A SafeFoundry methodology

The Foundry Atlas

A two-day sprint that finds the AI use cases that will actually work in your business — and the four to pilot first. No strategy decks. No maturity scores. Just a ranked, scored, action-ready map of where AI applies in one operational function, and the pilot blueprints to start next quarter.

Format
Two-day structured sprint
Output
A ranked Atlas + four pilot blueprints
Built for
Function leaders piloting AI in 2026
The deadlock the Atlas breaks

Most enterprises stuck on AI aren't short of ideas

They have ten use cases on a whiteboard, three pilots half-funded, and no clear way to choose. The Atlas exists to break that specific deadlock — to convert ambition into a ranked, defensible plan in 48 hours.

i
Too many candidate use cases, no way to rank them
Workshops produce long lists. Vendors pitch their favourites. Internal teams advocate for what they understand. Without a defensible scoring model, the loudest voice wins — not the highest-impact use case.
ii
Maturity assessments don't tell you what to do next
A 32-page maturity report tells the function it's a 2.4 out of 5 on AI readiness. It doesn't tell anyone which pilot to start on Monday. The gap between assessment and action is where most AI programmes die.
iii
Pilots launch without commitments and quietly disappear
A workshop produces enthusiasm. The function returns to its quarter. By month two, the pilot has no owner, no budget, and no senior sponsor. The Atlas closes by naming the owner in the room — or it isn't a pilot.
What you walk away with

An Atlas. Ranked, scored, action-ready.

Every Foundry Atlas produces the same five-section deliverable. By the end of Day 2, the function owner has a working document they can act on the next morning — not a slide deck that sits in a shared drive.

i
Function summary.The strategic question the sprint answered, the participants who answered it, and the framing the Function Owner committed to in the room.
ii
Use case inventory.Every candidate AI use case in the function — typically 12 to 20 — discovered through the six-surface operational framework. Nothing missed.
iii
Scored grid.Every candidate rated on Impact, Data Readiness, and Speed to Value — with the rationale for each rating captured in the room. Defensible, not opinion.
iv
Tier classification.Every candidate placed in one of four tiers — Show First, Strong Case, Mention, Enabler — using rule-based classification. No tribal politics.
v
Pilot blueprints.One-page plans for each Tier 1 use case: the operational change, the data inputs, 30/60/90-day milestones, success metrics with baselines, ROI estimate, and the named implementation owner.
Atlas · Retail Field Ops · Excerpt
Tier classification
Visit reportingTier 1
Lead prioritisationTier 1
Regional narrativeTier 1
Outlet anomaly detectionTier 1
Planogram complianceTier 1
SKU push selectionTier 2
Mystery shopper replacementTier 2
WhatsApp follow-upsTier 3
Visit-to-CRM data flowTier 4
From the Foundry Atlas worked example — see full document below.
See a complete worked example →
How it works

Two days. Eight hours of pre-work. A finished Atlas.

The Foundry Atlas is a tightly designed sprint. Pre-work is structured, the room is small, the timeboxes are enforced. By 1700 on Day 2, the room walks out with a finished Atlas and named pilot owners — or the sprint hasn't worked.

0
The week before. The Function Owner completes the Inventory Template — eight hours of structured discovery using a six-surface framework that works across any operational domain. The Data Person reviews and adds data notes. The Working Grid is sent to the Facilitator 48 hours before Day 1.
Day 1
Validate and Score
0900
Open and frameFunction Owner states the strategic question. Facilitator runs the methodology brief.
0945
Validate the inventoryWalk every use case. Operators ground-truth. Add what was missed.
1130
Score ImpactEvery use case rated High, Medium, or Low. Rationale captured.
1400
Score Data ReadinessData Person leads. Yes / Partial / No, with a one-line gap note.
1545
Score Speed to ValueImmediate / Short / Long, with the measurement question framed.
Day 2
Classify and Blueprint
0900
Day 2 frameRecap. Walk through the four tiers and the assignment rules.
0930
Tier classificationRule-based classification. Function Owner decides boundary calls.
1115
Pilot blueprints — first passPairs draft eight-field blueprints for each Tier 1 use case.
1400
Review and refineRoom critiques. Implementation owners named in the room.
1545
Atlas readoutFunction Owner presents to senior leadership observer. Pilots committed.
The methodology underneath

Four components. Defensible structure.

The Atlas is built on a four-component methodology that any SafeFoundry-trained facilitator can run. Each component has explicit rubrics, rules, and templates — so two facilitators running the same engagement produce the same Atlas.

i
Function Inventory
Discovery through the six-surface operational framework — transactional, decision, communication, intelligence, coordination, quality. Nothing missed because the framework forces examination of every surface.
ii
Scoring Model
Three dimensions, each with explicit rubrics: Impact, Data Readiness, Speed to Value. Every score has a rationale captured in the room. Defensible against any "why is this rated higher" question.
iii
Tier Classification
Rule-based assignment to one of four tiers — Show First, Strong Case, Mention, Enabler. The rules eliminate most arguments. The remaining debates are real and worth having.
iv
Pilot Blueprint
For every Tier 1 use case: operational change, data inputs, 30/60/90-day milestones, success metrics with baselines, ROI estimate, named implementation owner. Without an owner, the blueprint isn't done.
Three ways to run an Atlas

Pick the mode that fits your function

The methodology is the same in every mode. The difference is who holds the marker.

Consultant-led · Most popular
A SafeFoundry facilitator runs your sprint
For first-time AI engagements, politically sensitive functions, and senior leaders who want an outside voice to surface honest data and sharpen decisions.
  • SafeFoundry-certified facilitator runs the full two-day sprint
  • Structured pre-work with the Function Owner and Data Person
  • Final Atlas document delivered within 24 hours of Day 2
  • 30-day post-sprint check-in to validate pilot kickoffs
  • Optional: pilot delivery support as a follow-on engagement
Pricing on request — scoped per engagement based on function complexity and geography.
Request an introductory call
Self-run kit
Run it yourself with the Atlas Kit
For internal teams with operational rigour and prior exposure to design-sprint formats. Everything you need to run a sprint without a facilitator.
  • Methodology playbook (PDF)
  • Inventory Template + scoring rubric reference cards
  • Pilot blueprint template (eight-field)
  • Worked-example Atlas as reference
  • Email support for first sprint
From ₹49,000 · One-time per organisation
Get the kit
App-guided · In development
A facilitator app walks you through it
A facilitator — internal or external — is guided by the SafeFoundry app, which structures inputs, applies rubrics, and produces the Atlas as output. Built for organisations running multiple Atlases per year.
  • App guides facilitator through every sprint phase
  • Auto-generated scored grid and tier classification
  • Atlas document produced on Day 2
  • Cross-engagement benchmarking across functions
  • Early access — limited cohort
Currently in private beta with selected partners.
Join the waitlist
Honest fit

Who the Foundry Atlas is for — and isn't

The Atlas is built for a specific kind of moment in an enterprise's AI journey. If you recognise yourself in the left column, this is the right engagement. If the right column fits better, we'll tell you so.

A good fit if
  • You lead an operational function — field ops, distribution, customer service, branch ops, claims — with a clear P&L or operational outcome.
  • You have candidate AI ideas already — from internal teams, vendors, or workshops — and need to rank them defensibly.
  • You can commit a two-day sprint window with the right people in the room, including yourself.
  • You want pilots that start next quarter, not a strategy deck for next year.
  • You're prepared to own a pilot commitment by name at the end of Day 2.
Not a fit if
  • You're looking for an AI maturity score — that's a different product with different buyers.
  • You want a three-year strategic roadmap for AI across the enterprise — the Atlas is one function at a time.
  • You need an implementation engagement first — the Atlas produces blueprints, not running pilots. Implementation is a follow-on.
  • The Function Owner cannot attend both full days. Without them, the sprint produces recommendations, not decisions.
  • Your function has no operational data — we'd recommend foundational data work first, then come back.
A complete worked example

See what an Atlas looks like — before you commit.

A full Foundry Atlas built for retail field operations at a representative consumer electronics brand. Sixteen candidate use cases discovered, scored, classified, and five Tier 1 cases with full pilot blueprints. The methodology, the rationale, and the deliverable — exactly as you'd receive it.

Read the worked-example Atlas →
16
Candidate use cases discovered through the six-surface framework
5
Tier 1 use cases with full pilot blueprints
2
Days from open to Atlas — pre-work outside the sprint
3-4 Cr
Year 1 ROI envelope across all five Tier 1 pilots, indicative
Honest answers

Questions you'd actually ask

Why two days, not five?
A five-day sprint is too long for what is fundamentally a structured analytical exercise — and it's hard to get senior people to commit five full days. Two days is the honest length: long enough to do the work properly, short enough to actually happen. The pre-work is what makes two days possible.
How is this different from an AI maturity assessment?
A maturity assessment measures where you are. The Atlas measures what to do next. They're different products with different buyers. Maturity assessments produce a 32-page report and a score. The Atlas produces a ranked list and four pilot blueprints with named owners.
Why do you insist the Function Owner attends both full days?
Because without them in the room, the sprint produces recommendations, not decisions. The Function Owner is the tiebreaker on contested scoring, the decision-maker on Tier 1, and the person who commits to pilots in the readout. A two-day sprint with the wrong people produces a polished document nobody owns.
What happens after Day 2?
The Atlas is finalised within 24 hours and circulated. Pilot kickoffs should happen within seven days of sprint close — that discipline is what separates a sprint that produced a document from one that produced action. SafeFoundry can support pilot delivery as a follow-on engagement, but the Atlas itself is a complete deliverable on its own.
Will my data leave the room?
No. The sprint runs on whatever data tools your function already uses. Inputs and outputs stay inside your environment. SafeFoundry uses the structured artefacts (the inventory, the scored grid, the pilot blueprints) to produce the Atlas document, but does not retain underlying business data after engagement close.
Can we run the Atlas across multiple functions in parallel?
Yes — but each function gets its own sprint with its own Function Owner. The methodology is built around one operational function at a time. Trying to compress two functions into one sprint produces two thinner Atlases, not one richer one.

Find the AI use cases that will actually work in your business.

A 30-minute introductory call. We'll discuss your function, the strategic question worth answering, and whether the Foundry Atlas is the right next step. No pitch deck.

Book an introductory callSee the worked example first