Fair billing & AI trust
The two things merchants hate most about other AI chat tools — surprise bills and bots that hallucinate — fixed by design. You set the spending limit, decide what happens at the cap, and never pay for spam. And when the agent isn't confident, it says so instead of making things up.
- Spend controls: a monthly cap you set, with 50/80/100% alerts — we never auto-upgrade you — live
- Never-go-dark: at the cap the agent keeps capturing leads or hands off to a human, so you don't lose the sale — live
- Spam guard: abusive bursts are rate-limited and never billed against your quota — live
- AI guardrail: low-confidence answers disclaim or escalate instead of fabricating — live
- Knowledge Gaps report: every unsure answer grouped by topic, so quality never silently degrades — live
Deeper agent customization
More control over the agent's brain and behavior. Live now: choose GPT or Claude per agent (bring your own key), full system-prompt control, tone/guardrails, and a prompt inspector that shows the exact prompt your agent uses. Next: visual multi-step workflows and per-topic policies.
- Per-agent model choice + bring-your-own OpenAI/Anthropic key — live
- System prompt, tone & guardrails + prompt inspector — live
- Visual workflow builder for multi-step resolutions — coming
- Per-topic tone, rules, and escalation policies — coming
Business-system actions & integrations
Connect the agent to more of your stack — ERPs, OMS, WMS, CRMs, loyalty, and subscription tools — so it can take more real actions on the customer's behalf, securely and within your rules.
- Expanded action library beyond core ecommerce platforms
- Custom actions via API with granular permissions
- Deeper Klaviyo, Recharge, and CRM data sync
Richer human handoff
An even stronger help desk: smarter routing, agent collision detection, internal notes, SLAs, and seamless live takeover so customers never feel the seam between AI and human.
- Advanced routing and assignment rules
- SLA tracking and breach alerts
- Live co-pilot suggestions for human agents
Improved knowledge retrieval
Next-generation retrieval so the agent finds the right answer even in large, messy knowledge bases — with better handling of conflicting docs, freshness, and structured data.
- Higher-accuracy semantic search — hybrid scoring + diversity reranking finds the best answer, not just the closest match
- Automatic detection of stale or conflicting docs — prefers the most recently updated source
- Structured-data retrieval for specs, sizing, and policies — answers attribute questions precisely
Frequently Asked Questions
Want a say in the roadmap?
Join the ecommerce teams resolving more tickets, answering 24/7, and turning support into a revenue channel with Bookbag.