Skip to content

How Rivora works

One feature request becomes one complete solution, designed with the people who know and checked by your team. Every answer along the way can be kept as your product’s memory.

Illustrative example. The team, the policy and the feature in the pictures are made up.

Start with what you have

A feature request, plus whatever already exists: notes, call transcripts, screenshots, documents, Confluence pages and your code on GitHub. Rivora already knows what your team decided on earlier features.

  • Confluence: import the pages you choose.
  • GitHub: Rivora reads your code through a GitHub App with read-only access.
  • Privacy: before Rivora’s AI reads a file, personal details are stripped and secrets are blocked.

Illustrative example

It asks before it drafts

Rivora asks the questions it can’t answer, before and while it drafts. The people who know answer, correct and approve. Each answer can be kept as a decision, rule or clarification.

Illustrative example

Nine parts, chosen for each feature

Each feature’s solution has up to nine parts. Choose the ones a feature needs: the full feature, frontend and UX only, backend and data only, or your own mix. The Executive Summary and Feature Intent are always included.

Illustrative example

  1. Executive SummaryThe feature on one page: its value, how ready it is, and the main risks.
  2. Feature IntentThe problem, who has it, what success looks like, and what’s in and out of scope. It also holds the feature’s epic.
  3. Product ImpactThe screens, workflows, users and data the feature touches today.
  4. System ImpactWhich parts of the system change, with the risks your tech lead needs to know.
  5. Frontend and UXScreens, flows, states, permissions, accessibility and wording.
  6. Backend and PlatformAPIs, services, data changes, background jobs, permissions and what happens when something fails.
  7. Data and MemoryThe data the feature creates or changes, how long it’s kept, and what it links to.
  8. Implementation HandoffA step-by-step plan across frontend, backend, data, tests and documentation.
  9. Delivery BacklogOne epic and its stories, with acceptance criteria and a definition of done.

Only approved parts feed the parts that follow.

Your team checks every part

A person approves each part, sends it back with a comment, or edits it. An approved part changes only when someone chooses to change it. Each part shows the sources it used, so you can see what it was based on.

Illustrative example

Your product’s memory

When your team confirms an answer, Rivora keeps it as a decision, rule or constraint. Each one records where it applies, where it came from and who confirmed it.

  • It applies where it should: a product, one part of a product, or the whole workspace.
  • It changes when you say so: stop a decision or replace it when things change.
  • Nothing is learned silently: memory comes from what people say and confirm.
  • Every feature builds on the last: the next feature starts with everything your team has confirmed.

Decision · Active

Members on legacy NSW policies keep paper statements until June 2027.

Applies to
Member Portal
Confirmed by
Head of Claims Operations · 3 Oct 2026
Came from
Question on “Digital statements for members”
Used in
Backend and Platform · Delivery Backlog · Implementation Handoff

Illustrative example

Hand over and build

Approved work goes to Jira as stories ready to pick up, and to coding agents as a plan with the reasons attached.

  • JiraOne epic per feature. Stories go to Jira only after a person approves them.
  • ConfluencePublish approved work to Confluence, with a preview before anything is written.
  • Markdown and PRDExport approved parts as Markdown or as a PRD.
  • Coding agentsApproved work goes to coding agents such as Claude Code and Cursor, with the decisions and rules attached. Connect from Claude or Cursor with one click: Sign in with Rivora.
  • ZapierSend approved work to Zapier.

The agent handoff carries eight of the nine parts: all but the Executive Summary.

Illustrative example

Bring one real feature you’re planning in the next month.

Try it on one real featureBook a free 30-minute session