01
Read the request
On the issue view, Second Chance reads the summary, description, request type and labels, as the viewing agent, so it only ever sees Confluence articles they can already access.
Free and live on the Atlassian MarketplaceSelf-service doesn't end at submission.
The moment a request is submitted, Second Chance reviews it against your connected Confluence knowledge base and surfaces the articles that actually match, ranked by confidence, giving users one more chance to self-resolve before an agent picks it up.
Forge-native. Read-only. It never touches your ticket lifecycle.
Second Chance requires Confluence connected as the Jira Service Management knowledge base.
Why after submission
Most knowledge tools fire when nobody's reading.
Jira Service Management suggests articles while a customer is filling in the form, before they've committed. By then most people have already decided they want a ticket, so the suggestions get skipped.
Native JSM · before
Suggestions during the form
Shown while the user is still typing, before they've committed to raising a request. Most are already set on a ticket, so the prompt is dismissed.
Second Chance · after
Suggestions after submission
Analysed once the request exists and the user has a ticket reference: the moment they're most open to a faster route to resolution.
How it works
Read the request. Score the knowledge. Rank the answers.
Deterministic keyword scoring: no LLMs, no embeddings, nothing to train. Every recommendation is inspectable.
02
Score every candidate
It searches the Confluence space connected to the project as the customer-facing knowledge base, and scores each article on a transparent, weighted formula: request-type-in-title, keyword overlap, freshness.
03
Rank the top answers
The best three surface in the issue panel, each with a strength band and a plain-language reason. Requests with no strong match are logged as coverage gaps.
How the score is built
A weighted sum of signals. Deterministic, inspectable, and shown on every card.
Request type appears in the article title+30
Summary keyword overlap (×1.5 on a title hit)+25
Description keyword overlap+15
Label overlap · freshness+15 · ±3
Configured once, per project
The minimum match score and the weighting behind it live in Jira Service Management project settings, next to the rest of your JSM admin. Nothing to install or manage separately. Second Chance requires Confluence connected as the Jira Service Management knowledge base.
A genuine second chance to self-resolve
Matching articles surface right after the request is created, ranked by confidence, at the moment users are most likely to actually read them.
Catch known issues before an agent does
When a request matches an active incident or known problem, Second Chance flags it immediately with a status link, no waiting for triage.
See what your knowledge base actually covers
Analyse recent requests to find strong coverage, gaps, and recurring ticket types that deserve documentation. Turn deflection data into a roadmap.
Coverage intelligence
Every gap is a documentation opportunity.
Requests that arrive with no strong match are coverage gaps: the questions your Confluence knowledge base can't yet answer. Second Chance turns them into a ranked list of what to write next.
68%
Knowledge coverageCoveredStrong match
PartialWeak / medium
No matchCoverage gap
Lives in Jira admin, not a separate app to manage
Second Chance is a free, standalone app for Jira Service Management, enabled per project in JSM settings. There's nothing to configure to get started, and tuning the match score is optional.
Confidence, at a glance
Every card says how sure it is, and why.
Strong
≥ 35
A confident match, usually the request type or key terms hitting the article title directly.
Medium
≥ 22
A reasonable match on keyword or label overlap, worth a look before escalating.
Weak
> 0
Some signal, but thin. Shown for transparency, never gate-kept, always labelled honestly.
Runs inside Atlassian
Runtime
Forge-native
Runs entirely inside Atlassian Cloud. No external servers.
Permissions
Reads as the agent
Agents only ever see Confluence articles they already have permission to read.
Data
Nothing leaves Atlassian
No LLMs, no embeddings, no learned models. Scoring is rules-based and auditable.
Lifecycle
Never touches tickets
It never changes states, workflows or request lifecycles. Read-only, always.
Give every request a second chance.
Install Second Chance free from the Atlassian Marketplace. It's read-only, Forge-native, and live on the next issue view.