MazylonPulse
Product

The proactive delivery copilot.

Mazylon Pulse isn't a dashboard. It's a copilot that watches your projects continuously, coaches your teams at the right moment, and makes your Confluence processes (deployment, DoD, onboarding, incident…) answer in Teams.

Daily Digest

Every morning at 8am in your team's Teams channel: an actionable, prioritized summary with traceable evidence.

  • Critical tickets + severity
  • Active blockers + elapsed time
  • Concrete recommended actions
  • 👍 / 👎 buttons to calibrate the agent
  • Adaptive card format optimized for Teams

Automatic risk detection

~40 embedded business heuristics, written by agile coaches, tenant-calibratable via continuous feedback. Story structure, backlog health, cycle time, abandoned PRs, agile metrics and positive signals measured.

  • Stagnating tickets with sprint-aware thresholds (2d end of sprint, 7d start)
  • Overloaded / slow-start / day-3 slipping sprints (< 60% delivery)
  • PRs with no review, oversized PRs, saturated reviewers, PRs abandoned > 30d
  • Bugs older than 30d, bugs/features ratio > 30% of active tickets
  • Backlog too big (>100 TODO) or nearly empty, WIP > 20 simultaneous tickets
  • Falling velocity (3-sprint avg < 60%), high carry-over, throughput halved
  • Positive signals measured: sprint on-track at day -3, 0 critical bug, 0 blocked ticket

Cross-project exec view for EM (new)

If you steer 3 to 8 projects, atomic signals aren't enough. The executive layer detects meta patterns per-project heuristics miss — portfolio drift, someone overloaded cross-projects, multi-team blocker, sprint miss streak. Delivered as Monday 8am digest + 4h cron push + dedicated dashboard + Teams intent.

  • 7 cross-project exec signals: PROJECT_DRIFT, CRITICAL_BURST, ASSIGNEE_OVERLOAD_CROSS, HEALTH_DEGRADATION, MULTI_PROJECT_BLOCKER, SPRINT_MISS_STREAK, OVERCOMMIT_AVOIDED
  • Weekly EM digest Monday 8am with median delivery + top critical signals + per project
  • Proactive push every 4h as Teams DM (native 24h dedup, never spam)
  • /em-overview dashboard: 4 KPI tiles + grid cards per project + projects × weeks heatmap
  • Teams intent « exec view » / « where does my portfolio stand? » for the on-demand view
  • « EM Recommended » preset includes EM_EXECUTIVE_SIGNAL — one-click activation
  • DB archive of EM digests to reconstitute your 12-week portfolio trend

Portfolio KPIs — cross-project delivery view (new)

One step more analytical than the exec view: median lead time, cumulative throughput, total WIP, median escape rate, predictability — all aggregated across your active projects, with « worst offenders » leaderboards that answer in 3 seconds « which of my projects is slowest, which has the most escaping bugs, where are we overloaded? ». Available in dashboard and Teams bot.

  • Consolidated totals: cumulative throughput, total WIP across all active devs, medians for lead time / escape / completion ratio
  • 5 worst-offenders leaderboards (top 3 per category): lead time, predictability < 70%, escape rate > 10%, bugs ratio ≥ 40% of shipped, overloaded WIP (≥ 5 tickets/dev)
  • Detailed per-project table sorted by load (escape × lead time), columns lead time / throughput / WIP / pred / escape / bugs %
  • Dedicated dashboard page /portfolio-kpis + direct link to each project's Flow tab
  • Teams bot intent « portfolio KPIs » / « which projects are slowest » / « where do we ship worst »
  • Distinct from the exec view (instant signals): here we're talking aggregated flow/quality metrics to steer delivery

Per-project flow/quality KPIs (new)

On each project, the Flow tab gives the metrics modern EMs actually use — lead time, throughput trend, average WIP, PR/backlog age distributions, effort mix, 5-sprint predictability, escape rate. No DORA connector required: everything is computed from your Jira/GH/GL/BB data already ingested.

  • Lead time (createdAt → DONE) median + P90 over rolling 90d
  • Throughput this week + 4-week trend sparkline with 📈/📉 indicator
  • Average/max WIP per person + PR age distribution (0-2d / 2-5d / 5-14d / 14d+) and TODO backlog
  • Effort mix by type over 30d — auto-alert if bugs ≥ 40% of shipped (tech debt)
  • Predictability — completion ratio of last 5 closed sprints + median, colored by severity
  • Escape rate — post-sprint bugs referenced via WorkUnitLink, with thresholds 🟢 ≤5% / 🟡 5-15% / 🔴 >15%
  • Dedicated Flow tab in the dashboard + markdown rendering in the Teams bot (« Alpha KPIs »)

Weekly auto-refresh of overview + coach (new)

Before, the coach analysis and Discovery status went stale for weeks unless someone clicked « Refresh ». Now, every Sunday 8pm local of your organization, Pulse re-syncs all connectors of your active projects and regenerates the agile coach report — your analyses are fresh Monday morning, hands-off.

  • Step 1 Sunday 8pm: re-sync Jira + GitHub/GitLab/Bitbucket + regenerated DISCOVERY report
  • Step 2: agile coach (SAFe/Scrum recommendations) regenerated on the fresh baseline
  • Guards: DORMANT projects skipped, artefact < 6d skipped (anti-flap on timezone change)
  • Separate toggle to disable just the overview step if your Jira is slow
  • Predictable cost: ~$0.10-0.15 per project per week (LLM + sync)
  • No EM intervention required — the analyses land, the EM reads

Automatic coaching at creation

The moment a dev creates a Jira ticket or opens a PR, Pulse reviews against Definition of Ready / PR quality criteria and privately coaches the author in Teams — without going through the manager.

  • 7 DoR criteria on every created Jira ticket
  • 5 quality criteria on every PR (GitHub / GitLab / Bitbucket)
  • Private Teams message to the author, concrete examples provided
  • Quality training at the moment that matters
  • Zero heavy process, zero manual PM review

Conversational Teams bot

Nearly 30 natural-language intents (FR / EN): project status, sprint forecast, KPIs, on-demand digest, coaching, project mute, doc questions, ceremony prep, retro action tracking. No commands to memorize.

  • « Exec view » / « Where does my portfolio stand? » — EM cross-project snapshot
  • « Where does the Alpha project stand? »
  • « Will Alpha's sprint hold? »
  • « Prep Alpha's grooming » / « Prep the planning »
  • « Note that Alice needs to refactor CI by Tuesday »
  • « Action a1b2c3 is done »
  • « Mute Alpha » / « Reactivate Alpha »
  • Also works as chat in the dashboard

Enriched ceremony prep packs: grooming, planning, retro, daily

Pulse doesn't add meetings. It preps the ones you already have. 24h before each configured ceremony, a focused prep pack is pushed to the project's LEADs in Teams. After a July refactor, each pack now surfaces 3-5× more actionable signals — no more isolated stagnating tickets but a real action plan for the facilitator.

  • Grooming Pack: DoR fails · dangling deps (deps to a DONE ticket) · Jaccard duplicates · tickets > 13 SP to split · 28-60d stagnating vs drop-candidates > 60d
  • Planning Pack: 8 risk categories (BLOCKED explicit / by-link / incomplete DoR / linked abandoned PR / oversized / stagnating / no-priority) + falling velocity + bugs ratio + carry-over
  • Retro Pack: sprint metrics + delta vs sprint N-1 + scope creep + DONE effort mix + PR metrics (merged / review time / no-ticket) + top blocker signals
  • Daily Pack: per assignee (yesterday / today / blockers / WIP stalled > 3d / PR reviews pending) + cross-team blockers + PR IN_REVIEW stalled + sprint burn day-N %
  • Auto T-24h cron configurable OR on-demand trigger (« prep the grooming »)
  • 48h after the ceremony: Pulse measures how many tickets were actually groomed (measurable ROI)
  • History + outcomes visible in the dashboard's Impact Panel

Automatic celebration of avoided overcommits

2 weeks after a Planning Pack that warned TIGHT or OVERCOMMIT, Pulse measures the following sprint's closure. If the team shipped ≥85% of committed, a « 🎯 Overcommit avoided — well done! » message is automatically posted in the Teams meeting chat. The proactive loop goes all the way: signal → human adjustment → sprint delivered → measured recognition.

  • Trigger: outcome evaluated at sprint close (not 48h — the real timing)
  • Shipped/committed delivery ratio computed on story points (PRs excluded)
  • « Warn adjusted » flag when pack warned + sprint closed ≥85% = win
  • Push in the initial planning meeting chat (conversational continuity)
  • 1:1 fallback to LEADs if meeting chat unavailable
  • Impact Panel: dedicated « Avoided overcommits » tile with cumulative counter and average ratio
  • History in /prep-packs with dual sparkline overcommits + delivery ratio over 12 weeks

Retro actions tracked in Teams

Capture the action items decided in retro directly from Teams — Pulse persists them, lists them on demand, and automatically reminds LEADs when a due date approaches. No more actions forgotten in a shared doc.

  • « Note that Alice needs to refactor CI by Tuesday » → captured + owner + due date
  • « List open actions on Alpha » → list sorted by due date with 🔴 overdue / 🟡 due today emojis
  • « Action a1b2c3 is done » → immediate closure
  • Automatic daily 9am reminder to LEADs for actions due/overdue
  • Also visible in the dashboard with complete + delete buttons

Confluence RAG cross-referenced with your live delivery

Pulse indexes your project-linked Confluence spaces AND cross-references each answer with the current Jira tickets / GitHub PRs of the user asking. That's the real wedge: Glean reads the doc, LinearB sees the delivery — Pulse combines both in real time in Teams.

  • « How do we deploy the V33 migration? »
  • → Procedure cited from Confluence [1][2]
  • → « You currently have MLE-142 (in review) on this migration — that's probably where these steps apply. »
  • Delivery context injected only if semantically relevant to the question
  • Explicit refusal if the doc doesn't cover it — no invention
  • Nightly indexing + on-demand re-indexing + auto page summary

Cross-project Weekly Executive Digest

Every Monday 09:00 UTC, a « 🚀 Pulse status — last 7 days » digest aggregating all your Mazylon projects is posted to OWNER/ADMIN. Targets EM/CTO steering multiple projects and wants a TL;DR without opening 5 dashboards. Auto-skip when the week has been quiet.

  • Cross-project counters: avoided overcommits, packs pushed, open critical signals
  • Average delivery ratio of plannings evaluated over the last 7 days
  • Long-term context: « X this week · Y since day one » when the ratio is significant
  • Unlimited DB archive + /history endpoint to reconstitute a retrospective board file
  • On-demand dry-run preview (see now what will land Monday)
  • Send-now with 24h rate limit for board demo or impatient beta client
  • SVG sparkline of the overcommits + delivery ratio trend over 12 weeks
  • All-time CSV export (13 analytical columns) for Excel / Google Sheets

Unified delivery model

The engine correlates Jira ticket, GitHub PR and Teams conversation around the same unit of work. One timeline, always traceable.

  • Correlation via rules + LLM judge
  • Explicit evidence (never black-box)
  • Multi-project, multi-team, ACTIVE / DORMANT projects
  • Immutable audit log for every signal

Automatic reports

Nine reports generated on demand or on webhook — from the dashboard OR from Teams in natural language. Always from real data, never from gut feel.

  • Daily Digest — 8am every business morning, pushed to the project's LEADs
  • Grooming Pack — 24h before your grooming, top 8 tickets to prep
  • Planning Pack — load per person + overcommit verdict
  • Discovery — factual state of a project
  • Agile Coach — SAFe/Agile recommendations on 30 key tickets
  • Project KPIs — ticket & PR counters by status / type / provider
  • Sprint forecast — ETA + velocity + confidence level
  • Post-mortem — LLM root cause on closed sprint with carry-over
  • Release Notes + Demo Support — auto-generated at sprint close
  • 'Grounded only' mode: every sentence is sourced

Multi-project PMO view

For Head of Eng / VP Eng steering a project portfolio: consolidated roadmap, cross-project dependencies, global capacity per team.

  • Per-project risk heatmap
  • Cross-project dependencies detected
  • Projected vs committed capacity
  • Drill-down to the unit of work (ticket / PR)

Per-user notification preferences

Each user finely controls what Pulse pushes to their Teams. 12 notification kinds individually toggle-able, 3 one-click presets, control from dashboard OR from the Teams bot in natural language. An EM steering 5 projects doesn't just take it — they choose.

  • 12 granular kinds: daily digest, weekly org, critical/warning signals, prep packs, celebrations, ceremonies detected, retro actions, connector revoked
  • 3 one-click presets: « EM Recommended » / « Silence except critical » / « Reactivate all »
  • /settings/notifications UI with toggles grouped by category
  • Teams bot: « mute daily digest », « apply the EM Recommended preset », « what notifications do I get? »
  • Backward-compat default: all kinds active by default, no user silenced on rollout
  • OWNER can bulk-apply a preset to all LEADs of a project in one click
  • Admin analytics: « most muted kinds » = direct product signal on what feels like noise
  • Weekly auto-alert to OWNERs when a kind exceeds 50% opt-outs (rework content or timing)

Tool calling (opt-in)

From V2 and always opt-in: Mazylon can act on Jira, create sub-tasks, add comments, schedule Teams meetings.

  • Systematic user confirmation
  • Strict audit log for every action
  • Disableable per connector and per tenant

V2 and beyond

An agent-first roadmap.

The Scrum Master agent that preps your ceremonies is already here. Coming soon: a Delivery Manager agent that answers 'why is this sprint going to miss'. A PMO agent for multi-project visibility. Mazylon isn't one more tool — it's a team of agents that collaborate.

Ready to see your sprints differently?

30 minutes with one of our founders. Live demo on your Jira if you want. No commitment.