Resident agents for installed bases
Thread puts a resident AI agent on every account you've already sold, to activate the add-on, finish the rollout, complete the onboarding. It texts, emails, and calls with a professional's patience, hands the close to your team when that's what works, and only calls it done when your system of record says so.
Illustrative cohort. The comeback band is the product: 1,139 accounts went dark and the guide pulled 704 back. Green is earned by the read; the agent can't mark its own homework.
How it works
You write the outcome in plain language and name the system that proves it. The guide owns the path.
Point a guide at a cohort, 50 accounts or 5,000. Consent attested before the first touch. Each account gets its own resident agent with its own memory.
It texts first. Answers questions from your knowledge base. Nudges. Emails. After a quiet week it calls, or books your closer with the full story. Weeks of patience, not a drip.
On every claim of done, it reads the record, Snowflake, HubSpot, Pendo, Slack. "The customer says it's connected" is a lead, not a close.
The last card fills only on a fresh read of your system. Closes you confirmed by hand are counted apart. Your CFO can audit the difference.
The platform
One system: resident guides on every account, the harness that runs them, and the surfaces your team and your customers see. Your CRM and warehouse stay the source of truth, Thread reads and writes; it doesn't replace.
Long-lived resident agents, one per account, each with its own memory and a single goal, get this account live, pursued across text, email, and voice for weeks. Every answer grounded in your knowledge base, your docs, your policies, your words. Voice included: instructions, guardrails, and a number that answers as your brand, on the same floor.
How the loop runs The harnessThe orchestration layer over the lifecycle: holds the goal, the context, and the exit criteria at every stage, and reads your systems to close outcomes.
Explore Motions The routerRouting and segmentation inside a Motion: reads each account's state and sends it down the right path, so the right guide engages at the right moment.
See the builderText, email, and voice are one thread with one memory. The owner replies wherever they actually answer, and the guide follows, nothing re-asked, nothing lost between channels.
Your team supervises from one workspace and steps in only where a human converts better.
A portal under your brand, magic-link, no passwords, where documents land, steps complete, and progress is visible. The guide walks people to it; the record proves they arrived.
The floor
These aren't settings someone remembers to turn on. They're the runtime, enforced in code on every guide, every channel, every wake.
Add your own rules on top. The floor never comes off.
Where guides work
Five installed-base problems, one motion. Find yours.
Thousands of merchants already pay for the module. The guide activates them one at a time, and proves every activation in your warehouse, not in a dashboard of sends.
Every acquisition hands you a book of merchants stalled between signed and transacting. The guide runs the chase per merchant, collect, verify, push, confirm, until the record is in your CRM.
POS migrations, program launches, compliance pushes, every location gets a guide carrying the program to done, with an audit trail your counsel can read.
Activation verified in product telemetry while CSMs judge instead of chase. For operating teams: one playbook, N portcos, metered per verified outcome.
Roughly half the addresses you built past never take the service. One guide per serviceable address, pre-registration to kept install to billing-active, verified against OSS and billing.
Voice, in the same room
Design the voice agent beside the guide: instructions, guardrails, knowledge, a number. Calls follow the same floor, every dial asks a human first, and the call never verifies; only the read does. Try it right now: +1 (669) 348-3154.
Start here