Osmosis Frequently Asked Questions

The lag you're competing against: Official minutes are published 60 to 90 days after a meeting, and they're a heavily filtered derivative of what was actually said. Databases update even later.

The math: One $100M mandate at ~1% fees is roughly $5M in revenue over five years. The cost of missing the signal is measured in mandates, not subscriptions.

The difference: While your competitors are reading last quarter's minutes, you're referencing yesterday's IC discussion.

• 8:30 AM — Email alert: "Texas Teachers IC discussing increased private credit pacing from 5% to 8%. CIO wants options by next meeting."

• Includes video clip + exact timestamp

• Shows who said it (CIO vs consultant vs staff)

8:45 AM — Check the organization page:

• IC chair background and history
• Consultant relationships and recent recommendations
• Staff who cover your strategy

9:00 AM — Execute:

• Reference the exact quote from the meeting
• Align your pitch to stated constraints
• Loop in the partner with the relationship
• Week 2 — Meeting booked (they're impressed you knew their timeline)
• Month 3 — RFP issued; you're already in dialogue

Power users:

• Relationship managers hunting for warm leads
• PMs who want to know who's actually deploying capital
• Sales enablement and RevOps teams who operationalize signals into pipeline

You're ready for Osmosis if:

• ✓ You find out about mandates after competitors have pitched
• ✓ Your team works from static databases, not live signals
• ✓ Meeting prep means hours of manual research that's dated by the time it's done
• ✓ You're losing to firms that "somehow knew" about opportunities first

Allocation types we track:

• New commitments and re-ups
• Pacing changes and strategic shifts
• Manager graduations and emerging manager programs
• Consultant recommendations and searches
• Co-investment opportunities

The essence: A database entry is the end of the story: a commitment logged after it closed. The meeting is the beginning: the pacing debate, the consultant's nudge, the CIO's actual reasoning — months before any record updates. The window between what's said and what's published is where mandates are won.

Real timeline:

Monday: Oregon IC discusses increasing PE allocation

Monday afternoon: Osmosis alert to relevant managers

Tuesday: First movers scheduling calls

3 weeks later: RFP released

6 weeks later: Database updates show "considering PE increase"

How much deeper is the meeting than the minutes? Verified example:

• The CalSTRS January 2025 Investment Committee produced 23,625 words of searchable dialogue in Osmosis (403 transcript chunks), delivered same day. The official minutes: 1,456 words, published two months later.

• Freshness. Minutes publish 60 to 90 days after the fact, and they're a heavily filtered derivative of what was actually said. Click an allocator in a static database and the last meaningful update can be years old. Osmosis is same-day.

• Depth. An hours-long IC meeting becomes hundreds of searchable chunks of dialogue, not a one-sentence summary. Verified example: CalSTRS January 2025 IC — 23,625 words in Osmosis vs. 1,456 words of official minutes.

• Direction. A database tells you what an allocator did. Osmosis tells you what they're deciding — the pacing debates, consultant recommendations, and searches that haven't reached any published record yet.

Time savings: Prep that used to take hours of manual research happens in minutes, so your team spends the day in front of allocators instead of documents.

First-mover advantage: Alerts arrive the same day as the meeting. The first call that references what the CIO actually said sets the frame for every call that follows.

Competitive intelligence: You see who else presented, what consultants recommended, and how the board reacted — so you know where you stand before the RFP drops.

What we capture today:

• Public fund board and investment committee meetings (live + same-day video)
• CIO and staff interviews in public forums
• Consultant presentations at public sessions
• Regulatory filings, board packets, annual plans, and documents tied to allocation decisions

Coverage scope:

• Osmosis maps approximately $13.9 trillion across 58,600 U.S. institutional organizations and retirement plans. We have source-backed intelligence on $6.7 trillion of that universe, including high-signal meeting, audio, and media coverage on $6.0 trillion.
• Our deepest U.S. segment is public funds: $4.4 trillion across 250 institutions, including nearly every major state investment system.
• Separately, Osmosis tracks $7.3 trillion across 26 sovereign wealth funds, with source-backed content on $6.7 trillion.

What we never capture:

• Closed-door sessions
• Material non-public information (MNPI)
• Embargoed or privileged materials

• Anywhere an allocator talks publicly is fair game: IC and board meetings, CIO interviews, podcasts, regulatory filings, board packets, annual plans.

How we collect it: We never bypass paywalls, logins, or access controls. Where a public meeting or webinar requires registration, we follow the organization's prescribed process and accurately identify ourselves. We don't use proxies or obfuscation to mask who we are. Collection is reviewed against each source's terms of use, with outside legal counsel oversight.

Here's the part that matters: most meeting content is ephemeral. It's a video call that happens, and when it's over, it's gone. The only relic left behind is a set of minutes 60 or 90 days later.

• So while it's public data, it becomes proprietary data to us — because we were the ones who attended the meeting and captured it.

• US institutional allocators are the core: public funds, corporate and union plans, endowments, foundations, and insurers.

Endowments and foundations: Covered when they publish meetings, file, or speak publicly — including the university investment offices GPs target most.

Sovereign wealth: Covered when funds publish or their officials speak publicly.

Before a pilot: send us the accounts or coverage areas that matter to you and we'll onboard them ahead of the trial, so you're testing against your actual target list rather than ours.

• Our universe is US institutional allocators, excluding family offices and private wealth. Family offices are genuinely black boxes: if they're not talking publicly, there's nothing to index.

• We'd rather be excellent on the institutional market than thin everywhere.

What we track: Personnel. Every meeting we process is analyzed for hires, departures, and role changes, and that surfaces on the organization page alongside the consultants and managers they work with.

The pattern we see: Customers pair Osmosis with LinkedIn for contacts. For a growing number, the Rolodex was the last thing their legacy database was doing — and LinkedIn does it better.

• Every fact is referenced back to a source with a date, so you always know how fresh it is. Some things genuinely don't update often, and you'll see that.

• What you won't get is a summary that quietly rests on a 2022 document without telling you.

How we ensure quality:

Speaker attribution: We know who said it (CIO vs consultant vs trustee)

Source links: Click through to the exact timestamp or document page

Confidence scoring: We flag when we're uncertain

User verification: "Report Issue" button with rapid review

Models: Osmosis is powered by frontier LLMs accessed through commercial APIs — Google (Gemini), Anthropic, and OpenAI — with speech-to-text by Deepgram. We don't train our own foundation models, and we switch between models as they improve.

Your data never trains a model: Customer data is not used to train or fine-tune any model, and our AI providers operate under API terms that prohibit training on inputs.

No autonomous actions: The platform generates intelligence — summaries, rankings, alerts, answers. It never takes actions on your behalf. Every output goes to a person who decides what to do with it.

Quality monitoring: An automated system scores query responses as they're generated; lower-scoring responses are routed for human review. Weak answers also trigger coverage fixes — a thin result dispatches an audit of that allocator to find what we're missing.

• Through MCP, you control the model completely: you bring your own harness (Claude, Copilot), and Osmosis is the data underneath it.

Your data is yours:

• Your CRM, your investor data, and your documents never cross to us
• Customer watchlists, notes, and queries are ring-fenced to your account
• Never shared between clients
• Not used to train any model
• Full audit trail of who sees what

Personal data: We don't conduct behavioral monitoring or tracking of individuals, and we don't sell or rent personal data. Individuals appearing in indexed public content can object or request removal.

Week 1 timeline:

• Minute 1-2: Setup complete
• Hour 1: First relevant alerts arrive
• Day 1: Team trained on signal interpretation
• Week 1: First meeting booked from signal
• Week 2-4: Measurable pipeline impact

Monitor by email: Alerts come to you, personalized, within about an hour of a meeting wrapping.

Investigate in the app: When something's worth five minutes, that's where the raw data, citations, and video live.

Operationalize through MCP: Connect Osmosis to Claude or Copilot alongside your CRM, and your AI assistant works with a corpus that actually knows what CalPERS said last Tuesday. Customers use it today to pull fresh intelligence on active accounts and write it into their CRM records — your workflows, your environment.

Works with: Claude, Microsoft Copilot, and other MCP-compatible AI platforms. Native connectors available for enterprise accounts where required.

Read-only by design: We read from Osmosis sources, never from yours. Orchestration with your CRM happens in your environment; nothing proprietary crosses to us.

Real usage, first month: A $150 billion manager with ten people on the platform created 1,200 hours of research in their first month. Created, not saved — most of it was work that simply wouldn't have happened otherwise, because nobody had the hours.

What customers tell us:

• "Paid for itself in the first month"
• "One meeting covers the annual cost"
• "Should be 10× the price based on value"

Level 1 - Reactive (Industry norm):

• □ Learn about mandates from RFPs or news
• □ Chase searches after they're public
• □ Long fundraising cycles driven by cold outreach

Level 2 - Aware:

• □ Manually check meeting minutes and board sites
• □ Some proactive outreach, weeks behind the signal

Level 3 - Proactive (Osmosis users after 30 days):

• □ Alerts drive daily priorities
• □ Outreach references what the allocator actually said
• □ Faster qualification, fewer dead-end meetings

Level 4 - Predictive (Osmosis power users):

• □ Signals route automatically into team workflows
• □ In the room's head before the search exists

• Who else presented at recent board meetings

• Which managers consultants are recommending

• What strategies are gaining or losing favor

• When competitors get mentioned in meetings

Real example: One customer had pitched against the same two firms for five years and never seen their pitch. She found their presentation to a board in our data, watched it, and rebuilt her deck around six points of differentiation.

What you won't see: Private interactions, closed-door discussions, or anything that wasn't said publicly

Depth: High-signal meeting, audio, and media coverage on $6.0 trillion of U.S. institutional capital — including $4.4 trillion across 250 public funds.

Verified depth example: CalSTRS January 2025 IC — 23,625 words of searchable dialogue in Osmosis vs. 1,456 words of official minutes, delivered same-day vs. a two-month lag.

Real usage: A $150B manager's ten-person team created 1,200 hours of research in their first month on the platform.

• 15-minute preview — We'll show you today's signals for your strategy.

• Contact us to get started.

Architecture first: Osmosis is read-only. We read from Osmosis sources, never from yours. Your CRM, your investor data, and your documents never cross to us.

Attestation: We completed a SOC 2 Type 2 examination — operating effectiveness, not just design — covering Security, Availability, and Confidentiality. Full report available under NDA.

Security posture:

• Encryption in transit and at rest
• Role-based access control; MFA and SSO/OAuth supported
• Multi-tenant with logical data isolation — one customer's data and activity are never accessible to another
• Full audit logging
• Annual penetration testing

Data residency: The platform and its data are hosted in the United States, and AI processing is configured to US endpoints.

Data lifecycle: On termination, customer data is deleted or returned within 30 days of written request (per the MSA).

Compliance framework:

• Public information only (no MNPI)
• Source verification for every signal
• Customer data segregation
• GDPR/CCPA compliant
• SEC marketing rule aligned

Documentation available:

• SOC 2 Type 2 report (under NDA)
• AI DDQ and security documentation
• DPA execution
• Vendor assessment questionnaire