Become the Ollie of Asia —
before someone else does.
BOM BOM has the product, the ops, and the compliance moat that Butternut Box, Tails.com, The Farmer's Dog and Ollie spent years building. What it doesn't have is the modern, AI-first digital experience that lets those brands scale. This is the 12-month plan to close that gap — and leapfrog them.
A real business,
with real operational muscle.
Before we talk about what to change, let's be clear about what we're protecting. BOM BOM has earned a category position in Singapore and Korea that most Western pet-food startups would envy. The rebuild exists to accelerate it — not to reinvent what already works.
Nutritional Support
Complimentary after-sales support by a professionally certified nutrition team. This is the human layer Western competitors outsource to chatbots.
Ethically Sourced
SFA-approved, free-ranging, antibiotic-free, hormone-free prime cuts. A trust moat no Western brand has inside Asia.
Scientifically Balanced
Comprehensive meals, customised and portioned per pet. The engine is there — the AI layer will make it 10× smarter.
Cold Chain Management
Vacuum-packed, fresh pet food delivered weekly across Singapore. Ops discipline most D2C brands still can't match.
You're already feeding 10,000 pets a month — on infrastructure that wasn't built for it.
That's real revenue, real customers, real ops. And it's running on a stack that makes every new feature painful. The rebuild is how you turn that same team into a 50,000 meals/month machine — without burning them out.
This isn't a website rebuild.
It's an operations platform rebuild.
The brief describes a website and admin. What we see in the codebase is something far bigger: a full operational platform with hardware in the loop, two country-level businesses, and live revenue depending on it every day. The way we scope, price, and de-risk this project has to match that reality.
What BOM BOM actually is, from our read of the codebase.
Ten .NET projects. Over 116 database tables. Two customer-facing websites (SG + KR). An admin backend running production lines, QC tablets, inventory handhelds, and delivery routing. A mobile app API. A custom headless CMS. Six background jobs — Mailchimp sync, Intercom sync, production export, rotation ordering, onboarding, and order-label generation. Stripe payments. Twilio SMS. SendGrid email. Firebase. Google Maps. Hangfire queuing. Label and QR printing. Golden-sample history. Member pets. Allergies. Nutrition recommendations. Meat types. Delivery drivers and slot routing.
This is not a website with an admin attached. This is a vertically integrated food-production and subscription-commerce platform that happens to have a website on the front.
Which means the real brief is not "rebuild our website" it is "rebuild the platform that runs our business, without interrupting the business."
What we see today —
and what's quietly holding you back.
Based on the partial codebase, the architecture diagram, and the symptoms you've described (slow feature delivery, brittle integrations, agency friction), here is our honest read of where the platform stands today and why moving fast is painful.
The problem isn't the language —
it's the architecture and versions.
Every one of the symptoms you describe traces back to a small number of compounding root causes. We've grouped them by severity so you know what's actually dangerous vs. what's just dated.
.NET Core 3.1 is end-of-life
The entire stack targets netcoreapp3.1, which Microsoft stopped supporting in December 2022. No security patches. No modern language features. Not a rebuild concern — a compliance and risk concern today.
Vue 2 on the frontend — also EOL
Both websites ship vue@^2.6, vuex@^3.6, and vee-validate@^3.4. Vue 2 reached end-of-life December 2023. Every new feature is being built on deprecated tooling.
Single EC2 instance + Windows filesystem for uploads
Image paths are hardcoded to S:\Websites\... drives. This is why scaling horizontally is impossible, why failover can't exist, and why every deploy risks the whole system. This is the bottleneck.
Monolith with tight coupling
Website, Admin, and AppAPI share one data layer and one repo. A change in one module risks all three. Deployments are all-or-nothing. This is what makes feature delivery slow — not the language, the coupling.
Country-separation via connection-string toggle
SG and KR run the same codebase pointed at different databases by switching a connection string in appsettings.json. Fragile, makes per-country releases risky, and makes divergent country features nearly impossible.
Legacy frontend toolchain
Laravel Mix, AdminLTE, Bootstrap 4, jQuery, Select2, Slick, Parsley, Tooltipster. Modern browsers and modern hires expect modern tooling. Recruiting developers into this stack is a headwind.
Six background jobs, all separate EXEs
Mailchimp, Intercom, Production export, Rotation order, Onboarding, Order-labels. Each is a standalone Windows executable triggered from disk paths — not a modern queue. Fixable, but an architectural liability.
Hardware-in-the-loop complexity
Label printers, QC tablets, handheld inventory scanners, production-line devices, rack QR codes. Real-world integrations that are easy to break and hard to test without hardware present. This will dominate rebuild risk.
// No CI/CD visible in the repo. Manual deploys.
// No service boundaries — everything shares BomBom.Data + BomBom.Repo.
What the West has built —
and where BOM BOM can leapfrog them.
We looked at the four biggest global pet-food subscription platforms. Three are UK/US consumer darlings. One was acquired by Nestlé Purina. Here's how BOM BOM stacks up today, where they've pulled ahead, and where the rebuild puts BOM BOM on a trajectory to become the category leader in Asia.
The gap isn't product. BOM BOM's sourcing, nutrition, and cold-chain discipline are already at or above these brands. The gap is digital experience — onboarding, personalisation, self-serve subscription management, AI, and app-first behaviour. That's exactly what a modern rebuild fixes.
Butternut Box
UK · B-Corp315M+ meals served · 4.8 Trustpilot · 43k+ reviews
Won by making fresh food feel obvious. "Switch to fresh. 4 in 5 never go back." Copy is witty, transparent, human — and the ingredient story is told better than anyone else in category.
Tails.com
UK · Purina-owned1M+ kibble combinations · Vet + engineer built algorithm
Won on algorithmic depth. Their pitch is "your dog's food, literally not fed to any other dog." The personalisation moat was strong enough that Nestlé Purina acquired them to own the D2C future of pet nutrition.
The Farmer's Dog
US · #1 ranked100M+ meals · USDA kitchens · ~$251 / month AOV
Won on trust + delight. Pre-portioned per-dog labeled packs. A quiz that feels like a conversation ("she weighs ___ lbs"). Fresh-food framing pitched directly against kibble. Premium positioning, popular-price point execution.
Ollie
US · AI-first20M+ data points · Patent-protected AI vet imaging · LiDAR body scans
Winning on AI. Upload a photo of your dog's stool, coat, teeth, or weight — their AI analyses it and tunes the meal plan. Smartphone LiDAR scans replace guesswork on body condition. Microbiome insights. This is the blueprint BOM BOM should study and beat.
BOM BOM doesn't need to copy them.
BOM BOM needs to leapfrog them.
Every one of these brands has gaps. None has what BOM BOM already has — SFA-certified sourcing, a certified human nutrition team, multi-market ops discipline, vertical production. Pair that with a modern AI-first digital experience and BOM BOM becomes unreachable in Asia and exportable to the world.
- AI Mode · Conversation-first ordering. Nobody in category has built this yet.
- Photo health checks · Beat Ollie on Asian-breed coverage and Asian-vet partnerships.
- Multi-pet households · Per-pet labelled packs + one cart. FD does labels — no one does the orchestration.
- Asian-first nutrition science · Breeds, climate, dietary norms the West ignores.
- Ops-grade transparency · Show customers the golden-sample QC photos. Nobody does this.
- Voice & brand · Own a voice Asia hasn't heard yet in pet food.
Five places where a bad plan would hurt you.
These are the specific risk vectors that make this engagement unlike a normal website rebuild. Our phasing, pricing, and governance are all designed around these.
Live operations cannot stop
BOM BOM is processing real orders, routing real drivers, and feeding real pets every day. Any rebuild has to run parallel to production, with strangler-fig cutovers — never a "big bang" replacement.
Business logic lives in code, not docs
Meal plan formulas, pet allergy rules, nutrition rotations, combo-product logic, rack rotation, order-label generation — none of this is documented, all of it is in .NET. Discovery must capture it before it can be rebuilt.
Hardware integrations are brittle
Production tablets, QC tablets, handheld scanners, label printers, rack QR codes. These integrations are usually underdocumented and require being on-site to test. We treat them as their own workstream with their own risk budget.
Two countries, two markets, one platform
SG and KR are not the same business — different currencies, languages, suppliers, carriers, delivery rules, and compliance expectations. The rebuild needs to encode that difference into architecture, not paper over it with connection strings.
Data migration is the silent killer
116 tables. Decades of subscription history, pet profiles, allergy records, order archives, vendor relationships, production audit trails. Migration has to be reversible, idempotent, and re-runnable — not a one-shot cutover.
Codebase quality is still unknown
We've seen a partial codebase. Before committing to fixed scope, we need full-repo access, read access to production-adjacent environments, and interviews with the incumbent team. Without that, any fixed price is a fiction.
A modular, API-first platform —
built to move at the speed of the business.
The future-state stack is not about trading .NET for TypeScript as a religious decision. It's about moving from monolith → services, single-server → cloud-native, and EOL frameworks → long-supported ones. The recommendation below is what we are fastest and most opinionated in.
Architecture is split into six layers, each independently deployable, each with its own release cadence. The customer-facing web and mobile teams can ship every week; the production-line team can ship without touching the eCommerce team; infra changes can roll out without product risk.
BOM BOM Platform v2
// CI/CD via GitHub Actions with preview environments per PR.
// Observability: CloudWatch + Sentry + Grafana. Zero-downtime deploys from day one.
// CMS note: Sanity is used only for marketing pages & stories — never for order, production, or inventory data.
The legacy stack
The modern stack
All orders in one ask —
BOM BOM AI Mode.
A conversational, AI-native ordering experience. Describe your pet once. BOM BOM AI Mode generates the perfect meal plan, calculates portions, picks proteins around allergies, and upsells intelligently — all in a live, streaming cart. This is a working prototype. Try it.
AI-generated meal plan
Why this plan
BOM BOM · AI Cart
5 itemsSubscription is the business.
We're rebuilding it like the crown jewel it is.
Recurring revenue is not a checkout feature — it's the financial engine of BOM BOM. Every pause, every skip, every failed payment, every plan change, every dunning email is either strengthening or leaking LTV. The legacy stack treats these as edge cases. We treat them as the core product.
Pause · skip · swap
Self-serve lifecycle controls. Every action audit-logged. Churn telemetry on every click.
Multi-pet households
Per-pet meal plans, per-pet labelled packs, one subscription, one cart, one invoice.
Plan rotation & auto-adjust
Recipe rotation driven by pet profile. Auto-scale portions as weight or age changes.
Failed-payment recovery
Smart retry, alt-card prompts, in-app restore, targeted email sequences. Recover the revenue most platforms leak.
Cancel flow with save offers
Dynamic offers based on why they're leaving (price, fit, lifestyle). A/B testable. Proven churn-reduction.
Retention telemetry
Every customer scored on churn-risk. Triggers proactive nutritionist outreach before cancellation.
Referral engine
Two-sided rewards. Built into subscription management. Tracking, attribution, anti-abuse — in-house.
Gift subscriptions
Gift-a-month, gift-a-plan, gift-a-nutritionist-session. New acquisition channel most competitors ignore.
Every touchpoint — engineered.
Annual LTV lift
Target uplift from better retention + rotation + save offers post-rebuild.
Involuntary churn
From failed-payment recovery + smart retry. Typical industry win after this rebuild.
Feature velocity
New subscription experiments per month vs. the legacy stack.
Self-serve any action
Pause, skip, swap, change pet — zero-support-ticket target.
AI isn't a feature —
it's the operating layer.
The rebuild gives us somewhere to put intelligence. Every one of these capabilities is enabled by the new architecture and blocked by the legacy one. Ollie has three of them. Nobody has all nine. BOM BOM can.
AI Meal Plan Engine
Replaces hand-coded rules with an algorithm trained on BOM BOM's own data + breed science. Scales to millions of combinations — the Tails.com moat, but Asia-first.
Photo Health Check
Upload a photo of stool, coat, teeth, weight → AI flags issues → nutritionist reviews. Beats Ollie by adding Asian-breed coverage and local vet partnerships.
Smartphone Body Scan
LiDAR / computer-vision based body-condition scoring. No more guessing weight — the phone measures the dog. Auto-adjusts portions.
AI Demand Forecasting
Predicts next week's production demand per recipe, per region, per customer segment. Cuts waste and prevents stockouts in the factory.
AI Nutritionist Copilot
Augments your certified team. Drafts responses using BOM BOM's own knowledge base — team approves & sends. 5× the throughput without outsourcing.
AI Route Optimisation
Cold-chain delivery route planning that beats hand-rostering. Accounts for traffic, time windows, capacity, cold-chain duration.
AI QC Photo Verification
Production-line photo uploads scored against golden samples by a vision model. Catch deviations the human eye misses.
Churn-Risk Scoring
Every customer scored on likelihood-to-cancel daily. Fires proactive nutritionist outreach 2 weeks before a typical churn event.
AI Content & Copy
Product pages, blog, email, ads, landing pages — generated on BOM BOM's brand voice, reviewed by your team. Multi-language SG/KR from day one.
Recipe R&D Assistant
Models the nutritional impact of new recipe ideas before they hit the factory. Lets your nutrition team iterate 10× faster on new products.
Personalised Upsell Engine
Right supplement, right moment, right price. The engine behind AI Mode's live upsells. Compounding AOV week over week.
Pet Photo Diary AI
Customer-submitted pet photos auto-tagged, health-flagged, turned into social proof + nutrition feedback. Owners love seeing progress over time.
From 10,000 meals / month
to 50,000 and growing.
Here's the roadmap — month by month, quarter by quarter. What ships, what it unlocks, and what it means for the top line. This is the view we'd walk through with your board.
Target meal volume · 12-month projection
Foundation
- Discovery & architecture locked in Month 1
- Core API + domain services begin
- Data-migration strategy agreed
- Baseline ops + velocity stay stable
Core shipping
- New admin running parallel to legacy
- S3 + CDN replaces S:\ paths
- Subscription engine refactor starts
- First AI capability prototypes
Customer-facing
- SG + KR sites rebuilt on Next.js
- AI Mode public beta
- Factory tablet apps rebuilt
- Subscription LTV features live
Scale + AI
- Legacy decommissioned
- AI health check + photo diary live
- 50k meals/month target
- Year-2 growth partnership
12 months.
Four phases. One outcome.
A fixed-scope, fixed-fee 12-month engagement designed around how BOM BOM actually runs — live ops can't stop, production can't slip, and the team needs velocity, not ceremony. Each phase overlaps the next; every week ships something testable.
Discovery & Architecture
Month 1. A senior team lives in your codebase, on your production floor (SG + KR), and with your data. We exit this phase with a locked architecture, a risk register, a team plan, and the design system for the rebuild — all signed off.
Core System Rebuild
Months 2–6. The engine. NestJS domain services, Next.js admin, Postgres migration, event bus, S3 uploads, CI/CD, observability. New admin runs in parallel with legacy from Month 4. By Month 6 your ops team is using the new admin daily.
Customer, Apps & AI Mode
Months 6–10. Everything a customer touches. SG + KR websites rebuilt on Next.js, subscription engine shipped, BB Store modernised, factory tablet & handheld apps rebuilt — and the first two AI capabilities shipped including AI Mode public beta.
Launch, Hypercare & Growth Ignition
Months 10–12. Cutover, hypercare, growth. We kill the legacy stack, run 60 days of daily-standup hypercare, then stand up the Year-2 growth engagement (AI capabilities 2–10, multi-market expansion, performance). You end the year with a platform you can compound on.
$500,000 USD.
Twelve months. Fixed scope. Fixed fee.
No ranges. No "indicative." One number, one timeline, one outcome. The scope is detailed above, broken into four phases, with milestone payments tied to what we ship. Year 2 (the growth retainer) is a separate engagement — not baked into this price.
The BOM BOM Platform Rebuild
A single 12-month fixed-price engagement. Four phases, overlapping. Milestone payments tied to delivery. Senior team for the whole engagement — no bait-and-switch with junior hand-offs.
- Phase 1 · Discovery & architecture · $40k
- Phase 2 · Core rebuild · $220k
- Phase 3 · Customer, apps & AI Mode · $180k
- Phase 4 · Launch & hypercare · $60k
- Milestone payments · no upfront lump sums
- Weekly demos · biweekly production releases
Year 2 — Growth Partnership
Separate engagement starting Month 13. The team that built your platform keeps running it — shipping AI capabilities 4-10, multi-market expansion, performance engineering, and 24/7 on-call. This is how 50k meals/month becomes 150k.
- Dedicated senior pod · 3-5 engineers
- Biweekly production release cadence
- AI capability delivery roadmap
- Multi-market expansion (3rd country ready)
- 24/7 monitoring + on-call rotation
- Quarterly strategy with SD leadership
Payment schedule — $500,000 over 12 months.
// Milestone-based · ops-safeWhat could go wrong —
and what we'll do about it.
We'd rather tell you the risks now than discover them at month five. These are the real ones, ranked by likely impact on timeline and budget, with our mitigation approach for each.
Codebase quality is unknown
HighWe've reviewed a partial codebase. What we haven't seen — incumbent commit quality, test coverage, infrastructure scripts, deployment pipelines, undocumented integrations — could shift the Phase 1 estimate materially.
Hardware-in-the-loop integrations
HighLabel printers, QC tablets, scanners, production-line devices — easily the highest-risk workstream. Undocumented protocols, vendor-specific quirks, and no way to test without being on-site. This is where schedules slip.
Data migration complexity
High116 tables, two databases, live subscription and order history, pet and allergy records, vendor relationships, production audit trails. One botched migration could create tax, compliance, or customer-trust problems that take years to live down.
Business logic trapped in code
MediumMeal plan formulas, allergy rotation, golden-sample QC rules, combo-product logic — none of this is documented. The only sources of truth are the .NET code and the people who wrote it.
Incumbent agency handover friction
MediumEven with goodwill, incumbent teams under-cooperate when they feel displaced. Slow repo access, delayed answers, missing infra credentials — all common and all schedule-impacting.
Scope creep from live operations
MediumLive platforms keep evolving during a rebuild — new promos, new markets, new compliance asks. Without discipline, "just one more thing" compounds until the rebuild never finishes.
Country-specific compliance & payments
Low – MedSG and KR have different tax rules, different invoicing requirements, different carrier norms, and different data-residency expectations. Underestimating this turns Phase 2 into a mini-Phase 1.
Team capacity on your side
LowRebuilds fail when the client team can't carve out time to review, decide, and sign off. A senior product owner and an ops lead need to be available on a weekly cadence — not heroically, but reliably.
You don't need a bigger agency —
you need a sharper one.
Your current friction isn't a symptom of your agency being small. It's a symptom of them being slow, layered, and stuck on a stack that can't keep up. Here's what working with us looks like in practice.
A senior team that ships — not a PM with a Gantt chart
Every engagement has real senior engineers writing code from week one, not a layer of account management. Our typical feature cadence is biweekly. On Phase 3, you'll feel the difference within a month.
We run the platform like it's ours
We don't ticket-and-chuck. We own the platform end-to-end: architecture, infra, deploys, monitoring, incident response. If something breaks at 2am, it's on us to fix — not on you to raise a case.
Small enough to move fast. Big enough to do it right.
We're a 30+ person senior-heavy team in Ho Chi Minh City. Same time zone as SG. 4-hour overlap with KR. We scale pods up and down as the roadmap needs — no seat-warming, no bench tax.
We are not a Razor-replacing shop — we are an architecture partner
Most rebuild projects fail because the new thing looks like the old thing with a fresh coat of paint. Phase 1 (Discovery) exists because we treat architecture as the product, not the output.
Experienced with live, revenue-critical systems
We've delivered eCommerce, subscription, booking, and ops platforms for clients across four continents. We've never broken a revenue-critical system on cutover — and we've done more than twenty.
85% client retention — because we treat partnership as the product
Most of our clients have been with us 4+ years. The agencies who churn their clients aren't solving the hard problem. We stay because we deliver — not because we're locked in.
Saigon Digital have been fantastic throughout the entire website development process. They have been incredibly responsive, professional, and creative. The final product exceeded our expectations.Harry Peisach CEO · Ski.com
12 months.
$500k. Let's build it.
Kickoff Month 1: architecture locked, design system signed off, team on the ground. By Month 12, BOM BOM is the AI-first pet-food platform the category didn't see coming. Fixed scope. Fixed fee. No surprises — only outcomes.
Book a 30-min walkthrough →Or reply to Nick directly · nick@saigon.digital