Audience Builder turns plain-English briefs into structured queries against your warehouse — so analysts stop writing SQL for list pulls, and campaign teams stop waiting three days for them.
Audience Builder sits on top of your existing Redshift warehouse and your Civis Platform permissions. No new data store to maintain — the source of truth stays yours.
No SQL, no query-builder UI gymnastics. Just say who you want — Audience Builder handles the translation.
Each brief becomes a set of structured filters — list, boolean, or range — grounded in your real column names and descriptions.
Streaming preview, baseline statistics, and a clean hand-off to your ESP, ad platform, or Civis campaign tooling.
Describe the audience you want in plain English. Civis AI finds it — no SQL, no spreadsheets, no back-and-forth with your data team.
Just say what you're looking for — "Catholic donors in California who gave over $100 this year." Civis AI translates your request into precise targeting criteria so you always get exactly who you meant, not a rough approximation.
in ["Catholic"] · 1 of 7 categories selected
= "CA" · 1 of 50 categories selected
> $100 · max unconstrained
true · auto-added by suppression policy
Counts update in real time as you adjust your criteria. Change your mind halfway through? Just update your description — no need to wait for the previous search to finish.
Before you export, Civis AI automatically profiles your audience so you can see who you're reaching at a glance. No extra steps, no separate analysis.
Audiences bind to the dataset they were built against — so each team works in their own corner of the warehouse, with their own schema and their own permissions.
Fundraisers pull mid-donor and lapsed-donor lists in real time — instead of Slacking an analyst and waiting three days.
Campaign managers cut audiences for paid social, push, and email directly against the subscriber warehouse — with baseline stats attached.
Every audience carries its filter manifest, so compliance reviewers can inspect exactly which constituents were contacted, and why.
Care teams define outreach cohorts by clinical and demographic attributes — without exporting PHI to a third-party tool.
Build suppression lists, VIP cohorts, and lookalike seeds against the CRM you already trust. Export to any destination.
Campaigns, PACs, and party committees cut voter universes live — updated against the latest file, defensible on filter-by-filter basis.
Audience Builder runs as a Civis Platform Service inside your org. Warehouse queries use the viewer's Civis API key — so every row access is governed by the same RBAC you already trust.
No data leaves your environment to reach Civis AI. Only schema metadata and the natural-language prompt are sent to Amazon Bedrock.
Annually audited. Reports available under NDA.
Authorized for federal data workloads.
Only schema + prompt leave your VPC — never rows.
Queries run with the viewer's Civis API key.
As a Civis Platform Service, inside your Civis org. It queries your existing Redshift warehouse via the Civis API — there's no second data store to provision or keep in sync.
Your natural-language prompt, your dataset's schema (column names and remarks), and — for list filters — the distinct values for categorical columns. Row-level data never leaves your VPC to reach Bedrock.
Yes. Every audience exposes its full filter manifest. You can edit filters in a structured UI, keep chatting, or both. The natural-language interface is a starting point — not a lock-in.
Any Redshift table with a table comment and per-column remarks. Admins configure datasets per Civis org via the admin page; audiences are permanently bound to a dataset at creation.
Exports hand off to the destinations your Civis Platform org already uses — ESPs, ad platforms, SFTP, CSV. Every export carries its audience ID, so the filter manifest is traceable from campaign back to prompt.
Audience Builder is available to existing Civis Platform customers and to prospects as part of a guided pilot. Request a demo and we'll scope a short pilot against one of your datasets.
We'll run a working pilot against one of your Redshift tables. You'll see your first audience in the first session.
We'll run a working pilot against one of your datasets. Thirty minutes, your SQL person in the room — by the end you'll have a real audience.