Roadmap
What is next, and what we are deliberately not building
The last section is the useful one. Most roadmaps list only what is coming, which leaves you to work out whether a missing feature is late or refused, and those are very different answers when you are choosing a tool.
Being built
In the code right now.
Agent screen
Turns that outlive the browser
A sent message should run server-side to completion the way it does in a good chat product. Your machine can shut off mid-turn and the results still land in the workspace. Today a turn dies with the response stream, and a replayed running block becomes an error by design rather than a spinner that spins for ever.
Decided, not started
Decided and specified. Not started.
Runtime
The CEO steers the ship
Each workspace gets an operator-set goal, and a model-driven orchestrator decides each pass what needs doing toward it and directs the roster. Rather than five specialists and a coordinator each self-driving on their own cron. Phase one has shipped: the goal, and a daily steering pass that nudges at most two agents. Retiring the role crons is the remaining design.
Pricing
Pricing rebuilt on measured margin
Every metered action repriced against its measured vendor cost rather than an estimate, with allowances resized to match. This was blocked on the cost measurement being wrong; that is fixed, so the repricing is next. We will publish the new table before it takes effect.
Docs
Full-body documentation search
Search over page bodies rather than titles, sized for the person hunting an error code or a field name. The palette already does not know where results come from, so the swap is a ranking function plus an index.
Connectors
OAuth refresh for the three short-lived connectors
Salesforce, Zoho and Google Calendar currently hold credentials that expire on the vendor's schedule, so a card can degrade without you doing anything. Connect once, stay connected.
Designed and waiting
Written up in full, waiting on something ahead of it.
Inbox
The enhanced inbox
Four changes: the poller stops discarding inbound from people who are not contacts, every thread gets a profile rail beside the messages, a one-click draft runs the copy agent's real prompt stack, and every draft carries two exits. Send, or start a sequence. Specified in full; not built.
Channels
SMS
The named next channel, and the only one. It is not open yet and we would rather say that than list it as available.
Runtime
Per-workspace runners
Each workspace's orchestrator in its own container behind an invoke, status and cancel seam, so one workspace's run cannot starve or observe another's. Designed and worth doing once the model-driven orchestrator is the thing running.
Declined, on purpose
Asked for, considered, declined, and why.
Channels
Instagram, phone and paid ads
Instagram left the motion entirely. The platform requires the correspondent to start the conversation, so there was no compliant cold path to lose. Phone and paid ads are closed. Saying so is more useful than leaving them on a roadmap nobody intends to reach.
Sourcing
Sales Navigator integration
It would return structured fields and remove most of the second AI pass, and it needs a paid seat. Nothing in the product may assume a customer holds one, so we built against Classic's four searches instead.
Compliance
An EU deployment
We serve the US and say so. The controller-grade data model is built because it is cheap to keep and a rewrite to retrofit, but an EU region is a migration and a legal position rather than a setting. It reopens when there is an EU customer to open it for.
No dates, and that is not evasion
A dated roadmap from a team this size is a guess with a calendar attached, and the honest version of it changes every fortnight. What is stable is the order, and the order is above.
If something on this page is the one thing standing between you and using GTM, say so. It is the strongest signal we get, and it does move the order.
Everything above is next. Everything else already runs.
The roadmap is short because the motion is built. What is left is depth, isolation and pricing accuracy.
Start with sales. $2,000 or $5,000 onboarding, then at least $250 monthly credit funding.