Ri Koh — Your Organization's Intelligence Layer
Ri Koh turns your scattered documents, conversations, and tribal knowledge into instant sourced answers — and ships ready-to-use artifacts (white papers, briefings, slide decks, images) that sound like your team wrote them. Federal-grade sovereignty. Multi-tenant isolation. Your data never trains a model.
Ask Anything, Get Sourced Answers
RAG over your documents, knowledge base, conversations, and (optionally) the open web. Every answer cites sources you can audit.
Draft Anything in Seconds
12 artifact types — white papers, briefings, study guides, slide decks, narrated slides, FAQs, timelines, infographics — generated from your own documents in your own brand voice.
Generate Images on Demand
Nano-banana (Gemini 2.5 Flash Image) integrated into chat. Covers, hero illustrations, diagrams. Save what you like, discard what you don't.
Two AI Personas, One Workspace
Ri Koh — precise, professional knowledge advisor. Marvin — the depressed-but-brilliant Hitchhiker's Guide android. Both speak as members of your team, not outside vendors.
Federal Pipeline Built In
SBIR / STTR / grants.gov scanners, opportunity scoring, proposal drafting, and Slack pipeline digests. Designed for GovCon teams.
Per-Tenant Data Isolation
Each customer gets a physically separate D1 database, R2 bucket, and Vectorize index — not just logical row-level scoping. Architecturally impossible for cross-tenant leakage.
The pain Ri Koh solves
- Knowledge trapped in PDFs, Slack threads, Drive folders, and employees' heads
- AI tools that hallucinate or expose your data to vendors who train on it
- Federal customers requiring FedRAMP-aligned, region-pinned, opt-out architecture
- Marketing and proposal teams reinventing content from a blank page every week
- Onboarding new hires who can't find what they need in your knowledge base
What you get on day one
- Branded workspace with your logo, colors, tagline (white-label)
- Indexed knowledge base seeded from PDFs, Word docs, Drive folders, public URLs
- Chat & artifact pipeline grounded in your documents
- Three AI sovereignty tiers — pick the right one per query, per workspace, or per customer
- Live in 3–5 working days with operator-assisted onboarding
Three AI sovereignty tiers — picked per query
| Tier | Where inference runs | Compliance | Best for |
|---|---|---|---|
| Anthropic API | Anthropic-managed AWS, US-region | SOC 2 Type II · ISO 27001 · ISO 42001 · train opt-out | Standard commercial workloads — fastest, simplest, lowest latency |
| AWS Bedrock | Customer-pinned AWS region (commercial or GovCloud) | SOC 2 Type II · FedRAMP High (GovCloud) · HIPAA-eligible · train opt-out | Federal · Defense · Healthcare · regulated industries |
| Local LM Studio | On-premise — your hardware, your network | Inherits your controls · model and weights never leave perimeter | Restricted-sensitivity data · CUI · attorney-client privileged · air-gap |
Plans — modular by need
Core chat + knowledge base + basic artifacts + Contacts CRM. Anthropic tier only. Fair-use query allowance.
Starter + AI Action Plans + unlimited image generation + higher query allowance. Add modules à la carte.
Pro + Sovereignty Tier + custom DPA / BAA + dedicated support. Federal / regulated / GovCloud-eligible.
Ready to see it on your data?
30-minute discovery call. We'll show you Ri Koh on a real customer's seeded workspace and discuss what onboarding looks like for yours.