Balsam COE Founding Brief

Prepared for Claire and France by Goose Group.

This is a shareable working brief for discussion with Claire and France.

This brief is meant to make the COE concrete enough to discuss, adjust, and refine together.

Purpose

The strategy sprint identified more worthwhile work than Balsam can act on through a report alone.

A few examples from the sprint:

• Kate Hollywood mapped a 19-step product data workflow with 18 bottlenecks. She has already done the hard discovery work. What she needs now is engineering support and a way to turn that map into actual changes.
• Josh Falco built Figma plugins with Claude Code. Lindsay Shrum validated reconciliation agents against historical data. Andrea Gouw is running an AI automation effort with a team of two. These are not hypothetical future champions. They are already building.
• The content production workflow touches creative, eCommerce, international, and marketing at the same time. Multiple teams independently described content architecture and content handoffs as a binding constraint.
• Bob McDonald's group is still manually translating product data into 4-5 retailer-specific formats per retailer. Balsam is targeting 10 retailers by 2027. Every new retailer multiplies the same manual work.

The same structural pattern showed up across the business:

• important knowledge lives in people, spreadsheets, Slack threads, and disconnected systems
• problems in content and product data create delays for other teams later in the process
• people are already building useful things, but they are doing it in isolation
• Balsam has pods for delivery, but not a clear way to prepare and prioritize work before it reaches the pod roadmap

These are not just internal problems. They show up in what customers experience: a new market that launches without the full product catalog, a retailer partner waiting on product data, a holiday shopper who sees inconsistent content across channels. The internal friction is real, but what matters is what it costs the customer.

Balsam has more good ideas than it can act on.

It needs a practical way to:

• choose the right work
• define that work clearly before it hits a delivery queue
• generate enough evidence early to justify more investment
• spread the learning beyond the first few people who are already moving

That is the job of the COE.

Mission

The COE's mission is to help Balsam turn real workflow insight into clear initiatives, working improvements, and stronger internal capability — so that customers in every market get the full Balsam experience and retail partners get the data they need without delay.

In practice, that means:

• starting with the workflows that matter most
• working with the people closest to the work and most ready to invent
• producing useful artifacts and evidence early
• handing validated work into Balsam's existing delivery teams
• leaving Balsam stronger and more self-sufficient after each cycle

Founding Tenets

Five principles that guide what the COE does and how. These are the reasons we say no to specific things, and how we settle disagreements when two reasonable people see different paths.

1. Workflows, not tools. Start with real work that matters. AI fits inside the workflow, not the other way round.
2. Venture, not min-max. A portfolio of bets, chosen for learning and upside. Not the single highest-ROI initiative identified through analysis.
3. Upstream before downstream. Fix the source before accelerating what happens with it. Moving bad data faster is a worse outcome than leaving it alone.
4. Collaboration is contribution. Proposals beat comments. Pull requests beat peanut-gallery feedback. Decisions are recorded, not relitigated.
5. Build toward independence. Every cycle leaves Balsam stronger. By the end of the founding period, Balsam runs the COE.

Full reasoning for each: tenets.md.

What The COE Is

The COE is a small team and operating structure that sits between ideas and execution.

It should complement the Balsam Operating Council, the AI Council, and the existing pod structure, not compete with any of them.

Simple version:

• the Operating Council prioritizes
• the AI Council sets standards and enables the broader company
• the COE helps staff and leaders define and test the work
• the pods ship the work that is ready to be shipped

What It Is Not

The COE should stay clear about what it is not.

It is not:

• another static strategy phase
• a generic AI training program
• a shadow product organization
• a place where ideas go to wait
• a closed group chosen only from the top
• a permanent external dependency
• a technical support function for AI tooling

Goose Group's value is strategy, workflow design, and capability transfer. Cohort members should be people who are ready to build and experiment — people with energy and initiative, not people who need onboarding on basic tools. The COE will help people work differently, but it starts from a baseline of curiosity and willingness to try.

The goal is to create a way of working Balsam can keep using, not a permanent layer that depends on Goose Group staying in the middle forever.

How Goose Group Would Work With Balsam

Goose Group works alongside the people who know the work best. The job is to listen first, work with the people closest to the workflow, turn promising ideas into concrete briefs and working improvements, and help Balsam's teams own the wins and carry the pattern forward themselves.

1. Start with workflows, not tools

Goose Group would start from real work that already matters to the business, not from a generic list of AI use cases.

The useful questions are where a workflow is constrained today, what is limiting it, and what focused intervention would materially improve it.

2. Work with the people closest to the work

The best starting points are the people who already understand the work in detail, have credibility with adjacent teams, and have the energy to invent.

That is why the first cohort should come from the people closest to the work, not from an org-chart exercise. They need structure, collaboration, and a way to share and build on what they already know.

The list should come through a visible intake process, reviewed with leadership and prioritized with the Operating Council.

3. Use structured timeboxes to generate evidence and make decisions

Typical pattern:

• map the workflow
• define the constraint clearly
• build a small prototype or demo quickly
• review it with the people who use the workflow
• decide whether it should be improved, handed off, or stopped

Early versions help define what a later production version should do. The work gets clearer through doing, measuring, and deciding, not by describing.

4. Build compounding tools, products, and working assets

Each cycle should leave behind things Balsam can keep using:

• workflow maps and one-page initiative briefs
• prototypes, internal tools, and lightweight products
• prompts, playbooks, documentation, and operating notes
• data products or shared internal knowledge layers that make later AI work easier
• readouts and handoff material for pods, business owners, or implementation partners

That might take the form of a small internal tool, a data product, a shared markdown library, an MCP-compatible context layer, or a portal page that helps other teams understand and reuse the work.

5. Create compounding value beyond the cohort

The cohort is the starting point. Value flows outward — through the workflow, into growth teams, and to the customer.

• Cohort: roughly six members working directly with Goose Group, spending focused time each week on real workflows, prototypes, and decisions.
• Workflow teams: content production, product data, creative ops. Their work gets faster and cleaner. This is where coaching, rollout, and practical adoption happen.
• Growth teams: international, marketing, indirect sales. They move faster because upstream work actually flows. Fewer meetings, more agility, faster decisions.
• Customer: a shopper in Germany or Australia gets the same beautiful, complete product experience as a US customer from launch day. A retail partner gets clean product data on time. Peak-season buyers get the right offer when it matters, not two days later. The value does not stop inside Balsam.

The goal is not efficiency for six people. It is better outcomes for the teams and customers downstream.

6. Stay close to the work

The COE will be most effective when it is close enough to the work to see what is actually happening.

That means:

• direct working sessions with cohort members
• async collaboration in Slack or the systems Balsam already uses
• access to the relevant tools, docs, and safe data
• regular sponsor syncs with Claire and France

7. Build toward independence from the beginning

The COE should be designed so Balsam owns more of it every month.

That means:

• using Balsam's people and systems wherever possible
• documenting the process as it runs
• identifying the likely long-term internal owner early
• making capability transfer part of the work, not an afterthought

What Else The COE Should Cover

The cohort incubation work is the center of the model, but it should not be the whole model.

To get the full value from the COE, it should also include the surrounding activities that make the work visible, usable, and easier to scale across the company.

1. Training and outward-facing internal activities

The COE should give Balsam a practical way to bring the wider company along.

That can include:

• talks at company all-hands or leadership meetings
• hackathons or build days around specific workflows
• office hours and working sessions for people outside the active cohort
• starter kits, playbooks, and internal enablement material
• coordination with AI Council communications into the broader company

The goal is to make the work visible, accessible, and repeatable.

2. Operating Council and AI Council collaboration

The COE works closely with the Operating Council as it prioritizes, reviews opportunities, and makes decisions, and with the AI Council as the enablement and standards body for AI across Balsam.

For the Operating Council:

• preparing agendas, pre-reads, and decision memos
• helping teams improve and enrich incoming ideas before review
• helping compare initiatives across functions
• turning priorities into actual next steps
• reporting back on what cohorts, champions, and experiments are producing

For the AI Council:

• sharing monthly COE readouts so tooling, policy, and training decisions are informed by real work
• surfacing skills, playbooks, and patterns worth standardizing across Balsam
• flagging AI Council decisions (licensing, data access, policy) that are blocking cohort progress

The COE gives both bodies better-prepared choices, clearer comparisons across functions, and a steady view into what the cohorts are producing.

3. Collaboration, vendor management, and AI technology strategy

Part of the job is helping Balsam coordinate the teams, vendors, and technology decisions around this work.

That includes:

• connecting business teams, product, data, and engineering around the same workflows
• helping Balsam get the right level of engagement from vendors
• making sure platform decisions help this work move
• identifying when a vendor can solve a problem, when Balsam should configure something itself, and when custom build work is the right move

Example:

• if Snowflake should provide forward-deployed help for analytics work coming out of the COE, the COE can help organize that conversation and move it forward

4. Fractional VP of AI leadership

Balsam also needs senior cross-functional AI leadership around the program. Goose Group can provide that leadership during the founding run by helping Balsam:

• connect the work across functions
• identify where the next useful initiative is
• navigate tool, vendor, and implementation decisions
• keep the company moving from isolated experiments toward a more coordinated way of working
• act as connective tissue across the broader AI landscape — Pablo's initiative, Bernie's efficiency work, company-wide training — so the moving parts see each other

This is active operating leadership around the program, not project administration.

Sunset on the coordination role. Through Month 4, Goose Group carries the cross-landscape coordination. By Month 5, this transitions to an internal owner emerging from the program. Goose Group stays available in a lighter advisory posture afterward. The goal is not permanent connective tissue from the outside — it is getting Balsam's own connective tissue in place.

5. A visible home for the work

The COE should have a visible home inside Balsam.

We should help build a simple internal portal or microsite where employees can:

• see what the COE is for
• submit or review ideas
• see who the current champions are
• review current and past cohort work
• access readouts, starter kits, and playbooks
• see visible examples of impact

This gives people a consistent place to understand the COE, follow its work, and decide how to engage.

6. Marketing strategy support

Pascal Staud is a Goose Group collaborator focused on brand, growth, and creative operations. He ran STAUD STUDIOS for over a decade — a marketing and creative operations firm working with BMW, Mercedes-AMG, and other premium brands across European markets. More recently he was a co-founder at Monks, a global marketing services company.

Pascal is part of Goose Group and will be involved in COE work where his skills are relevant.

Separately, Balsam's marketing function could benefit from a dedicated marketing strategy partner — someone to work with Karima on positioning, spend allocation, channel strategy, and team structure. That work is an input to the COE, not part of it. We recommend an introductory conversation between Pascal and Karima to explore whether that makes sense. It does not need to be decided before the COE launches.

7. Bring in additional implementation capability when needed

The COE should help Balsam get the right implementation capability behind validated work.

That may include:

• internal Balsam pod capacity where the work is ready
• trusted external partners
• Goose Group contractors
• direct support from key technology vendors

Goose Group can help scope the work, compare options, and introduce the right people when a workflow problem needs more specialized capability or more delivery capacity than the core COE team should provide.

How Work Should Enter The COE

The COE should combine a focused set of initial opportunities with an open intake.

The sprint already surfaced several strong candidates for the founding investment opportunities: content production, product data and retailer onboarding, and a decision-support track in marketing or analytics. Those should be written up as short opportunity briefs and used as the initial slate for review.

Open intake should begin at the same time so staff and leaders can bring forward additional ideas.

Every opportunity, including the obvious starting ones, should go through the same intake discipline. Goose Group should help teams write the case, make the value legible, and bring it to the Operating Council in a comparable format.

A practical intake pattern is:

1. Put out a clear call to Balsam staff asking for workflow opportunities, ideas, and volunteers.
2. Let leaders and sponsors nominate work they believe matters.
3. Use Goose Group to help teams turn the already-identified opportunities and any new submissions into a short one-pager and a simple business investment brief.
4. Review those briefs with the Operating Council and sponsors so opportunities can be compared side by side.
5. Select the first cohort and first workstreams through that decision process.
6. Start the work and publish the choices and the reasons so people can see how decisions are being made.

This keeps the COE open, credible, and useful to people beyond the initial champions.

What The COE Does Each Cycle

Every cycle should include a mix of six activities.

Opportunity review and selection

• gather opportunities from staff, leaders, and active champions
• identify the workflows that are worth attention now
• help teams write one-pagers and business investment briefs that make the value legible
• help the Operating Council and sponsors prioritize opportunities and select the cohort
• define each selected item clearly enough to act on

Workflow mapping and requirements

• document how the work actually moves today
• surface the handoffs, bottlenecks, and missing context
• turn broad observations into specific points of intervention

Rapid validation and lightweight build

• produce quick demos, prototypes, helpers, automations, or internal tools
• test the intervention with the people who would actually use it
• use the result to decide what should be pushed further

Office hours and unblockers

• give participants practical help as they hit friction
• keep useful work moving between formal sessions
• help connect business context to technical possibilities

Handoff into execution

• package validated work for the appropriate pod or owner
• provide enough documentation, evidence, and requirements that the work can move cleanly
• avoid making pods do weeks of rediscovery before they can start

Workflow rebuild at the handoff layer

A workflow is a series of handoffs between people and tools. Many of the delays in Balsam's workflows sit in those handoffs, where a person is standing in for an integration that should exist.

The content workflow is a good example:

• a brief gets re-entered manually into ClickUp structures
• assets still move between tools through desktop saves and uploads
• translated content gets manually re-entered into the CMS

The cohort's job is to map those handoffs together and improve the ones that create the most delay or rework.

Josh Falco's Figma CSV plugin is a useful early example of this way of working. It did not just make one person faster at an existing task. It removed a recurring manual step from the workflow.

A strong cohort should leave behind a workflow with fewer manual handoffs, clearer operating patterns, and a team that can keep applying the same approach. The test is not just whether the team is faster — it is whether the customer gets a better, more consistent experience because the work behind it actually flows.

Readout and knowledge sharing

• publish what was learned, what was built, and what changed
• let later cohorts start with better context
• make the work visible beyond the six people closest to it

Suggested Operating Rhythm

This rhythm gives people a consistent way to understand the COE, participate in it, and follow its progress.

Recommended cadence:

• Weekly cohort session: working session with the active cohort
• Weekly sponsor sync: Claire, France, and Goose Group review priorities, blockers, and progress
• Weekly office hours: open collaboration time for participants and adjacent teams
• Monthly readout: what was learned, what was built, what should move next
• Async collaboration throughout: Slack, docs, and tool-level follow-up between sessions

Done well, this makes the COE part of Balsam's operating infrastructure rather than a series of disconnected sessions.

The structured sessions are contact hours with Goose Group. The goal is for the rest of the week to change too — cohort members should be doing their existing jobs differently: documenting workflows, testing new approaches, coaching their nearest colleagues, and building on what gets created in sessions. The COE is not a side project. It is how the work gets done. Time commitments vary by tier — see Cohort participants for detail.

Recommended Starting Shape

A good starting structure is a 6-month founding run.

The first four months should be the intensive building period. The final two months should be used to carry the strongest work further into implementation, measure what changed, and transition ownership inside Balsam.

That is long enough to:

• launch the COE with clear intake, cadence, and ownership
• run six or more cohort cycles
• produce visible wins on real workflows
• build a body of artifacts and examples
• identify and equip an internal owner

It is short enough to feel concrete and bounded.

The natural launch moment is the April 29-May 1, 2026 onsite in Monterey Bay.

Balsam is already planning baseline AI training for all 300 employees at that onsite. The COE is the operating program that can follow that baseline training and turn it into actual work.

Recommended framing:

• Month 1: announce the COE around the onsite, open intake, publish a simple home for the work, and gather the first set of ideas
• Month 2: select the first cohort, confirm the first two workflow areas, and begin the first cycle
• Months 3-4: ship the first mapped workflows, prototypes, and initiative briefs, then expand the second ring around the work
• Months 5-6: keep the cycle running, strengthen the internal repository and operating rhythm, and transition toward an internal owner with lighter Goose Group support

If the exact month-by-month plan changes, that is fine. The important thing is to agree that this is a founding period with a clear handoff horizon.

If the model is working and Balsam wants a longer runway, that can be decided later. The first discussion should stay focused on the founding run.

Month-By-Month Outputs

The concrete artifacts and shipped changes to expect at each milestone. Not activity metrics — outputs the Operating Council can hold the COE to.

Month 1 — May (launch)

• COE announced at the April 29–May 1 onsite. Internal home (this repo) published.
• Intake process live, with submission template and criteria.
• Cohort 1 (Product Data) roster confirmed and mission agreed.
• Workflow map for the full product data lifecycle, in progress or done.
• First prototype or skill shipped and in use by Kate and Mary.

Month 2 — June

• Cohort 1 cycle concludes. Published readout covers what was built, what was learned, and what moves next.
• At least one validated initiative handed off to a build partner or internal team with full requirements, evidence, and engagement model.
• Cohort 2 roster confirmed (likely content production or analytics).
• Operating Council reviews and prioritizes open intake.

Month 4 — August

• Three cohort cycles complete.
• Three mapped workflows, three or more shipped prototypes or improvements, a growing set of skills and playbooks in the repo.
• At least one external build partnership active and delivering.
• First alumni coaching new cohort members.
• A named internal candidate emerging as the long-term COE owner.

Month 6 — October

• Six cohort cycles complete.
• Self-sustaining operating rhythm. Goose Group in a lighter advisory posture.
• Internal owner running the program with Goose Group support as needed.
• Measurable change against at least two of the founding ambitions below.

In-Season Operating Approach (Sept–Jan)

Balsam's in-season crunch overlaps with the back half of the founding period. The COE does not pause, but it adapts.

• Smaller cohort size (two to three core owners rather than four or five).
• Consolidation cycles focused on shipping and handoff, not new discovery.
• Alumni from earlier cohorts run more of the work; Goose Group shifts to support.
• Readouts continue monthly; cohort cycles stretch to six weeks.
• Non-critical new intake queues for February 2027.

The 2027 planning window is when the biggest compounding wins show up. The founding period exists so that planning cycle is different from last year's.

Founding Ambitions And Starting Measures

The right mental model is venture-like, but it should stay concrete. Balsam is not funding a generic program. It is funding a portfolio of high-upside workflow bets, where a relatively small amount of focused work can change a much larger business outcome.

The Operating Council and sponsors should agree on a small number of ambitions for the founding run and use those ambitions to steer what gets priority.

Examples of the kind of ambitions that fit this model:

• Increase international opening velocity for the 2027 season. If content and localization no longer block every new market, Pablo's team can open geographies faster with the same people — and customers in those markets get the full Balsam brand experience from day one, not a partial catalog with missing descriptions.
• Reduce the manual work required to onboard new retailers and channels. If product data can move cleanly from PIF to retailer format, Bob's team spends more time growing accounts and less time doing data entry — and retail partners get accurate, complete product information that helps them sell.
• Cut peak-season decision latency in marketing. If Karima's team can move budget and creative support faster than the current ~48-hour lag, the company can react during the weeks that matter most — and holiday shoppers see the right products and promotions while they are still shopping, not after the moment has passed.
• Create a reliable path from validated workflow ideas into France's delivery capacity. The COE should not end at prototype stage. It should produce work that can actually move.

Each ambition should have a few simple starting measures attached to it. For example:

• International / content: time from English source content to localized publish, number of manual handoffs in the localization path, number of geographies that can launch without bespoke workarounds
• Product data / indirect sales: hours spent translating PIF data into retailer formats, number of manual transformations per retailer, time from retailer opportunity to usable product data package
• Marketing / analytics: current ~48-hour signal-to-budget-reallocation lag, time from creative request to live asset, number of vendor reports that still require manual reconciliation
• COE to pod handoff: time from intake to sponsor decision, time from validated initiative to pod handoff, number of initiatives that receive engineering follow-through

These do not need to be finalized in the document. The important thing is that the COE is judged against a small set of real business ambitions, not only against activity.

Cohort 1: Product Data

Product Data is the first cohort. It launches in May 2026 at the Monterey Bay onsite.

Why Product Data first:

• it affects merchandising, indirect sales, inventory planning, vendor communication, and launch work
• the current manual process is already well understood
• Kate has already done meaningful discovery work in this area, which makes it a strong candidate for early intake
• Bob's team is living the daily manual cost of the current process
• fixing it improves work for multiple teams at once
• when product data flows cleanly, retail partners get accurate listings and customers shopping through those channels see the right product, the right description, and the right price
• Balsam decided not to move forward with a PLM this year, which makes Kate and Mary's workflow carry more weight than it was designed to

Upstream before downstream. Cohort 1's work is upstream — the PIF, the master data, the producer-to-consumer hand-off. Moving inaccurate product data into ten retailer formats faster is not an improvement. Fixing the source is. Retailer-specific work and downstream consumer-team adaptations are out of scope for Cohort 1 and come in later cohorts once the upstream is stable.

Core owners: Kate Hollywood and Mary Corrick (to be confirmed by Claire). Participants and adjacent ring to be named before kickoff. The full cohort package lives at cohorts/01-product-data/.

Candidate Areas for Cohort 2 and Beyond

Cohort 2 starts in Month 2–3 once Cohort 1 is in rhythm. The Operating Council picks Cohort 2 from the backlog. Strong candidates based on the strategy sprint:

Content production / localization / creative operations

Why:

• it affects creative, eCommerce, international, and marketing at the same time
• the workflow has already been mapped in meaningful detail
• there are obvious manual handoffs that can be improved quickly
• Josh has already shown the pattern with working Figma plugins
• multiple teams have independently identified content architecture as one of the main constraints on growth
• when content flows, customers in new markets get the full brand experience at launch — not a patched-together subset that undermines the quality Balsam is known for

One decision-support track

Once two workflow cohorts are moving, a decision-support track makes sense. Candidates:

• campaign analytics / marketing decision support
• self-serve analytics on top of the data layer
• a focused channel or retailer operations workflow

Marketing is worth reimagining, but it comes in as a parallel engagement rather than a COE cohort during the founding period. Pascal–Karima conversation runs separately. See the Marketing strategy support section below.

Roles And Commitments

Clear roles will help the COE get started, and those roles can evolve as Balsam's needs and internal capabilities change.

Claire

• sponsor from the business side
• help decide what matters now
• help identify the right cohort participants
• help remove cross-functional blockers when the work needs executive involvement

France

• sponsor from the product / data / technology side
• help align the COE with the Operating Council, the AI Council, and pod reality
• help secure access to the relevant systems, tools, and technical context
• help make sure validated work has a path into implementation
• help reserve a practical path into business-accelerant engineering capacity for COE-validated work

Cohort participants

Cohort membership works in three tiers. Deeper for fewer, lighter for more — so depth can match what the work actually needs.

Core owners (~50% of their time during an active cycle). One to three people per cohort. They drive the work — they are doing the workflow today or directly responsible for it. Fifty percent for a four-to-six-week cycle is what turns a workshop into a shipped change. After the cycle they graduate into alumni status and return to their regular role.

Participants (~20% of their time during an active cycle). Two to four per cohort. They contribute specific workflow knowledge from a producer, consumer, or adjacent angle. Enough time to stay in rhythm and test what gets built.

Adjacent ring (~5%). Five to ten colleagues who use what the cohort produces, book office hours when they hit friction, and coach back into their teams. People being considered for the next cohort also sit in this ring. This keeps the pipeline warm and the COE visible beyond the active group.

Across all tiers, participants should:

• bring the real workflow context, judgment, and day-to-day knowledge
• bring what they learn back into their day-to-day work — the COE should change how they do their existing job, not add a side project on top of it
• test what gets built
• help bring adjacent colleagues into the next cohort

Goose Group

• act as the founding operator of the COE
• facilitate, map, prototype, build, document, and coach
• keep the work moving between sessions
• help turn discoveries into useful plans, prototypes, or handoff material that Balsam can act on
• provide broader fractional VP of AI leadership around the program, not just manage the cohort calendar

Internal long-term owner

This person does not need to be finalized on day one, but they should be identified during the founding run.

The goal is to have a real internal owner by the end of the founding run, even if Goose Group continues to support the program in a lighter way.

What Good Looks Like After The Founding Run

If the COE is working, Balsam should have more than a set of documents.

It should have:

• six or more cohort cycles completed
• each cohort producing a mapped workflow, a working prototype, a validated initiative brief, and trained alumni
• working tools, prototypes, or automations in the hands of real teams
• a backlog of defined initiatives that pods can actually pick up without months of rediscovery
• a growing internal repository of workflow maps, playbooks, and readouts
• alumni from the first cohort who can help bring the next people along
• a named internal owner with a clearer path to run the COE with lighter Goose Group involvement

The main outcome is that Balsam gets better at identifying this kind of work, testing it, and moving it into execution, while also surfacing opportunities that would otherwise stay buried in the organization.

That should include both cost structure improvement and upside potential. Some wins will remove manual work. A smaller number should materially improve how quickly Balsam can grow, launch, and learn. The clearest sign of success is not internal speed — it is that customers in more markets get a better, more consistent Balsam experience, and retail partners get what they need to sell well.

Things To Watch

A few practical headwinds are worth naming early.

Participant adoption

People are busy, and a new program becomes background noise if the first cohort does not produce something people can actually use.

Short cycles help here. The first cohort needs to produce something visible and useful in the first few weeks so people can see how this connects to their work.

Access and approvals

The work slows down if each cycle spends its time waiting for access, approvals, or favors from system owners.

France's sponsorship helps here. Data needs should be scoped before each cycle starts, and blockers should be surfaced in the first sponsor sync rather than discovered halfway through.

Cross-functional ownership

Some workflows will be politically sensitive. Some problems will cross ownership boundaries. Some teams will be protective of their ways of working.

This is one reason the sponsor layer matters. Claire and France can help make decisions and remove blockers when the real friction turns out to be organizational rather than technical.

Long-term ownership

If Goose Group becomes the only group that knows how the COE works, the program stops when Goose Group steps back.

Capability transfer has to happen from day one. Each cycle should leave behind artifacts, trained people, and clearer internal ownership so the work stays inside Balsam over time.

Commercial Shape

A simple commercial structure is a monthly Goose Group retainer with clearly separated extras.

Base retainer

Base retainer: $50K/month for a 6-month founding run.

That is $300K total for the founding period.

The base retainer covers Goose Group acting as the founding operator of the COE.

That includes:

• sponsor cadence with Claire and France
• COE operation and cohort facilitation
• opportunity review and intake support
• Operating Council and AI Council collaboration
• office hours and unblockers
• lightweight prototypes and internal tooling where needed to keep the work moving
• planned internal enablement activities tied to the COE
• building and maintaining the basic internal home for the COE
• the broader fractional VP of AI leadership around the program

This is the core layer Balsam is retaining Goose Group to provide.

Team

• Mike Joyce and Alex Finnemore lead the program week to week

Goose Group's broader expertise — including marketing strategy and creative operations — is available for complementary engagements when the work calls for it.

Framing

This is not meant to be hourly billing. Balsam is retaining Goose Group for time, judgment, and operating support across the founding run.

The useful mental model is to treat the founding run like an incubator for a small number of high-upside bets. The point is not that every opportunity produces the same return. The point is to identify the few opportunities where a relatively small amount of focused work can create disproportionate value.

Examples of the kind of upside this should target:

• faster international market launches because content and localization can actually move — customers in new geographies get the complete Balsam experience from day one
• faster retailer onboarding because product data does not have to be translated manually for every account — retail partners can list and sell sooner with accurate product information
• same-day or near-same-day marketing decisions during peak instead of a ~48-hour lag — holiday shoppers see relevant promotions while they are still buying
• a clearer path for France's team to back validated business initiatives instead of absorbing vague requests

That should show up in outputs such as:

• six or more cohort cycles
• mapped workflows
• working prototypes
• validated initiative briefs
• trained alumni
• a backlog the pods can pick up without months of rediscovery
• an internal owner running the program by the end of the founding run

If the program works well, Balsam will have a better way to identify useful initiatives, test them, and move them into delivery.

If the founding run works well, Goose Group can continue afterward in a lighter advisory role with Claire, France, the Operating Council, and the AI Council while Balsam takes on more of the day-to-day operation internally.

What sits outside the retainer

The retainer covers the COE and Goose Group's leadership around it. A small number of things sit outside:

• travel and on-site costs (billed at cost)
• software and infrastructure costs for tools or sandbox environments (typically small, flagged transparently)

Marketing strategy support through Pascal Staud is available as a complementary engagement on its own terms — see the marketing strategy section above.

Implementation projects that come out of the COE — engineering work, data pipelines, system integrations — are normal business investments. Balsam already makes these decisions through its existing process. Part of Goose Group's value is helping define that work clearly and overseeing the right partners to deliver it. But those costs sit on Balsam's business budget, not on the COE line item. The COE surfaces and validates the opportunity. Balsam decides what to fund.

What Changed

April 14, 2026

Updated based on Claire's response to the updated brief and the follow-up Alex/Mike discussion:

• Governance layered. Operating Council (prioritization) now sits alongside the AI Council (enablement and standards). The "What The COE Is" table, the intake flow, and the Council collaboration section all reflect the two-body model.
• Time model tiered. Core owners ~50% during an active cycle, participants ~20%, adjacent ring ~5%. Deeper for fewer, lighter for more. Replaces the earlier flat 20% framing.
• Product Data named as Cohort 1. PLM-free-2026 context added. Upstream-before-downstream principle made explicit — fixing the source before accelerating downstream retailer-specific work. Kate and Mary as proposed core owners.
• "Three tracks from day one" dropped. Cohort 1 runs solo; Cohort 2 is picked by the Operating Council from the backlog in Month 2–3. Cohort candidates remain content production and a decision-support track.
• Fractional AI leadership scope broadened with sunset. Goose Group acts as connective tissue across the broader AI landscape (Pablo, Bernie, training) during the founding period. Handoff to an internal owner by Month 5.
• Month 1 / 2 / 4 / 6 outputs added as concrete milestones the Operating Council can hold the COE to.
• In-season (Sept–Jan) approach added — smaller cohorts, consolidation focus, alumni-led continuation.

April 6, 2026

Updated based on Claire's questions and our discussion:

• Time commitment clarified. 20% is structured contact time with Goose Group. The goal is for that time to change how cohort members do their entire job — not add a side project. Adjacent collaborators and next-cohort candidates get bookable office hours.
• Marketing strategy separated from COE. Pascal is part of Goose Group and will be involved in COE work where relevant. A marketing strategy engagement for Karima's function is a separate, complementary path — we recommend an introductory conversation, independent of the COE launch.
• Team and commercial structure simplified. The retainer covers Alex and Mike. Travel and software are the only extras. Implementation projects that come out of the COE are normal business investments on Balsam's budget — Goose Group helps define the work and oversee partners, but the cost isn't a COE line item. Complementary engagements (like marketing strategy) are scoped on their own terms.
• Cohort selection sharpened. Added that the COE is not a technical support function — cohort members should be people ready to build and experiment.

Questions To Align On

These are the remaining questions worth discussing together.

1. Does this updated structure match what you want the COE to do?
2. Roster for Cohort 1 — confirm Kate and Mary as core owners, identify the right merchandiser and consumer participants, confirm Clint Borrill's availability.
3. Who is the Operating Council liaison for the monthly COE readout, and what is the best meeting cadence to fit the Council's existing schedule?
4. What does Balsam need for procurement on the Goose Group agreement?
5. Engineering partner estimate — what budget range does Claire want to plan against for the first follow-on build project?

Practical Next Step

If the operating model feels right, we would like to start moving:

• Build the COE home. We start on the internal microsite now — program structure, how to submit ideas, readout archive. Something real to show at the late-April offsite rather than talking about it abstractly.
• Simple agreement. Goose Group is a Dutch entity with US bank accounts. We are thinking a short, clean contract. What does procurement look like on the Balsam side?
• Pascal-Karima intro. Happening separately — an introductory conversation, no commitment.

Appendix: Implementation Partners And Specialist Support

If a validated COE problem needs deeper engineering work or faster delivery than the core COE team should provide, Goose Group can help Balsam bring in additional implementation capability.

That may include:

• Balsam delivery teams
• trusted partners such as Spatial Edge or Substantial
• Goose Group contractors
• support provided directly by technology vendors

Goose Group's role in those situations is to help define the work clearly, compare the available options, and make sure the delivery team stays connected to the workflow problem that created the need in the first place.

These engagements would be scoped separately from the Goose Group retainer. As a reference point, a focused external data or integration engagement is often 4-6 weeks and may fall in the $80-120K range depending on scope.

Spatial Edge as a specific option

One specific partner option is Spatial Edge.

Spatial Edge is a data and AI engineering company focused on production analytical systems, data pipelines, ML models, and system integrations. They are most useful when a validated COE problem turns out to be a real data engineering or integration problem rather than an advisory or prototyping problem.

For Balsam, the problems most likely to fit that profile are:

• PIF-to-retailer format translation automation for Bob McDonald's team, where the problem is a repeatable transformation pipeline rather than better prompting
• Reference SKU matching for inventory planning, where attribute-based matching could replace intuition-heavy manual work
• PIF version control and change management, where the current Excel and email process needs a governed production data layer
• Figma-to-CMS or adjacent system integrations where the real work is API-level integration between tools

If Goose Group maps and validates one of those problems through a cohort cycle and it is clearly an engineering problem, Goose Group can bring in Spatial Edge and manage the engagement. Balsam does not need to create a separate vendor relationship unless it wants to.

As a directional reference point, a focused Spatial Edge engagement is often 4-6 weeks and roughly $80-120K, scoped separately from the Goose Group retainer.