Product / Smart automation

Automation that cites its sources.

Renewal drafts written from each member's actual history. Lapse detection 30 days out. Engagement scores update nightly. Citations on every output. Toggle off and the platform still works.

Get started See how drafts work
Capabilities

What the assistant can do.

Five concrete capabilities. Each one is grounded in your data. Each one is auditable.

Draft renewal outreach

"Write a renewal email for Sarah Chen who lapsed 14 days ago." Cites the member's 3 prior interactions, references their actual history.

Flag at-risk members

Lapse detection 30 days before renewal date. Engagement scores update nightly. Staff-facing only; never visible to members.

Answer database questions

"Who hasn't paid dues this quarter?" Results come with the underlying query shown so you can verify.

Clean CSV imports

Catches duplicates, format mismatches, and obviously bad data before it lands. You approve before commit.

Summarize threads

A long email thread or member note history gets a one-paragraph summary with linked source messages.

Citations, always

Every output includes a "Drafted from X" footer pointing to the exact database fields, events, or messages used.

The honest pitch

Most automation in incumbent platforms is a "generate" button.

It produces generic content with no grounding. Members notice. Staff lose trust. The feature gets unused after the third draft mentions a fundraiser that never happened.

Sembr's automation is grounded in your data, cites its sources, and respects your choice to opt out completely. Every output is a draft, not an action. Drafts always require staff review before they send, post, or trigger.

  • Output cites the source fields it used. You can always verify.
  • No hallucinated member names, history, or relationships. If insufficient data, it says so.
  • Drafts are drafts, not actions. Staff approves before send.
  • Audit log records the draft, the staff edit, and the final sent version separately.

Bring your own key. Or toggle off entirely.

Sembr defaults to Anthropic Claude (claude-sonnet-4-6). If you'd rather use OpenAI or Google Gemini, paste your API key in settings; your data flows directly to the provider you chose. Or toggle automation off at the tenant level and the platform still works (every workflow that uses automation has a manual path).

Get started See the DPA
# Automation toggle (per tenant)
> default: ON (Anthropic Claude)
> bring_your_own_key: optional
> providers: Anthropic, OpenAI, Gemini
> tenant.disable_ai: false
> member.exclude_from_ai: per-member opt-out
FAQ

Common questions.

What model does Sembr use by default?
Anthropic Claude (claude-sonnet-4-6). Selected because it follows grounding instructions reliably and refuses to invent data when context is insufficient. You can swap to OpenAI or Google Gemini by pasting your API key in settings; your data flows directly to the provider you chose.
Where does the data live during a draft?
The database query and the resulting context (the member's history, prior interactions, and field values) are sent to the AI provider over TLS. The provider returns a draft. Sembr does not retain the prompt or response after the draft renders. The provider's own retention policy applies (Anthropic: zero-retention for API by default).
Can a member opt out of having their data used in automation?
Yes. Each member can toggle "exclude from AI features" in their portal settings. When set, that member's data is filtered out of any AI prompt, and AI features that need data about them (drafted outreach, engagement scoring) skip over them silently.
What if automation gets it wrong?
Every AI output is a draft, not an action. Drafts always require staff review before they send, post, or trigger. The audit log records the draft, the staff edit, and the final sent version separately so you can audit divergence over time.
Does this count as "AI in production" for board / governance purposes?
The default config for AI features is "off" at the tenant level. Your board can review the DPA and the AI provider's data handling docs before enabling. You can also restrict AI features to specific staff roles via the permission matrix.

Smart automation, on your terms.

Toggle it off and the platform still works.

Get started See pricing