← Member help · Chat

Product front door — market story (OmniLore)

Member SoT key: tribal:shared:member:product-front-door
As-of republish: 2026-08-17T20:30Z
Also: docs/public/PRODUCT_FRONT_DOOR_MARKET_STORY_2026-08-11.md
Maintenance: docs/RUNBOOK_MARKET_FRONT_DOOR_MAINTENANCE.md · bin/market-front-door-check.sh

Law: A good front-door story explains OmniLore in 30 seconds without requiring HOOTL, MCP, SoT, dual-seat, OpenBao, or CoTrinity first.

Not: clean governance-complete public release · multiparty HSM · plant ops encyclopedia.

Headline

> Private AI for teams that need answers they can trust and actions they can control.

Strongest short story

> OmniLore helps teams use AI on private knowledge without losing control.
> It combines grounded answers, task-specific agents, permissioned tool use, and private continuity across sessions—in a workspace designed for sensitive operations.

30-second story

> OmniLore is a private AI workspace for teams that need trustworthy answers from approved knowledge — and controlled help with documents, tasks, images, and internal tools.
> It runs under your organization’s control, grounds answers in sources you approve, gives each agent only the permissions it is allowed to use, and can pick up where you left off without treating chat history as company truth.
> Join by invitation into a signed-in workspace — not anonymous public chat.

Short product pitch

OmniLore gives your team a private AI workspace that can search approved company knowledge, help with documents and code, generate downloadable documents (including PDFs), create and edit images when enabled, create task-specific agents, and interact with internal tools under strict permission controls.

Unlike generic chatbots, OmniLore keeps product truth grounded in approved documentation, separates user/member access from operator access, and fails closed when a tool or source is not authorized. Use the full OmniLore chat mode for product answers — speed-only modes are not a substitute for approved sources. You can optionally connect your Drive and Gmail for private search (never shared as the company encyclopedia). Personal continuity (preferences and prior work) stays scoped to you — it does not replace approved product documentation.

Pillars

| Pillar | Plain-language meaning |
|--------|------------------------|
| Trusted knowledge | Answers are grounded in approved sources — not empty-search invention. Prefer full OmniLore mode for product truth. |
| Permissioned agents | Each agent only gets the tools and data it is allowed to use — and stops when not authorized. |
| Private control | Knowledge, models, secrets, and workflows stay under your organization’s governance. |
| Personal continuity | Agents can resume your prefs and prior work across sessions — private to you, never a substitute for approved company knowledge. |

Why OmniLore is better (honest)

No slogans. Concrete differences you can verify without seeing our plant internals.

| Claim | What we mean | What we do not claim | How you can check (proof surface) |
|-------|--------------|---------------------------|-------------------------------------|
| Grounded product answers | Org / product questions are answered from approved documentation, not from inventing a story when sources are missing. | Absolute accuracy / “no errors ever” | Ask product questions in full OmniLore mode; expect sources or an honest miss — not a polished fiction. |
| Permission ends the path | Agents and tools run under least privilege. When a tool is not allowed, the system is designed to stop, not improvise admin power. | Unbounded company agent with admin power | Try a denied action in a least-privilege workspace; you should get a refusal, not a silent bypass. |
| Continuity without policy forgery | The system can resume your work and preferences across sessions. | Chat history is not the company encyclopedia | Use continuity / resume features; compare with product docs — prefs ≠ policy. |
| Private knowledge stays private | Optional personal Drive/Gmail search is yours only — never treated as shared company truth. | “We train a public model on your mail” | Personal Google is optional, scoped to you, and labeled private (see member how-to). |
| Org-controlled workspace | Join by invitation into a signed-in workspace under your organization’s control. | Unrestricted public free-for-all GA | Onboarding is invite-based; there is no anonymous public “open chat” as the product. |
| Honest maturity | We ship a controlled member / team beta. Capabilities vary by role and what your org enables. | Clean public GA / “everything finished” | Front matter: clean_release_go: false · member_beta_go: true. |

Taste rule: if a sentence needs hype adjectives or absolute guarantees, delete it and write the limit instead.

Best first use cases

  1. Ask organizational knowledge — policy, project status, product truth — with sources / fail-closed when unavailable (use OmniLore mode for these questions).
  2. Build a private task agent — research, document review, triage, team workflows.
  3. Connect tools under permission — read, draft, summarize, prepare actions — not uncontrolled automation.
  4. Resume your work — private continuity across chats (not inventing org policy from old threads).
  5. Create documents & images — generate PDFs/docs from chat; create or edit images when your workspace enables it.
  6. Search your private mail & Drive — optional personal Google connection; yours only, never company encyclopedia.
  7. Research the public web (when allowed) — cited external research; drafts stay untrusted until you promote them.

Also available (when enabled for your role)

These are real product paths for members — not homepage lead claims. Availability depends on membership tier and what your organization has turned on.

| Capability | What you can do | Honest limit |
|------------|-----------------|--------------|
| Scheduled jobs | Schedule private notes, personal briefs, or capped prompts for later | Results land in your workspace files — v1 does not guarantee a chat popup at fire time; timed digests are not “automated deep research” |
| Project cards & handoffs | Start a project intent card, track next steps, hand off work between sessions | Retrieval/posture helpers — not a predictive ML user model |
| Speech in / speech out | Speech-to-text and text-to-speech where installed | Capability varies by install; not every seat has every engine |
| Desktop screenshot assist | With your enrolled machine and a grant, capture a screenshot for the agent to inspect | Not free remote desktop or click-control of any PC |
| Public-law orientation | Practice-shaped workflows over public law text when offered | Not legal advice · not case files / PHI |
| Sidebar library | Bookmarks, saved prompts, personal chat skills / memory panels | Personal chat UX — not fabric product encyclopedia |
| Password self-service | Change your own password from chat under your signed-in identity | Bound to you — not admin resets for others |
| Answer feedback | Thumbs / ratings help operators improve the system | Does not auto-edit product truth or train a public model from your private chats |

How-tos: FEATURES_USER · SCHEDULED_JOBS_CHAT_V1 · PROJECT_INTENT_POSTURE_AND_CONTINUITY · DESKTOP_SEAT_CAPTURE · PUBLIC_LAW_WORKFLOWS · PASSWORD_SELF_SERVICE · CHAT_SIDEBAR_SKILLS_PROMPTS_MEMORY_BOOKMARKS · CONTINUITY_MEMORY.

One-page structure (public)

  1. What it is — Private governed AI workspace.
  2. Who it is for — Teams with sensitive docs, internal tools, code, or compliance needs.
  3. What it does — Search knowledge, create agents, work with docs/code/tools/images, personal Google when linked, resume continuity, careful web research when allowed, plus schedules, speech, and grant-bound desktop capture when enabled.
  4. Why different — Source-truth grounding, least-privilege tools, private control, and personal continuity that never pretends chat history is company policy.
  5. Available now — Controlled member / team beta via invitation (not unrestricted public GA).
  6. What it is not — Not a public chatbot. Not legal/medical/financial advice. Not uncontrolled automation. Not a chat-history “brain” that replaces approved knowledge.

What not to put on the front page

Do not lead with: AI OS · 700 tools · HOOTL · MCP · dual-seat fabric · Qdrant/Goldie/Sisyphus · OpenBao · clean release · residual accepted.

Those belong in architecture / security / operator docs.

Developer door (pointer)

CoTrinity is the IDE product line (trusted programming assistant). See
PRODUCT_POSITIONING_FRONT_DOOR_AND_COTRINITY · COTRINITY_IDE.
Do not make CoTrinity the only company homepage story.

Honesty residuals (market)

| Topic | Honest line |
|-------|-------------|
| Actions | Permissioned / often draft-prepare; not free merge/deploy autonomy |
| Email/code | Real paths; maturity varies by role and install — lead with knowledge + agents |
| Deployment | Organization-controlled / private workspace — not “random public SaaS chat” |
| Cursor IDE | CoTrinity supports Cursor; Cursor may still require a Cursor account (VS Code preferred for fewer third-party walls) |
| Continuity | Private session continuity is real; it is not product encyclopedia authority |
| OmniLore mode | Product/trust answers need full OmniLore (tools) mode — not Auto/speed-only alone |
| Personal Google | Optional; private to you; not tribal / not product SoT |
| Web research | Citations when allowed; still untrusted external content until you promote |
| Images / PDFs | Product paths exist when the plane is available to your role |
| Schedules | Workspace artifacts; not guaranteed chat popups; digests are not automated deep research |
| Project / posture | Intent cards and handoffs help continuity — not ML prediction |
| Speech | STT/TTS when installed — not universal on every seat |
| Desktop capture | Grant-bound to your enrolled machine — not remote control of any PC |
| Public law | Orientation on public text only — not legal advice or case PHI |
| Chat sidebar | Bookmarks/prompts/skills are personal UX — not fabric SoT |
| Password | Self-service for your identity only |
| Feedback thumbs | Improves ops awareness — does not auto-rewrite product truth |
| Availability | Controlled member / team beta — not unrestricted public GA |
| Front-door freshness | Process + machine check keep market story lag-bounded vs living product narrative during heavy development |

Process: RUNBOOK_MARKET_FRONT_DOOR_MAINTENANCE · decision DECISION_HONEST_MARKET_FRONT_DOOR_PROCESS_2026-08-17 · check bin/market-front-door-check.sh.

End market front door.