Every SylvieSI feature
SylvieSI is made by THE SHRIKS. This page lists what it does, part by part, in plain words.
Read and sort
Three message sources
Reads messages written in your app, in Slack channels you choose and in a Gmail inbox you choose.
Labels
Marks each message as a bug, a feature request, a question or feedback, and shows how sure it is.
Duplicate groups
Groups messages about the same bug, keeps every original and lets a person split a group.
Clarifying questions
Drafts the question to ask when a step or a place is missing. Nothing is sent without a send confirmation.
Sender trust
Marks a sender as signed in-app, verified by Slack or Gmail, or unverified. An unverified sender is never linked to an account automatically.
Inbox filters
Filter by bugs, feature requests, questions, feedback or needs clarification. Mark messages reviewed or dismiss them with a reason.
Decide what to build
Ideas
Gathers repeated requests into one Idea with the arguments for and against, who asked and how many.
Six decisions
Already solved, ask for details, reject, defer, build a component for staging, or hand to engineering. A decision needs a source.
Insights
Shows patterns across messages: an emerging bug, a growing request, a conflict between requests and a gap in the data.
Overview
Shows what changed since you last looked, with a source on every line.
Users list
Shows each sender's messages, bugs and requests on one page, with tabs for messages, usage, cases, outcomes and a timeline.
Find the cause
Cases
Opens a Case for a confirmed bug and links every matching message to it.
Reproduction
Runs the steps the sender described and watches what happens.
Evidence
Reads code and error data, and saves each finding with its source, its age and the gaps.
Root cause assessment
States a hypothesis, the evidence for and against, what is missing and the next step, with one of four states.
Product Graph
Holds what SylvieSI knows about your product: pages, components, APIs, roles and processes, each with its source and age. A person can correct a fact.
Resolution Graph
Links a bug, its cause, the work, the change and the outcome, so a new investigation can start from past Cases with the same cause.
Prepare and approve
Work
Lists the steps, owners, attempts and blockers needed to resolve a Case.
Linear handoff
Sends an engineering issue with the reproduction, the cause, a test plan and the exact change. The issue's status syncs back.
Draft pull requests
Creates a branch and a draft pull request with a fix and a test on a repository you approve. A person merges.
Approved components
Prepares a versioned component from a fixed catalog for a simple read-only feature, tests it in a browser and delivers it to staging.
LiveLayer
Delivers approved components to your staging site as a signed bundle that loads only for the audience you choose.
Approvals
A person approves one exact change with its risk and how to undo it. An edited change needs a new approval.
Switch off
A delivered change can be disabled, and SylvieSI checks in a real browser that the original page is back.
Check and tell
Staging check
Re-runs the failing step on staging and confirms that the fixed build is the one deployed.
Outcomes
Tracks six states: awaiting, measuring, verified, not achieved, inconclusive and regressed.
Replies
Drafts an update for each person who reported the bug. It goes out on the channel they used, after the check.
Re-open
Proposes to re-open a Case when a matching message arrives after it closed. The earlier outcome stays.
The Sylvie workspace
Four actions
Ask, Investigate, Execute and Schedule, chosen when you write.
Chats
Threads are grouped as Today, Yesterday, the previous 7 days, the previous 30 days and Pinned. Rename, pin or delete a thread.
Attachments
Attach files to a chat. SylvieSI checks the type and size, and removes personal data before a model reads the file.
Objectives
A goal with a plan, evidence, linked chats and an outcome, with its blockers and the Cases it touches.
On Going panel
Shows running work, approvals waiting for you and upcoming schedules beside your chat.
Assist
A panel on every Customer Intelligence and Product Operations page that answers questions about the open item and suggests prompts.
Command K search
Searches messages, customers, cases, ideas, work, changes, approvals, objectives and chats. Results follow your role.
Notifications
Choose decisions and failures, everything, or nothing. Notices arrive in the app and by email, and optionally in Slack.
Schedules
Run a task on a schedule, such as a weekly review of new bug reports. Each run lists its result or its failure.
Connectors
Connect Slack, Gmail, Linear, GitHub, Sentry and your model provider. Each shows its state, permissions and last health check.
Control and settings
Seven roles
Owner, Admin, Product, Engineering, Support, Reviewer and Viewer decide who can triage, approve and change settings.
Ten agents
Orchestrator, Assist, Triage, Evidence, Idea, Investigation, Runtime, Engineering, Communication and Brief. Turn each on or off.
Models and limits
Choose the model behind each role and set spending limits.
Audit history
Lists every action by a person or an agent.
Twelve settings pages
Workspace, Team, Product and Environments, Product Graph, Runtime, Integrations, Agents, Automation and Policies, Permissions, Data and Privacy, Notifications and Audit.
Setup in five steps
Account, workspace, message sources, product details and permissions.
Data controls
Set how long evidence is kept, freeze a workspace, export its data or request deletion.