FAQ
Common questions about the platform. For self-host operational questions, see the Self-host docs.
General
What is CCD-Suite, in one sentence?
A multi-module agent platform built on a knowledge graph, where every agent run sees the same four-namespace context (memory + skills + tasks + user) — so behavior stays consistent across surfaces, sessions, and modules.
Who is it for?
Agencies (10–50 people) that have outgrown point tools — a CRM, a PM board, time tracking, invoicing, SEO/social tools — and want their data
- agents to operate as one system instead of 8 glued-together apps. We expand from there into the broader services-firm and SMB market.
How is it different from ChatGPT + my CRM?
Three properties ChatGPT + CRM can't combine:
- Persistent, citation-backed memory — the agent remembers your tenant, and every claim links back to source rows.
- Confirmation-gated actions — agents propose, you dispose; no auto-firing emails.
- Cross-module reasoning — same graph layer means CRM × Content × Analytics queries work without glue code.
Cloud or self-host?
Cloud if you want managed everything (auth, scaling, model routing, weekly cross-org benchmarks). Self-host if you need data residency, want to run on your own hardware, or want the $0/mo TCO (BYO LLM key or bundled Ollama).
See Self-host vs cloud comparison for the full inventory.
Agents
Does the agent ever take action without my approval?
Read operations (semantic search, dashboard queries) run inline. Write operations (send email, post tweet, create deal) NEVER execute without your confirmation through the Inbox.
The single exception is the per-tenant auto-confirm allowlist for action kinds you've explicitly opted in (typically internal notes / internal task creation). Off by default; opt-in deliberately.
Can I bring my own LLM?
Yes, on self-host. Set AI_LLM_PROVIDER=openai (or anthropic)
in .env with your API key. Cloud edition uses the platform's
managed brain — your LLM provider choice is abstracted away.
Both editions support BYO embedding providers too — useful if you're standardizing on a specific dimension.
What about model determinism?
Agent runs are not deterministic — same input doesn't guarantee same output. We mitigate via:
- Strict tool-call contract (typed tools, validated outputs)
- Citation enforcement (post-generation pass)
- Confirmation gating on writes (you see what's proposed before it executes)
For workflows where determinism matters (e.g. compliance processing), use the REST API directly rather than the agent chat — the API is fully deterministic.
How much does an agent run cost (cloud)?
Depends on plan tier + run complexity:
- Starter: simple chat ~$0.01, dashboard card ~$0.03, complex research ~$0.10.
- The Scale tier has higher caps and access to better models (Sonnet, Gemini Pro) — costs scale roughly linearly with capability.
Per-tenant spend caps prevent surprises; you can set both monthly and daily caps in Settings → Tenant → Billing.
Self-host with bundled Ollama: $0. With BYO API key: whatever your provider charges.
Data
Where does my data live?
Cloud: AWS US-East-1 by default. EU residency available for Enterprise tier.
Self-host: wherever you install. The bundled Postgres holds all tenant data; files (uploaded content) live in the storage volume.
Can the agent see my data from another tenant?
No. Every query is RLS-enforced at the database level + every RPC takes an explicit tenant_id parameter. See Security model.
The single cross-tenant flow is the cross-org benchmarks feature (cloud only, opt-in), and it operates on aggregates with k-anonymity (k≥5) — never on raw rows.
What about training data?
We do NOT train models on your data. Cloud edition uses managed LLM providers (Anthropic, OpenAI, Google) under contracts that prohibit training on customer data. Self-host with BYO key — your provider's terms apply.
Can I export everything?
Yes. Cloud Enterprise has a guided export workflow (Settings →
Tenant → Privacy → Export). Lower tiers can request via support.
Self-host: ./ccd backup create gives you a complete tarball;
own it as you wish.
What's the retention policy?
| Data kind | Retention |
|---|---|
| Live entities (deals, contacts, posts) | Indefinite, until you delete |
| Agent run traces | 30/90/365/7y days by plan tier |
| Audit log | Same as run traces |
| Episodic memory | 90 days then promoted or archived |
| Tool call result caches | 24 hours |
Self-host: all of the above configurable.
Integrations
Which CRMs do you integrate with?
Native: none — CCD-Suite IS a CRM. We don't try to dual-write with HubSpot/Salesforce because the data model doesn't fit cleanly.
Read-only via MCP servers: hubspot-mcp, salesforce-mcp exist in the community and are simple to connect — useful if you're migrating.
What about email?
Read via OAuth (Gmail, Outlook). Send via... your existing email client. The agent drafts; you send. We deliberately don't try to be a sending tool — that's a different product category with different scale + deliverability constraints.
For automated sends (transactional, marketing), connect a sending provider via MCP (Resend, SendGrid all have MCP servers) and let the agent write to it under confirmation gating.
Can I use it from my CLI?
The REST API is fully usable from any HTTP client. We don't ship an official CLI today, but a thin shell over the REST API would be ~200 lines — happy to accept community contributions.
Pricing (cloud)
How much does it cost?
Community: $0 — self-hosted, 1 user, all 9 modules, bring your own LLM key (or run the bundled local model). Free licence key. Starter: $79/seat/mo — cloud, 5,000 agent runs/month included. Scale: $149/seat/mo — cloud, 10,000 agent runs/month included. Enterprise: custom — SSO/SAML, ABAC, deep audit, DSAR, EU residency.
See ccdsuite.com/pricing for the current published rates. The above may be stale.
Is there a free trial?
14 days of Scale features on cloud, no credit card. Pick a plan to continue, or self-host the free Community edition.
What counts toward agent runs?
Each time an agent generates a plan or response on your behalf is one agent run. Your cloud plan includes a monthly allotment (5,000 on Starter, 10,000 on Scale); Settings → Tenant → Billing → Usage shows the current month. Reading data and browsing the app never count — only the orchestrator's agent runs do. On self-host you bring your own LLM key, so there is no run cap.
Operational
How do I get support?
Cloud: in-app chat (response within 1 business day on Starter, 2 hours on Scale, 30 minutes on Enterprise) + email support@ccdsuite.com.
Self-host: community support by email and the docs knowledge base. Paid support contracts available for Enterprise self-host.
What about SLAs?
Cloud Scale + Enterprise have published SLAs (99.9% / 99.95%). Lower tiers are best-effort; recent uptime visible at status.ccdsuite.com.
Self-host: you operate it; your SLAs are your call.
How do I migrate from another tool?
The REST API accepts standard CSV imports + structured bulk endpoints. Most tenants migrate piecewise — CRM first, then content, then analytics layered on. We have a migration playbook for HubSpot → CCD; happy to write more if there's demand.
Can I get a demo?
Cloud: sign up for the 14-day trial. Self-host: curl https://get.ccdsuite.com/install.sh | sh. The bar for "demo"
is genuinely low — both paths work in under 30 minutes.
For high-touch demos (enterprise procurement), book via ccdsuite.com/demo.
Read next
- Overview — back to the docs hub.
- Self-host docs — operator guide.