Start with evidence, then carry it into delivery

• Turn a website into a reviewed Product Brief, confirm up to three starter Capabilities and keep existing Product Context or workspace data intact.
• Move through Nova feedback, GitHub setup and the workspace tour as complementary steps. GitHub can be deferred, valid imports are reused and progress survives reload.
• Nova can prepare eight Demo customer messages from the accepted brief. They enter the ordinary feedback pipeline, where summaries, evidence and reviewable Problem proposals appear with visible processing progress.
• Remove the Demo set together or delete individual Demo Feedback without removing real feedback, Company Context or unrelated relationships. Reset can prepare fresh examples.

The guided flows were exercised in DEV; production rollout is not verified.

• Problem verification considers attributable Initiative Decisions alongside delivered Changes. A deliberate, documented decision can resolve evidence, while deferral or exploration does not replace implementation evidence.
• Saved verdicts become stale when their linked delivery, Decisions, evidence or Initiatives change. Initiative details can audit open Problems in a batch while the user keeps the final closure decision.
• Closure previews show the total Initiatives, Problems, Feedback items and Evidence chunks affected. Partial resolutions call out the remaining scope against linked Changes and pull requests.
• Feedback analysis keeps focused customer evidence and separates background context from the primary evidence that must be resolved.

• Proposed Changes appear in the Delivery Timeline with implementation context, desired outcome and success criterion. Create work item is the primary action, and one proposal can connect several Work Items and implementations.
• Prepared briefs accept up to 50,000 characters and keep oversized drafts editable with a validation message. Proposal exports remain hidden until each provider has passed live verification.
• Pull request cards distinguish Open, Draft, Merged and Closed. Older webhook deliveries and import batches cannot overwrite a newer stored GitHub state.
• Discovery Decisions use a focused queue beside the Decision Brief, with search, counts and sorting. Initiative details remain alongside the brief on wide screens and move into a drawer on smaller screens.

• Authorized MCP connections can create and update Capabilities, create Problems from evidence, prepare draft Campaigns and link existing Feedback, Evidence, Problems, Capabilities, Initiatives and Changes within their documented permission and feature-switch limits.
• Initiative goal suggestions are available to connected clients for inspection and confirmation, including custom outcomes and idempotent receipts. Initiative links use the canonical Initiative workspace URL.
• Work Items follow confirmed GitHub Issue status. Connected tools distinguish direct product links from inherited relationships, and Feedback updates or evidence-backed Change claims do not silently mark a Problem as resolved.
• Capability creation, Problem creation, Campaign drafts and other guarded writes retain their existing default-off or permission boundaries. Older connections need to approve added scopes.

• Campaign member details support member and touchpoint-history imports, inline touchpoint logging and editing, next-action scheduling and rich-text selection rationale.
• Linked Contacts and Companies open inside the campaign detail stack. Campaign pipelines can be filtered by assignee, while cards, tables and touchpoint history show ownership.
• Contacts can be filtered by workspace LinkedIn network labels. Connected MCP clients can save network provenance and tags, match normalized LinkedIn profile URLs and return ambiguous identities for review.

• Feedback, Problems, Initiatives, Product, Strategy, Competitors, Sources, Contacts, Campaigns, Stories and Metrics share more consistent selection, spacing, reading surfaces and narrow-screen behavior.
• Keyboard editing separates titles from menus, preserves failed drafts, blocks duplicate saves and returns focus after a successful save. Long dialogs and detail headers keep actions reachable on smaller screens.
• Bridge highlights active competitor risks and low Product Quality scores, while Crew shows real agent history without placeholder deployment actions or fabricated workload totals.
• Story source search, workspace tabs, comments and read-only titles received focused navigation and reliability fixes.

Teams can start with grounded context, inspect the evidence behind a decision and carry the chosen work into delivery with fewer detached records. The same relationships are also becoming usable through connected clients, with explicit permissions and review boundaries still in place.