Custom business software for operations that no SaaS product understands.
I build production applications around the workflow the business actually runs: admin surfaces, customer flows, integrations, ledgers, inventory, documents, and deployment.
For owner-led businesses whose operational truth lives in spreadsheets or a stack of disconnected SaaS tools, and whose process logic is too specific, valuable, or interconnected for off-the-shelf software to model honestly.
Who this engagement is for
Who this is for
- $5M to $50M owner-led businesses whose operational truth lives in spreadsheets or a stack of disconnected SaaS tools.
- Process-driven companies (custom manufacturing, luxury retail, regulated industries, professional services) where the workflow is the product.
- Teams that have outgrown low-code or off-the-shelf platforms and can no longer get the business logic they need to ship.
- Owners who want to own the application end to end (code, infrastructure, data) instead of renting a SaaS that owns the workflow.
- Businesses sitting on a 10+ year legacy operational stack where staff productivity depends on workflow patterns the replacement must preserve.
- Companies where a single failure mode in the operational stack would cost more than the build.
Who this is NOT for
- Greenfield SaaS startup ideas. Custom application development is for established operations, not idea validation.
- Generic SaaS-style apps where existing platforms (Salesforce, Monday, etc.) already model the workflow well.
- Teams that want a custom build but cannot commit to operating it after launch.
- Buyers who want a project quote in 30 minutes. Real custom application work starts with a paid Architecture Sprint.
Where this engagement starts
The problem
Most custom-software conversations start the same way. The business has a workflow that genuinely matters: how an order moves to fulfillment, how a custom quote becomes an invoice, how inventory commits to a deposit, how staff actually do the work. The workflow lives in a stack of spreadsheets, a 15-year-old internal database, a homegrown departmental app, or four SaaS tools wired together with copy-paste discipline. It works. It also breaks every quarter, requires a senior person to operate, and blocks every reasonable scaling decision the owner wants to make.
The wrong fix is a generic SaaS replacement that does 60% of what the business actually does and tells staff to change the other 40%. The wrong fix is a low-code platform that lets a non-engineer build the V1 and leaves the company with a maintenance liability nobody can read.
The right fix is a platform-native application designed around the actual workflow, owned end to end by the client, runnable on their AWS account, and built well enough that a future developer can read the code, find the data model, and ship a change without breaking the operation.
The approach
How this engagement runs
Treat the existing operational workflow as a first-class requirement. Build the smallest custom platform surface that absorbs the operation into a sovereign, owned application without losing the workflow that makes the business work today.
Operational discovery
Document the actual workflow: who does what, in what order, with what data, against what constraints. Interview the staff who run it. Read the spreadsheets and the legacy database. Find the rules that aren't written down anywhere.
Platform architecture
Design the Next.js + Payload CMS + PostgreSQL + AWS architecture that owns the workflow end to end. Multi-account AWS Organization, ECS Fargate, RDS Postgres, S3, CloudFront. SQS + Lambda for any operation that should not block a request.
Migration-aware build
Ship the application in slices that can run alongside the legacy stack. Real data flowing through real workflows long before cutover. Production-clone rehearsals for every meaningful migration step.
Zero-downtime cutover
Move data with dry-run and apply scripts, audit reports, ledger reconciliation where it applies, and rollback criteria defined before the cutover starts. Production cutover runs against a clone first.
Operate and evolve
CloudWatch alarms, observability discipline, GitHub Actions CI/CD with STS-only credentials, and an Architecture Decision Record trail the team can inherit. Optional Fractional CTO retainer for ongoing roadmap work.
Outcomes
What you walk away with
- $5M+ ARRAnchor delivery: custom commerce platformSource: Fine's Gallery production platform (Conti Digital build, ongoing Fractional CTO retainer)
- Zero downtimeProduction cutoverSource: Fine's Gallery custom platform launch (June 2025)
- 20+ yearsLegacy operational data migrated cleanlySource: Fine's Gallery legacy-system retirement (28,000+ historical invoice records into platform-native invoicing)
- ~3,800 LOC PostgreSQLConfidential project: regulated registry ledgerSource: Confidential energy-sector certificate registry (Lead Architect, 6 months)
- 1,000+ integration testsTest discipline (anchor + confidential combined)Source: Real-database test suites against PostgreSQL containers (no mocks)
- 223,000+ LOC TypeScriptSolo delivery scopeSource: Fine's Gallery production codebase (single-architect, 19 months to first cutover)
Concrete deliverables
- Custom application architecture: Next.js 16 + Payload CMS 3 + PostgreSQL on RDS + ECS Fargate + S3 + CloudFront, deployed as a sovereign AWS Organization the client owns end to end.
- Operational workflow audit + Architecture Decision Records documenting the data model, business rules, and the legacy patterns that needed to be preserved.
- Event-driven async backbone (SQS + Lambda + EventBridge) for any operation that should not block a request: email dispatch, document generation, third-party syncs, scheduled jobs.
- First-party integration adapters with typed contracts (Zod-validated), retry semantics, and dedicated tests. No off-the-shelf plugins.
- Migration scripts with dry-run + apply modes, audit reports, ledger reconciliation, rollback criteria, and production-clone rehearsals.
- Staff operating surface (Payload admin) that mirrors the workflow patterns staff already use, with the platform limits underneath them removed.
- Observability: CloudWatch alarms, custom metrics, Playwright E2E coverage on critical user paths, and incident-readiness runbooks.
- GitHub Actions CI/CD with GitHub OIDC trust to AWS (STS-only, no long-lived credentials), Terraform-managed infrastructure, and handoff documentation a future developer can pick up.
FAQ
Frequently asked questions
Agencies sell teams and timelines. Custom application work needs an architect who is also accountable for shipping the code and operating the result. Solo delivery by an AWS Solutions Architect Professional, ongoing through a Fractional CTO retainer if the engagement scopes that way.
The economic incentive is to own the architecture decision and the implementation that gates it. No phase handoffs, no agency-to-agency knowledge loss, no plugin marketplace dependencies.
Yes. The safest path is parallel operation: real data flowing through real workflows in the new platform, validated against the legacy system, with cutover deferred until the new platform's behavior is proven against actual production data. The Fine's Gallery cutover ran with zero operational downtime and zero loss across 20+ years of legacy operational data.
Next.js 16 + Payload CMS 3 + PostgreSQL + AWS (ECS Fargate, RDS, S3, CloudFront, SQS, Lambda, EventBridge, KMS, IAM via GitHub OIDC). Terraform-managed infrastructure. The same stack the consultancy's own site runs on; the same stack used for the $5M+ ARR Fine's Gallery commerce platform.
Stack choice is opinionated for ownership: nothing in the stack is hosted by a vendor that holds the keys to the business.
Yes. The application runs in your AWS Organization. The code lives in your GitHub. Terraform state lives in your AWS account. CI/CD pipelines run under GitHub OIDC with no long-lived AWS keys anywhere. You can revoke any access I have without taking the platform down.
Three to six months for an initial production cutover, depending on workflow complexity, data migration scope, and integration count. The Fine's Gallery platform shipped its first production cutover in that window and has continued to evolve through a Fractional CTO retainer since. Smaller targeted applications can ship in less time; complex multi-workflow operations (regulated registries, multi-tenant platforms) take longer.
Custom application builds from $80K. Final scope depends on workflow complexity, data migration, integration count, and the staff operating surface's depth.
Discovery happens in a paid Architecture Sprint ($15K, 3 weeks) so the build estimate is grounded in your actual operation, not a generic intake form. Architecture Sprint output is a written architecture decision document — current-state risk register, target architecture, migration plan, and cost model — that you can run the build against with anyone, including yourself.
Solo delivery by Peter T. Conti, AWS Solutions Architect Professional + Associate. Lead Architect on a national-scale energy-sector certificate registry. Principal engineer on the $5M+ ARR Fine's Gallery commerce platform (223,000+ lines of TypeScript over 19 months to first cutover, ongoing as a Fractional CTO retainer).
No agency, no offshore, no junior team. The same person who scopes the engagement ships the code and operates the platform.
Optional Fractional CTO retainer ($15K/month, 3-month minimum) covers architecture review, vendor decisions, AWS posture, roadmap sequencing, and direct implementation when the critical path needs it.
If you have an internal team or an agency that operates the platform, the retainer scopes around them as senior architecture oversight rather than replacing them.
Related work
See this in production
Fine's Gallery Platform Modernization
Solo-built end-to-end commerce platform for a luxury marble and stonework gallery. $5M+ annual revenue running on a sovereign, client-owned AWS Organization. Zero-downtime production cutover. 20+ years of legacy operational data migrated into the new platform without loss. Marketing dominance on the new platform supports $500K+/month in commerce volume. Systematic removal of vendor lock-in across SaaS commerce, payment, and warehouse tooling. Now operates on a Fractional CTO retainer.
Peter T Conti Consulting Platform
The consulting practice's own site, built as the reference architecture for what the practice sells. Same Next.js + Payload + Postgres + ECS Fargate + Terraform stack as the anchor engagement. STS-only IAM, no long-lived credentials anywhere. Cost-tuned to run for a small fraction of the naive default AWS footprint, with full editorial workflow operated as code.
Read more
In-depth on this topic
The SaaS Ceiling: Why High-Ticket Brands Need Sovereign Commerce Infrastructure
For high-ticket ecommerce, you need infrastructure built around your business model, not the other way around.
Stop Renting Your Revenue Engine: The Sovereign Infrastructure Blueprint for High-End E-Commerce
The goal is straightforward: you own the platform that owns your revenue.
Building a Google Merchant Center Sync Engine Plugin for Payload CMS
Merchant Center sync engine that powers $2-400k+/month in Google Shopping revenue for a 5,400-product luxury ecommerce catalog.
Building a Production-Grade GA4 Analytics Plugin for Payload CMS
How I built a production-grade GA4 analytics plugin for Payload CMS, and the architectural decisions behind making it reliable at scale.
Credentials and Fit
Directly relevant experience for high-trust delivery
I lead architecture and implementation personally, with technical depth that spans product, cloud, and execution operations.
AWS Certified Solutions Architect - Associate
Cloud decisions are grounded in secure, cost-aware architecture with practical production tradeoff management.
Dual technical foundation
B.S. Computer Science plus B.S. Chemistry with undergraduate research and scientific presentation discipline.
AWS Certified Solutions Architect - Professional
My ability to use AWS to solve complex business requirements has been demonstrated at the highest enterprise level.
Custom Product Engineering Inquiry
Share your product vision, required technical capabilities, and target timeline for MVP or V2 so I can scope the fastest viable implementation path.
Pressure-test the platform decision before you commit budget.
Schedule a complimentary 30-minute consultation to align on objectives, stress-test your architecture, and leave with a concrete set of recommendations. No obligation, no sales pitch. Just actionable technical guidance.