SShortSingh.
Back to feed

Spec for Handing Website Chatbot Conversations Off to Human Staff

0
·10 views

Many website chatbot projects treat human handoff as an afterthought, even though it is often where leads are won or lost. The recommended approach is to define handoff triggers before coding, covering explicit requests for a person, low retrieval confidence, restricted topics, buying signals, and conversation loops, each tagged with a named reason. Handoffs should collect only essentials such as preferred contact method, consent, and a short problem summary, with consent stored as a timestamped record. The bot should also adjust its messaging to business hours, avoiding promises of quick replies when the team is offline. A versioned JSON contract between the bot and the team's tools, including a transcript link rather than the full transcript, keeps data organized and controlled.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Log in to join the discussion and vote.

Log in

Related stories

0
ProgrammingDEV Community ·

Hoki: Open-source waving lucky cat app adds Windows and Android versions

Hoki, a virtual Chinese lucky cat (招财猫), has released version 3.2.0 with a Windows app featuring a live 3D desktop cat that follows the mouse. An offline Android app and home-screen widget are also available, along with an installable PWA. Tapping the cat gives a luck boost, and lucky numbers trigger a bell and blessing alongside a daily fortune. The Windows app is single-instance, so relaunching hands focus to the running app, and all data stays on the user's device with no account, ads, or tracking. The project is MIT licensed, written mostly in JavaScript, and hosted for free on GitHub Pages.

0
ProgrammingDEV Community ·

Docker network clash on integration host cut off carrier traffic after new project deployed

An integration host running partner adapters in Docker Compose lost connectivity to a carrier reached over a VPN after a new project was deployed. Docker automatically assigned the new project a subnet, 172.25.0.0/16, that overlapped with the carrier's address range, which the host's default gateway routed through the tunnel. The new local bridge route redirected the carrier's traffic into an empty network, so bookings timed out until the project was stopped. The team fixed the issue by setting Docker's default address pools to a reserved block, recreating existing projects, and adding a pre-start check that compares bridge subnets with partner ranges.

0
ProgrammingDEV Community ·

Why Identical Firewall Rules Behave Differently Across AWS, Azure, Google Cloud and Oracle

Copying a firewall rule between cloud providers can unintentionally widen access, even when the port and intent appear identical. AWS security groups are stateful and allow-only, while Azure relies on priority-ordered rules with explicit denies and built-in defaults. Google Cloud applies an implied egress allow at the lowest priority, which can undermine a deny-everything policy if egress is not handled explicitly. Oracle security lists are stateless by default, so a rule copied from a stateful platform can break return traffic and tempt teams into overly broad fixes. The article warns that translating security group references into broad CIDR ranges and differing logging behaviours can silently expand exposure and hide evidence until an audit or incident.