Real-time attribution · P95 closure under 60s
Multi-touch attribution closes in real time, all 7 models in parallel. The foundational architecture.
Publishing the roadmap publicly is a commitment, not a marketing flourish. We update this page quarterly. If something you depend on is in "Exploring," vote for it via /contact — we read every input and prioritise accordingly.
Multi-touch attribution closes in real time, all 7 models in parallel. The foundational architecture.
All 13 action handlers ship with snapshot, handler, and undoHandler exported. 7-day mandatory dry-run on customer-facing actions.
Google + Meta + TikTok unified plan execution with hourly budget reallocation based on attribution rollups.
GSC delta → 9 decay detectors in parallel → AI refresh draft → CMS editor queue. P95 detection-to-queue under 60 minutes.
OpenAPI v3.1 spec, HMAC-SHA256 outbound, exponential backoff, dead-letter queue after 4 retries.
Multi-tenant approval workflow for high-blast actions. Per-rule + per-shop scoping. Reviewer + audit-trail on every approve.
Routes between Anthropic Claude, OpenAI GPT-4, GPT-4o-mini based on task class. Per-shop monthly budget caps with 80%+100% alerts.
Scheduled enterprise-tier export of attributed events to customer's data warehouse. Encrypted in transit, KMS-managed.
Per-cohort LTV models. Weekly retrain. Available as a webhook event for downstream consumers.
Audit in progress with [Auditor]. Target completion 2026 Q3; report available under NDA to enterprise customers.
Begin ISO 27001 certification process. Target completion 2026 Q4.
Enterprise-tier deployment with all data + processing constrained to EU regions. Documented in DPA addendum.
All 13 action handlers parameterise on B2B tier. Reorder, message, discount, bid — tier-respecting by default.
Optional MMM layer on top of MTA. For customers requiring board-level allocation methodology. Likely 2027.
Auto-generated client libraries from OpenAPI. Currently exploring distribution model (npm + PyPI + Go modules).
Considering Iterable, Customer.io, Braze. Selection driven by customer requests; vote via /contact?track=roadmap.
Shop-specific brand-voice fine-tuning on a per-customer base model. Privacy + cost implications still being modelled.
Customer requests are the strongest input to the roadmap. If something here is in "Exploring" and matters to your deployment, tell us — that's how items move up. Equally, if we're missing something important, propose it.