Built for the inbox that never stops
For small nonprofits, the same questions arrive week after week: programs, registration, eligibility. Someone has to answer them. For larger organizations, it's about keeping responses consistent as programs and staff change. We know both situations from the inside. This is the managed email system built for both.
Consistent responses, without the overhead
The system classifies incoming emails and matches them to pre-written responses that your team controls. It doesn’t generate responses from scratch. It selects the right canned reply based on what the email is asking. Your tone, your accuracy, your voice.
You start in draft-only mode. Nothing sends automatically until you’ve reviewed enough drafts to feel confident in how the system performs. Confidence thresholds are adjustable. You decide when and whether to enable auto-sending.
The system selects from templates you write and approve. AI handles classification; humans handle authorship. The voice, tone, and accuracy of every response stays in your hands.
When a staff member leaves, the templates stay. Onboarding the next person is a training exercise, not a knowledge-recovery project. The institutional knowledge is in the system, not in someone’s head.
What happens when an email arrives
A series of checks runs before any AI is involved. Replies go to humans immediately. New inquiries get classified, matched, and handled, with you in control of every threshold.
Reply detection
If the message is a reply to a previous email, it is flagged for human review immediately. No AI processing.
Classification
New inquiries are classified into one of your defined categories based on what the email is actually asking.
Draft or send
A matched canned response is saved as a draft for your review or sent automatically, depending on your comfort level.
Notifications
You receive a notification for every action the system takes: drafts awaiting review, escalations, and auto-sends for your records.
Hosted or dedicated: how much volume, and where does your data need to live?
Core and Active are differentiated by email volume and number of inboxes. Start there. The dedicated options exist for organizations where the question of data location isn't optional, and where the budget to answer it properly is in place.
For small organizations where routine email is eating time your team doesn't have. Handles inquiries automatically, flags anything complex, and keeps running whether you're at full capacity or stretched thin.
- 1 inbox (Gmail or mailbox.org)
- Up to 200 emails/month
- Up to 10 canned response templates
- Self-serve template editor with test mode
- Draft-only mode to start; auto-send optional
- Reply detection and human flagging
- Notifications via email or Telegram
- Anthropic API (your own key)
Email content processed by Anthropic on US-based infrastructure, disclosed in service agreement.
Book a free callFor organizations running seasonal programs or multiple communication channels. Active context management means you can update the system with current program details (a new intake period, a registration deadline) so responses stay accurate without reclassifying everything manually.
- Everything in Core
- Up to 3 inboxes
- Up to 600 emails/month combined
- Up to 20 canned response templates
- Active context management for seasonal programs and events
Email content processed by Anthropic on US-based infrastructure, disclosed in service agreement.
Book a free callA dedicated private server running a local AI model. No email content leaves Canada or touches any third-party API. Hosted by a Canadian-owned provider, in Canadian data centres. For organizations where the answer to 'where does this data go?' is not optional.
- Dedicated private server. Your data is never co-mingled.
- Hosted by a Canadian-owned provider, in Canadian data centres
- Local Ollama model. Email content never sent to any external API.
- Up to 3 inboxes
- Up to 600 emails/month
- Up to 20 canned response templates
- Active context management for seasonal programs and events
- Priority support
Ollama runs on hardware physically located in your office, maintained remotely by Cascadia South. Data does not leave your building. For organizations with the most stringent data requirements.
- Ollama and n8n installed on client-owned hardware
- Hardware in your office, managed remotely
- No data leaves your premises
- Setup, configuration, and ongoing remote maintenance by Cascadia South
- Scope and pricing determined after a discovery conversation
Hardware is client-supplied and not included in service pricing. Depending on your email volume, expect $400–2,500 CAD in hardware costs. See us for guidance on what your situation requires.
Book a free callStart with a conversation
We'll walk through your inbox, your situation, and which option makes sense. No commitment required.
What you get from day one
Your one-time onboarding fee covers initial setup: configuration, inbox connection, template authoring, and supervised testing before anything goes live. Your monthly subscription covers everything below.
Initial setup
Covered by your one-time onboarding fee. Workflow configuration, inbox connection, template authoring session, and a supervised testing period. Nothing sends automatically until you've reviewed how the system handles your actual incoming mail.
Template editor
A self-serve interface to add, edit, and test your canned responses. Update seasonal context (a new program starting, a registration closing) without contacting us. Includes a test mode to preview classification before enabling.
Reply and escalation handling
All replies to outgoing messages are detected automatically and routed to human review. Low-confidence classifications are flagged as drafts, never auto-sent.
Notifications
Every action generates a notification. Drafts awaiting review, messages flagged for follow-up, and auto-sends for your records. Delivered via email or Telegram, your choice.
Built for the inbox problems nonprofits actually have
- Small nonprofits getting more email than they can answer consistently: programs, registrations, eligibility questions that pile up because everyone's already stretched
- Organizations where staff turnover has disrupted inbox management and the institutional knowledge walked out with whoever left
- Any organization running seasonal or time-limited programs that generate bursts of similar questions
- Larger organizations looking to maintain consistent, professional responses across programs and channels without dedicating staff hours to routine correspondence
This is not a replacement for human communication. Anything nuanced, complex, or requiring follow-up is flagged for your team. The system handles the routine so your people can focus on the work that actually needs them.
We believe organisations should make informed choices about where their data goes. That’s why we offer options with different data handling characteristics and explain the trade-offs plainly rather than burying them in fine print.
For organisations where data sensitivity is heightened, or where funder or regulatory requirements specify how data must be handled, we recommend starting with a conversation rather than a tier selection.
This service is available on its own or as an add-on to an existing Cascadia South digital infrastructure retainer.
The AI doesn't write your responses. It just decides which one to send.
The well-known problem with AI is that it can make things up, producing confident, plausible-sounding responses that are simply wrong. That happens when AI is asked to write: to come up with an answer from scratch, explain something it may not actually know, or generate content on the fly.
This system doesn’t write anything. When an email arrives, the AI reads it and decides which of your predefined response categories it belongs to, essentially sorting it into the right folder. The response that gets sent is always a message your team wrote, reviewed, and approved in advance. The AI’s job is to pick the right one, not to come up with something new.
When the system isn’t sure which category fits, it doesn’t guess; it saves the email as a draft and notifies your team to make the call. You control how conservative that threshold is. The worst-case outcome is a draft sitting in your queue, not the wrong message going out to a client.
Consistent responses shouldn't depend on who's in the office.
Book a free call. We'll walk through your inbox, your templates, and what option makes sense for your situation. No commitment required.