roadmap-2026
Every quarter we publish an open roadmap. No "coming soon", no marketing mystery. Approximate dates, clear scope, and — most importantly — a list of things we decided not to do, because good prioritization is saying "no" with courage.
Principles guiding the roadmap
Three fixed principles; every feature is tested against them:
- It works for content creators who live off their work. That's our ICP. We don't optimize for enterprise running 50 different agents.
- It preserves BYOK and creator control. No new feature can require lock-in, proprietary data, or opaque token resale. (Details at why BYOK matters.)
- It reduces operational work, not adds. If a feature requires the creator to "configure more stuff", it fails the test — even if it's pretty to demo.
Q2 2026 — under construction
Everything here is in active development, with release planned for June/July.
- Voice on live calls. The agent picks up calls direct on the hub's number (via Twilio), responds in voice with ~1s latency, and logs the transcript to history. Closed beta in May.
- Automatic invoicing (Brazil). Native integration with NFE.io and eNotas. Every Stripe Connect sale triggers automatic NF issuance, no human in the loop. Heavily requested. Shipping in May.
- Multi-language in the same conversation. The member can switch between Portuguese/English/Spanish in a single thread — the agent detects and replies in the language of the turn, keeping unified memory.
- Native export to Notion/Airtable. Two-way sync of persisted facts — useful for those running a CRM in Notion.
Q3 2026 — planned
Spec locked, build starts in July.
- Collaborative hubs. Multiple admins on the same hub with granular permissions. Main use case: creators with a 3-6 person team running the same agent.
- Public component library. Prompts, triggers and integrations shareable across hubs — with attribution. Like a GitHub for agent configuration.
- Native A/B experiments. Automatic 50/50 split between two prompts or two offers, measuring revenue conversion per variant.
- Async weekly report. Automatic email every Monday with the 3-5 insights that matter (conversion, retention, highlight conversations).
Q4 2026 — exploring
Hard to promise here. Still in research, customer interviews, proofs of concept.
- Proactive agent (outbound). The agent starts conversations with members under specific conditions (e.g. 30 days without login, plan anniversary, external event). Challenge: not turning into spam.
- Shared memory across hubs (opt-in). If you run 3 different hubs for the same audience, the option to cross-reference facts — still studying privacy implications.
- Voice fine-tuning on the creator's content. Not the model itself, but a lightweight "style embedding" that makes the agent sound even more like you in tone.
What we decided NOT to do
Short list, but important. Requests for these come every month — the answer stays consistent.
- Third-party agent marketplace. Turns into a race-to-the-bottom on price and quality. Not our game.
- An in-house AI model trained by us. Training models is expensive, slow, and would scatter focus. We'd rather be the best orchestrator of frontier models.
- Token resale with markup. We already explained here why. BYOK stays.
- Native mobile app for members. Webview inside our own app is enough — building and maintaining native iOS/Android doubles effort with no real gain.
- Visual flow editor n8n-style. Drag-and-drop flows encourage unnecessary complexity. We prefer smart agent + declarative configuration.
How the roadmap adjusts
Every request that comes in via a public channel (Discord, support, Twitter) becomes an entry on a board visible to everyone. Every Monday morning, the product team triages what came in that week. Every 6 weeks, the roadmap is re-prioritized based on:
- Request frequency (how many different customers asked);
- Estimated revenue impact for customers (not always the most-voted one);
- Coherence with the three principles.
If your request didn't show up here, it's not because it was ignored. It's because it hasn't earned priority yet — or it landed on the "not doing" list and we decided to tell you to your face.
Got an idea for the roadmap?
Send it via in-app support. Every request gets read — and most become an entry on the public board.
Open app
A roadmap that doesn't say "no" isn't a roadmap. It's a wishlist. We'd rather disappoint 50 people now than ship a half-baked version of everything six months from now.
Read more at memberai.pro/en/blog/roadmap-2026.
Learn more: plans and pricing · about Member AI · real customer cases · full blog.