Frequently asked questions

Pick one high-volume workflow and launch it in a single channel. Prove that answers are accurate and escalations reach the right people, then expand to more channels and packs.

General-purpose copilots are tied to one vendor's model. Threada is a work platform: every request becomes a tracked WorkItem with policy controls, governed execution, and outcome analytics — across any LLM.

No. Threada works alongside your current tools. It captures and structures requests, then executes approved actions in the systems you already run.

AI handles understanding and structuring. Policy controls, review queues, and approvals handle the exceptions. People stay in the loop for anything high-risk.

Containment, time-to-resolution, escalation quality, approval cycle time, and action success — outcomes on the record, not interaction counts.

Run evaluation gates that test extraction, routing, grounding, and action behavior. Roll out gradually with canary traffic, shadow mode, and one-click rollback.

Teams that run high-volume operational work under accountability: customer operations and support, IT and procurement, security and compliance, and finance operations. If work arrives as requests that must be answered or actioned with a record, it can run on Threada as governed WorkItems.

The systems where work and actions live — support and ticketing tools like Zendesk, issue trackers like Jira, chat like Slack, plus email and API or webhook intake. Requests come in and approved actions go out as governed actions. See the integrations page for the current list.

No. A chatbot answers a message and forgets it. Threada manages the work: each request becomes a tracked WorkItem with cited evidence, policy controls, human approvals, governed actions, and an audit trail — so the work is governed and auditable end to end, not just a reply.

A WorkItem is the central unit of work in Threada — a typed, lifecycle-managed object that captures intake, decisioning, execution, and outcome for every request.

Traditional message-only automation tools manage requests one message at a time. Threada manages work: intake, policy, approvals, actions, and outcome measurement, with audit trails and governance controls.

Yes. Most teams begin with a single workflow scoped to one channel or workspace and expand incrementally as confidence builds.

Websites, uploaded documents, API connectors, and feed-based sources. Each is registered as a managed knowledge asset with its own freshness and retention policy.

You set sync cadence and stale thresholds per source. Threada alerts on freshness drift and failed syncs so evidence stays current.

Yes. Citation requirements can be applied per workflow and intent. Low-evidence responses are routed to review or escalation automatically.

Yes. Decision steps and approval requirements are configuration-driven.

Yes. Channel-specific policy overlays can override baseline workflow behavior.

Yes. Kill switches can disable execution per channel, workflow, or integration scope.

Yes. Telemetry dimensions include both channel and workflow identifiers for cross-sectional analysis.

Yes. Execution telemetry includes integration and action dimensions with success, retry, and failure states.

Yes. Policy version is available as a first-class reporting dimension for quality and outcome analysis.

Threada drafts security-questionnaire answers from your cited evidence and routes them for review. Low-risk, well-grounded answers can auto-approve by policy; sensitive ones wait for a named security owner. You never send an unreviewed answer unless your policy explicitly allows it.

Threada draws answers from sources you approve — previous questionnaires, security policies, SOC 2 reports, and DPAs — retrieved and cited per answer. If the evidence does not support an answer, the WorkItem flags it for a human instead of guessing.

Threada captures every drafted answer, edit, and approval as a time-stamped event with the evidence cited, so a later audit, renewal, or changed control can be reconstructed — who answered what, on what basis, and who approved it.

Threada grounds finance answers with retrieval over your sources, cites the specific page or record, and uses provenance labeling for high-stakes claims; when context is insufficient and abstain mode is on, it returns a no-answer fallback instead of guessing.

Threada's decision steps model human review, manager approval, security approval, automated gates, and exception queues, and policy overlays set approval requirements with deterministic precedence.

Yes — Threada's governed actions execute with idempotency keys and retries, are reversible, and produce auditable execution records, alongside a unified telemetry envelope and CSV/NDJSON exports.

Threada takes typed intake from web, in-app, Slack, Teams, email, API, and custom endpoints; inbound provider channels include Gmail, Twilio SMS, WhatsApp, social, Discord, and Teams webhooks, all normalized into WorkItems.

With abstain mode enabled, Threada uses a relevance threshold and returns an explicit no-answer fallback rather than fabricating an answer, and any answer it does produce includes citations.

Threada can gate support actions such as updating a ticket or notifying a customer behind decision steps; approvals are recorded, actions are reversible with idempotency keys, and execution is captured as auditable history.

A metered runtime event, such as intake, draft, approval, action, or WorkItem operation, counted by the platform usage ledger.

A processed page from uploaded or synced content that counts against the document-page quota used for indexing and evidence.

Yes. Anytime. Changes take effect next billing cycle.

You'll get warnings as you approach the limit; past it, usage is capped (overage credits apply where available). A separate 7-day grace period applies if a payment fails.

Save 20% with annual billing.

No. It's indexed for retrieval only.

We operate to internal availability targets with auto-scaling, automated backups, and tested restore playbooks. Contractual uptime targets and any service-credit remedies are agreed per order form.

No. Threada enforces policy within AI work automation — it integrates with your existing GRC and identity tools rather than replacing them.

Policy overlays are versioned and scoped by tenant, channel, pack, and workflow. Every execution records the policy version that was active.

Yes. Lifecycle, policy, and action events are exportable for external audits, SIEM integration, and regulatory reporting.

99.9% monthly availability with explicit RTO (4h) and RPO (24h) objectives and documented incident runbooks.

Stripe subscriptions + usage counters; grace-period warning automation and controlled disablement after grace.

Yes. Export feeds (GCS and BigQuery) and on-demand downloads are available for compliance reviews.

No. Threada is designed to orchestrate around existing systems through integrations and governed actions.

Yes. Approval requirements can be applied by workflow, channel, risk class, and action type.

Use EvalSuite gates for extraction, routing, grounding, and action behavior before broad rollout.

Yes. Web can be your first channel while you keep the same workflow primitives for future channels.

Yes. Integration execution is asynchronous and supports retries with auditable error classes.

Use correlation IDs, WorkItem history, decision events, and execution logs to reconstruct end-to-end flow.