The Delivery Manager's Guide to Archgram Flow

RAID logs linked to architecture nodes, confidence-scored milestones, and a Handoff Pack that replaces knowledge-transfer meetings

Delivery managers live and die by visibility. Is the team on track? Which risks are materialising? Which dependencies are going to slip? In most organisations this information is scattered across Jira, a RAID spreadsheet, a milestone slide in PowerPoint, and a weekly status email that is out of date by the time it arrives.

Archgram Flow gives delivery managers a single structured view of delivery health — with one differentiator that spreadsheets can never offer: every risk, dependency, and workstream is anchored to the architecture it affects. When you log a risk about the payment gateway, it is linked to the Payment Service node in the diagram. When a stakeholder asks "what is exposed if that risk materialises?", the answer is a click away.

This article covers how delivery managers specifically use Flow — RAID log, workstreams, milestone confidence scoring, and the Handoff Pack.

Setting Up Workstreams

Large Initiatives are almost always delivered by parallel teams. Archgram Flow models this with Workstreams — named delivery tracks, each with an owner and a set of assigned slices.

Create workstreams from Flow → Workstreams → New. Typical workstreams for a platform initiative might be:

Frontend — UI, accessibility, browser testing
Core API — business logic, domain services
Data — schema migrations, reporting, compliance
Platform — infrastructure, CI/CD, observability

Once workstreams are created, slices are assigned to them. The Timeline view shows all workstreams side-by-side, with their slices and milestones rendered as a Gantt-style layout. Dependencies between workstreams appear as arrows — and are automatically mirrored in the RAID log as Dependency entries.

Running the RAID Log

The RAID log in Flow is a live structured register — not a spreadsheet. Every entry has a type (Risk, Assumption, Issue, Dependency), a status, an owner, and — crucially — a link to the diagram node it affects. This means the DM can answer "which part of the architecture is under the most risk?" by looking at the diagram rather than mentally cross-referencing a spreadsheet.

The DM reviews the RAID log at least weekly. The recommended workflow:

Open the RAID log — filter by status Open and review each entry
Update statuses — risks that are mitigated move to Mitigated; assumptions that are validated move to Closed
Adjust confidence — after reviewing open risks, update milestone confidence scores
Log new items — anything that surfaced in the weekly standup or steering update is added immediately
Archive resolved items — resolved RAID items move to Archived but are retained for the Handoff Pack

A concrete RAID entry from a payment platform delivery:

Type: Dependency
Title: Identity provider SDK v3 release from Platform team
Owning team: Platform Engineering
Expected delivery: 2026-03-14
Impact if missed: Auth Service cannot be completed; M1 delayed by at least 2 weeks
Workaround: Use SDK v2 with compatibility shim (adds 3 days of rework)
Linked node: Auth Service
Status: Open — escalated to Platform EM on 2026-03-01

RAID items linked to diagram nodes flow into milestone confidence — surfacing risk before it becomes a missed date

Milestone Confidence Scoring

This is the practice that separates delivery managers who surface problems early from those who report missed dates after the fact.

Every milestone in Flow carries a confidence score from 0 to 100%. You set it. You update it every week. It is your professional judgement, not a formula derived from story points or velocity. A milestone at 90% two weeks out says "I expect this to land on time, there are minor open risks." A milestone at 55% one week out says "we are in trouble and leadership needs to know now."

Confidence scores create a historical signal. When you chart confidence over time for a milestone, a smooth decline from 85% to 45% over four weeks tells a different story than a sudden drop from 80% to 20% in a single week. Flow renders this history as a chart and includes it in the Handoff Pack — so the next DM on the programme can understand what happened and why.

Open high-impact RAID items surface automatically as milestone warnings in Flow:

⚠ Milestone M2 — Notification Service Live (2026-03-28)
Current confidence: 70%

Open risks affecting this milestone:
• HIGH: Email provider rate limit under load (linked: Notification Service)
  Suggestion: reduce confidence by 15–20% until mitigation confirmed
• MEDIUM: SQS queue depth monitoring not yet configured
  Suggestion: reduce confidence by 5–10% until resolved

Recommended confidence: 50–55%

You are not obligated to accept these suggestions — the DM always sets the final score. But the prompts ensure you are at least considering the open risks before reporting to stakeholders.

Weekly Delivery Cadence with Flow

A practical weekly rhythm for delivery managers using Flow:

Monday — RAID review (30 min): Open each Risk and Dependency. Confirm owners have updates. Log anything that came up over the weekend. Flag any items that need escalation.
Wednesday — Slice board check (15 min): Review In Progress slices. Identify any that have been In Progress for more than two weeks without movement. Flag to workstream owner.
Thursday — Milestone confidence update (15 min): Update confidence scores for all milestones within 6 weeks. If any score drops by more than 15% from last week, prepare a summary for the steering update.
Friday — Status artefact: Export the milestone summary from Flow (single page: milestones, confidence trend, open RAID count) for the weekly status email. Takes 2 minutes instead of an hour.

The Handoff Pack: The DM's Closing Artefact

When the Initiative is complete — all MVP slices Done, milestone exit criteria met — the delivery manager generates the Handoff Pack from Flow → Export → Handoff Pack.

The DM's contributions to the pack are:

delivery/raid.md — the complete RAID log, including archived items, showing how risks were identified and resolved throughout delivery
delivery/milestones.md — milestone history with confidence score charts, actual vs. target dates, and exit criteria records
delivery/slices.md — all slices with statuses, showing the scope that was delivered vs. deferred

The AI-generated README synthesises the delivery story from these inputs — what was built, what risks were managed, what was deferred and why, and what the recommended next steps are. The DM reviews the README and can edit it before the pack is finalised.

When Flow Makes the Most Difference for DMs

Flow provides the most value to delivery managers when:

Multiple workstreams are running in parallel — interdependency tracking in RAID + the Timeline view prevents the "each team reports green until they collide" failure mode
Architectural decisions have delivery consequences — a superseded ADR that changes the system design mid-delivery surfaces immediately in the linked milestones and slices
Stakeholders need regular visibility — confidence-scored milestones are a more honest signal than RAG statuses, and they are generated from live data rather than a PowerPoint update
There will be a handoff — the RAID history and delivery narrative in the Handoff Pack replace weeks of knowledge transfer meetings

Getting Started as a Delivery Manager

Once the architect has created the Initiative and attached diagrams, open Flow and create your Workstreams
Set up milestones with target dates and initial confidence scores
Assign slices to workstreams as they are created by the PO
Log your first RAID item — even if it is a low-risk assumption — to establish the habit
Review and update confidence scores weekly without fail; the trend is more important than any single score

Full reference for RAID fields, milestone configuration, and Handoff Pack generation is in the Archgram Flow documentation.