to her team in the same window, closes the laptop, and heads to her 9am.
Nothing she just did required logging into another tool. Nothing the agents told her came from outside the boundary of what she's allowed to see.
Morning Pulse — Your opportunities
Tuesday, May 20 · Generated 4 minutes ago
Acme renewal moved to Final Negotiation
Driven by Marcus's note yesterday on procurement alignment.
Northwind expansion: 2 new risks filed
Legal flagged change-of-control language in the master agreement. Finance flagged a 14% gap between forecast and signed pipeline.
Helix Bio has been quiet for 8 days
Typical response time on this account is under 48 hours. Last contact: Tuesday last week (David, on pricing).
Stellar Logistics moved back to Detailed Review
The technical review surfaced an integration concern.
Showing 4 of 11 opportunities · 12 references · View all →
Maya, due diligence on a 3pm deadline.
Maya is an associate at a mid-market private equity firm. She's been pulled into a deal review where some of the documents in the data room—the carve-out financials, the founder employment terms—are above her clearance. The platform knows this. Her view of the deal room is filtered to the documents she's allowed to see, and the agent she chats with is constructed from the same view. It doesn't filter restricted documents out of its responses. It was never given access to them in the first place.
Reading through a customer reference call summary, she notices the named customer represents nearly a third of the disclosed revenue. She tells the agent in the chat—"add a risk: customer concentration. This one customer is 30% of revenue, and the contract has a 90-day termination for convenience clause." The agent has the tools to do this because she's on the deal review page—different pages give the agent different tools, all running with her permissions. It files the risk against the opportunity with citations back to the two documents that surfaced the concern.
The Risk now exists as a tracked item. It moves through a workflow—Open → Mitigated, Accepted, Escalated, or KilledTheDeal—with rationale required at every transition. The partner reviewing the deal at 4pm sees the risk, the evidence, and Maya's reasoning.
She has limited access to the data room. She still contributed to the deal.
Deal Pulse — Project Cobalt
Generated 12 minutes ago · 9 documents · 5 findings
Stage Detailed Due Diligence
Lead Partner: J. Lindqvist
Findings so far
2 Risks — 1 High, 1 Medium
2 Open Questions
1 Inconsistency between pitch deck ARR and audited financials
Recently filed
Customer concentration risk (Maya, 8 min ago)
One customer = 30% revenue · 90-day termination clause. Sources: Customer Reference Call, Master Services Agreement
Open question on cap table treatment (Maya, 22 min ago)
Awaiting answer from deal team
You have access to 6 of 9 documents · 3 restricted · View all findings →
Both stories are built from the same primitives. Declarative data models. Governed retrieval that respects classification at chunk-time, not after. Agents whose tools come from the page they're on, whose view of the world is constructed from the user's permissions before any data is touched. Workflows that turn agent output into tracked, governed work. A security model defined once in metadata and enforced everywhere—in the UI, in the API, in the chat, in the agent's view of the world. Osyrin is the platform underneath all of it.
Osyrin is not yet publicly available. Public access will be granted soon.
Questions or want to get in touch? Reach out at contact@osyrin.com.