System of Context
In modern revenue organizations, context lives everywhere:
Private data in CRMs, product usage logs, billing systems, sales calls, emails, Slack threads.
Public data from news, blogs, hiring announcements, social signals, company websites, job postings.
Traditionally, these sources remain fragmented. Sellers and revenue teams waste time stitching them together manually, missing signals or duplicating effort. A System of Context solves this by unifying and continuously organizing all of that data into one coherent layer that can power agents, apps, and workflows.
What is the Data?
Private Data
CRM: Salesforce, HubSpot, Dynamics.
Product telemetry: feature usage, logins, error events.
Communication: Gmail, Slack, Zoom transcripts, call notes.
Financial systems: subscriptions, invoices, ARR.
Internal docs: Google Drive, Notion, Confluence.
Public Data
Market signals: press releases, job postings, funding rounds, news
Web activity: company blogs, LinkedIn updates, etc..
3rd-party intelligence providers.
Roxs' own enrichment layer
Data arrives via batch pipelines (via batch drops) or real-time streams (webhooks, event APIs).
Storage & Unification Layer
Once ingested, data must be organized in a way that supports both scale and reasoning.
Data Lakehouse holds raw and cleaned data across formats.
Entity Resolution & Deduplication: multiple systems may refer to the same “Company” or “Person.” A unification process resolves identities and merges duplicates.
Knowledge Graph is built:
Entities: Companies, People, Opportunities, Activities, Events.
Relationships: Person ↔ Company, Opportunity ↔ Account, News ↔ Company.
Attributes: ARR, usage trend, title, email, funding round date, etc.
This graph is continuously updated as new data arrives, so the context is always live and fresh.
Context Access Layer
Applications and agents don’t need all data — they need the right data, at the right time.
APIs expose queries like:
“Show me all recent interactions with Company X.”
“What are the latest public signals about this prospect?”
“Which accounts have rising usage but no recent sales touch?”
Context Windows can be scoped by:
Seller + Account (all context relevant to that pairing).
Time horizon (e.g., past 30 days).
Task at hand (e.g., pre-meeting brief vs. renewal forecast).
This ensures agents and apps pull in just enough data without noise.
Reasoning & Agent Layer
This is where the intelligence sits. Agents consume the unified context to decide, suggest, or act.
Scoped Agents: each aligned with a seller + account, or workflow (e.g., “Pipeline Agent,” “Research Agent”).
Capabilities:
Summarize context (pre-meeting briefs, account overviews).
Generate insights (churn risk, expansion signals).
Automate actions (draft outreach, log CRM notes, recommend plays).
Tools: APIs for writing back to CRM, sending emails, creating tasks.
Agents use both private signals (usage spike, missed renewal call) and public signals (new VP hire, funding announcement) together to provide rich, timely recommendations.
Engagement Layer (Apps & Interfaces)
The System of Context surfaces value where users already work.
Web App: full dashboard for searching accounts, reviewing signals, running plays.
Desktop / MacOS App: quick contextual insights, notifications, meeting prep.
Mobile (iOS/Android): on-the-go context before meetings.
Agent Interfaces: command palette (e.g. Cmd+K) to summon context instantly in Slack, CRM, or calendar.
This makes context ambient — always available, not a separate destination.
Governance, Security & Trust
Since the system touches sensitive data, strong controls are essential.
Role-based access control (field-level, source-level).
Hybrid deployments: SaaS vs. warehouse-native vs. customer VPC.
Data provenance & audit logs (know where every fact came from).
Encryption in transit & at rest, retention policies.
This ensures compliance while still allowing broad data usage.
The Flywheel Effect
Every new data source enriches the System of Context. Every action an agent takes creates more context (notes, decisions, outcomes), which feeds back into the system. Over time, the System of Context becomes the single most complete, live representation of customer reality, powering all revenue workflows.
Last updated