Common Room

Breadth across classes, tiered rather than uniform, with a published library behind it. Every plan connects to a CRM, a sales engagement platform and Slack out of the box, with a filterable integration directory and comprehensive coverage reserved for the top tier. Inbound, the platform pulls first-party data from CRM, product, marketing and engagement systems, second-party activity from GitHub, Slack, Discord and social platforms, and third-party intent through Bombora Company Surge.

Outbound, it writes into the CRM, fires Slack and email alerts, posts webhooks, surfaces agents inside Salesforce and Slack, and puts output into the browser through a Chrome extension. Custom integrations are available on every tier, and the MCP server extends reach to assistants. Tiering: select integrations on Essential and Advanced, comprehensive on Enterprise, with product signals an add-on below Enterprise.

Sourcecommonroom.io/integrations and commonroom.io/pricingread 2026-09-04

Named agents hand work to each other, with an automation layer underneath and a governance layer above. RoomieAI ships as four entitled capabilities rather than one assistant: Capture for account and contact research against granular criteria, Research, Activate for real-time personalized outreach built from that research, and Messages. A DataAgent runs alongside on data quality and deduplication.

Actions, automated workflow templates and automated custom workflows execute the result, with unlimited workflows and segments on every plan, and a Revenue Control Plane deploys plays and governs execution across large rep populations. Reps act on the output in the same interface rather than switching consoles. The pricing matrix names the individual agents as separate entitlements.

Sourcecommonroom.io/product/actions, commonroom.io/product/ai and commonroom.io/pricingread 2026-09-04

The core mechanism is identity resolution, not retrieval alone. Person360 turns partial identities and fragmented signals into resolved people and accounts, joining a GitHub handle, a Slack member, a website visitor and a CRM contact into one buyer, and Context360 keeps the resulting picture enriched and current.

Grounding draws on the customer's own systems (CRM, product usage, marketing and engagement) alongside second-party community activity and third-party intent, with enrichment and premium phone number lookup layered on. The company states the purpose plainly: buyer intelligence built for consumption by language models, accurate enough that agents can act on it without inventing. Custom product entities let a customer model their own objects into that graph.

Sourcecommonroom.io/product/person-360 and commonroom.io/product/enrichmentread 2026-09-04

Control is exercised at configuration and by the rep, with no gate in between. Real constraints exist: scoring weights are adjustable so the team sets which accounts agents work, workflows and segments are customer-defined, the Enterprise tier is positioned around governing execution across large teams, and role-based access controls what each user may configure.

In the common motion the agent researches and drafts and a rep acts on the result, rather than the platform dispatching unattended. None of that is an approval mechanism. No review queue, sign-off step, confidence threshold that withholds an output, blocked-topic policy or escalation rule is documented anywhere on the product pages or the pricing matrix, and the agents are described as running continuously.

Sourcecommonroom.io/product/enterprise and commonroom.io/pricingread 2026-09-04

Both certifications and access controls are documented, and the access surface is specific. The security page states support for single sign-on and role-based access control alongside SAML and SCIM through multiple identity providers, naming Okta and Microsoft Azure Active Directory. The pricing feature matrix carries SSO with GSuite and GitHub and a SAML/SCIM security line as entitlements.

The customer also holds data controls: the vendor says it provides the controls needed to manage data access, retention and deletion across a team. Behind those controls sits a SOC 2 certification covering security, availability, processing integrity, confidentiality and privacy, along with stated GDPR and CCPA compliance and encryption in transit and at rest.

Governance around person-level identity resolution is a stated design focus, which matters for a product whose core function is resolving individuals across platforms, and the vendor states all integrations use official APIs and conform to third-party terms of service. The SOC 2 type is not stated on the security page, and the report is available on request rather than published.

Sourcecommonroom.io/security and commonroom.io/pricingread 2026-09-04

Decision transparency at the prioritization layer, and no record of what the agents did. Account and contact scoring is transparent and adjustable, so a team can see how a ranking was derived and change the weights rather than accept a black box. Custom reports plus in-product and auto-recurring data exports let the underlying data be inspected outside the console.

Nothing traces agent behavior: no log of what RoomieAI researched or wrote for a given contact, no record of which workflow fired and why, no configuration change trail and no retention control over such a record. A customer can reconstruct why an account was ranked where it was, not what was then said to whom. Data exports move the data out for analysis elsewhere, which is not an audit surface over the agents.

Sourcecommonroom.io/product/lead-scoring and commonroom.io/pricingread 2026-09-04

A resolved identity corpus, not agent memory. Person360 and Context360 hold resolved people and accounts enriched continuously, so signal history and enrichment build up across sessions. That corpus is grounding data, and an accumulating data store is not memory. No memory construct the RoomieAI agents write and read back as their own state, no per-agent recall across sessions and no retention or forgetting control at the agent level is documented.

Sourcecommonroom.io/product/person-360read 2026-10-01

Multi-tenant cloud with no location or topology control offered. No region selection, EU or other in-region option, private VPC, single-tenant tier or on-premises route is published, even among the Enterprise plan's entitlements.

What the vendor documents on data is control rather than location: encryption in transit and at rest, GDPR and CCPA compliance, and customer controls over data access, retention and deletion.

Worth noting for a buyer: this platform resolves named individuals across public and private sources, and residency is a common requirement for that kind of processing in Europe.

Sourcecommonroom.io/security and commonroom.io/pricingread 2026-09-04

Assets that arrive built and assets the customer picks up, both present. Shipped and named: the RoomieAI agents as separate entitlements (Capture, Research, Activate and Messages), plus DataAgent, Prospector, Spark Briefs and Ask CR Anything.

Adoptable in product: automated workflow templates are a matrix line item on every plan, so a customer starts from a supplied automation rather than a blank canvas. Browsable outside it: a Playbooks library, a Signal Guides collection and an AI Prompts library, all published as standing resources rather than one-off posts. Buyers can browse and select on both fronts.

Sourcecommonroom.io/pricing and commonroom.io/resources/playbooksread 2026-09-04

Wide intake and output, listed as product surfaces. Signals in: job changes and champion tracking, website visits with de-anonymization, dark funnel activity, product usage signals, Bombora Company Surge intent topics tiered by plan, custom signals, aggregated signals, and social and community engagement.

First-party feeds arrive from CRM, product, marketing and engagement systems. Output and channels: alerts, workflows and segments, Slack and email alerts, Spark notifications and briefs, a Chrome extension, embedding inside Salesforce and Slack, webhooks, and the MCP server. Agents run continuously against that stream rather than waiting for a manual pass.

Sourcecommonroom.io/product/signals and commonroom.io/pricingread 2026-10-01

The customer does not choose the model, and no provider is named. No model list, routing policy, per-agent selection or bring-your-own-key option appears on the RoomieAI product pages, the pricing matrix, the security page or the developer documentation, and RoomieAI capacity is sold as research credits rather than as access to a choice of models.

The company's positioning explains the shape: it presents itself as the data layer feeding language models, buyer intelligence built for consumption by AI, so the models are an implementation detail it does not expose. The MCP server lets a customer's own assistant call Common Room, which is a choice of interface, not of the model Common Room runs. Because no provider is disclosed, this rests on consistent silence rather than a named single dependency.

Sourcecommonroom.io/product/ai and commonroom.io/pricingread 2026-09-04

Four external surfaces, all documented and all first-party. A versioned REST API with a published reference at api.commonroom.io/docs/api-v2.html, reached from a developers page and a docs hub.

An MCP server shipped as its own product page and listed in the pricing matrix as included on every plan, so a customer's assistants and agents reach buyer intelligence wherever they work rather than only in the console. A CLI, also a matrix line item. And webhooks, included across plans.

Around them sit a Chrome extension and embedding inside Salesforce and Slack, which are delivery surfaces rather than extensibility. No SDK or client library is published.

Sourceapi.commonroom.io/docs/api-v2.html, commonroom.io/product/mcp-server and commonroom.io/pricingread 2026-09-04

An optimization loop the customer holds, with nothing that tests before the fact. Scoring models are transparent and adjustable, so a team tunes weights against what actually converts and reads the effect on ranking, and custom reports plus data exports let outcomes be analyzed. That is a mechanism with a criterion aimed at the business outcome, not just a dashboard.

Nothing is aimed at the agents themselves: no controlled comparison of one agent configuration against another, no scored evaluation set, no benchmark, no regression check when scoring or prompts change, and no sandbox or dry run before RoomieAI output reaches a prospect. The company's claim that its data layer is accurate enough to eliminate hallucinations is unquantified, a property claim rather than a test the customer can run.

Sourcecommonroom.io/product/lead-scoring and commonroom.io/product/airead 2026-09-04

The extension delivers into the rep's browser rather than driving it. Common Room's Chrome extension surfaces resolved buyer intelligence and RoomieAI output alongside whatever a rep is already looking at, so the person acts and the agent informs.

Everything the platform does on its own runs through programmatic routes: CRM and sales engagement writes through integrations, Slack and email alerts, webhooks, the REST API, the CLI and the MCP server. The vendor states that all integrations use official APIs and conform to third-party terms of service, which is the opposite of driving an interface without one. No headless browsing, screen or click automation, portal navigation or form filling in third-party software appears anywhere on the property.

Sourcecommonroom.io/product/chrome-extension and commonroom.io/securityread 2026-09-04