On July 14, 2026, Microsoft ended support for InfoPath, SharePoint Server 2016 and 2019, and SharePoint Designer — the same day, with no extension and no migration tool. Marcus, Kinetech's AI Mendix developer, reads your forms, workflows, Access databases, and VB6 source directly — and rebuilds them as modern, governed Mendix applications, with evidence for every release.
01 · The clock
Between April and July 2026, Microsoft retired the entire departmental-app toolchain a generation of business processes was built on. This isn't a roadmap warning; every date below is already behind us.
The honest framing: the risk of waiting stopped being a roadmap preference on July 14. It is now a security-exposure line item — unpatched farms, unrunnable workflows, and forms that no longer render.
02 · The unlock
Every one of these technologies is machine-readable — the form definitions, the schemas, the code, the security model. Marcus ingests the actual artifacts, not a questionnaire, and derives a complete functional specification you review before anything is rebuilt.
| What you have | What Marcus extracts | What it becomes in Mendix |
|---|---|---|
| InfoPath form .xsn | The .xsn is an archive of XML: views, rules, validation, conditional formatting, and data connections are all declarative in its manifest | Pages with validation microflows; submit-to-SharePoint chains become native Mendix workflows |
| SharePoint Designer workflow | Declarative approval chains, notifications, and field updates over list events | Mendix workflow engine: user tasks, escalations, notifications — with an audit trail |
| SharePoint lists | Schema and data via the REST API; Microsoft's own assessment tools (PnP M365 Assessment, SMAT) inventory the estate | Domain-model entities with associations — and the historical data migrated with them |
| Access database .accdb / .mdb | Full schema, relationships, and saved queries via the database engine; every VBA module and form definition exported to text | Entities and associations on a real server database; queries become datasources; VBA logic becomes microflows |
| VB6 application .vbp / .frm / .bas | Plain-text source: every form layout, event handler, and module — typically over a SQL Server backend that carries the real data model | Pages and microflows over the same SQL data, with modern authentication and roles |
| Excel/VBA tools | Sheet structures as tables, formulas as business rules, VBA via the same extraction path as Access | Governed multi-user apps — the spreadsheet's logic, without the spreadsheet's copies |
The extraction produces a written specification — data model, page inventory, business rules, and a role-permission matrix in plain English. You approve the spec before a line of the model is written. That's the difference between AI-assisted migration and AI guessing.
03 · How a migration runs
04 · The date
Here is the commitment we will put in writing. The assessment does not end with a range — it ends with a date, sized from your actual extracted complexity. Miss that date and the assessment fee comes back. We can promise a date because we never quote a calendar before reading your source.
The assessment names the date. Miss it, and the assessment fee comes back.
That is a real term, not a slide. It is affordable to us for one reason: the clock below starts only when a complete artifact pack is in our hands, and it pauses only for your own approval gates.
| One form set plus its workflow and lists | Complete artifact pack → UAT-ready |
|---|---|
| Single form — one .xsn, one workflow, ≤5 lists | 10–15 business days |
| Form set — 3–8 forms, Designer workflows, 5–20 lists, 1–3 integrations | ~21 days typical |
| Departmental portfolio — 8+ forms, nested or state-machine workflows, Forms Services, custom code | 6–10 weeks |
| Each further form set in the same portfolio | 40% faster |
Sizing model, not a price list — the assessment reads your actual source and returns one date. Bands assume a single rebuild boundary agreed at the scope gate.
The date is only honest if the inputs are complete, so here is the whole list up front. Nothing on it is unusual, and most of it exists already.
Three approval gates are yours: scope, permission ratification, design sign-off. We answer each within three business days. Time beyond that on your side moves the date by exactly that much — and nothing else does.
Production cutover on your change calendar. SSO and network provisioning by your IT. Third-party licenses or vendor contracts we don't control. Remediation of source-data quality problems the assessment flagged. Anything added after the scope gate.
Functional parity against the approved specification, running on your migrated data in a parallel environment, with production security enforced and tested per role, all nine quality gates green, and documentation published. Your users click through it.
These are the fastest rebuilds in the campaign, because InfoPath forms and Designer workflows are small, entirely XML, and were built quickly in the first place. What moves the date is never the form — it is the undocumented list-to-list dependencies, item-level permissions somebody set by hand years ago, and code-behind that reached into the farm.
05 · The obvious question
Sometimes — and we'll tell you when. For a simple form inside an E3/E5 tenant, Power Apps is the low-friction answer and we won't pretend otherwise. The apps that outgrew Access and InfoPath are usually not that.
| Where it bites | Power Platform reality | Mendix |
|---|---|---|
| Data at scale | Canvas-app delegation limits (500 default / 2,000 max rows, per Microsoft's own documentation) silently return wrong results on non-delegable queries | Server-side queries with no delegation ceiling — the database does the work |
| Licensing | Premium runs $20/user/month; the $5 per-app plan was pulled from the licensing guide in January 2026; The Register documented "restricted table" usage forcing $95/user/month Dynamics licenses | Platform licensing that doesn't multiply per app — the departmental-portfolio shape is the point |
| Governance | Documented sprawl: one published audit found 47% of an enterprise's Power Apps had no documented owner — Access-style shadow IT, relocated to the cloud | One governed platform: model-level security, a single deployable app per process, an audit ledger |
| Complex apps | Multi-entity, multi-role apps with real workflow are where per-user premium pricing and connector patchwork stack up | Real relational modeling, native workflow engine, role-based security down to the row |
This is not an anti-Microsoft choice. Mendix runs happily on Azure, against SQL Server, behind Entra ID single sign-on. Your Microsoft platform stays; the abandoned app toolchain on top of it is what gets replaced.
06 · Straight talk
Schemas, relationships, validation rules, workflow logic, and who-can-see-what all extract cleanly — and land in the spec you approve. Historical data migrates; where SQL Server is already the backend, it barely moves.
You get a modern, responsive, accessible web app — not a pixel clone of a 2009 form. Design directions are presented for your sign-off first, so "modern" is a choice you make, not a surprise you receive.
COM and Office-interop dependencies in VB6/VBA need redesign, not translation — the assessment names them app by app. Excel formulas extract, but re-verification against real cases is part of testing, not skipped. And forms worth retiring get retired, not billed.
07 · Quality & security
08 · Engagement & pricing
Marcus reads your estate — forms, workflows, databases, VB6 source — and delivers a scored inventory: rebuild / retire / archive per app, complexity ratings, extracted specs for the priority apps, and a fixed-bid quote for each rebuild.
Runs on artifacts you export; read-only credentials where needed. The deliverable is yours either way.
Each app rebuilt as a Delivered engagement: working, fully tested, documented, with published evidence — priced from its assessed complexity, not from hope.
Typical departmental apps are days-to-weeks of Marcus work, not consultant-quarters.
For portfolios, a dedicated Marcus instance works the migration backlog continuously — either your team drives, or a Kinetech engineer reviews every release, with an SLA.
Bring your own Claude access — a direct Anthropic key or your AWS Bedrock / Google Cloud commitment — and reduce the monthly by roughly 15–25%.
One-time onboarding: $6,000–$10,000 — provisioning, secure credential setup, brand intake, first approval cycle. Pilot-program pricing — locked for 12 months for early partners.
09 · The questions you should ask
Keep the autonomy; retire the artifact. The people who built those databases are exactly who we want in the approval gates — they know the process better than anyone, and the extracted spec gives them the first complete picture of their own app they've ever had. What they give up is the 2 GB file on a share with no backup discipline; what they keep is ownership of how their process works.
Because the farm under them is now unpatched. SharePoint 2016 and 2019 left extended support on July 14, 2026, and there is no ESU program for SharePoint — the option Microsoft sells for Windows and SQL Server simply doesn't exist here. Every month on that farm is a month of accumulating unpatched exposure on a system that, by definition, holds your business processes. "Still works" and "still defensible" have parted ways.
For InfoPath, Microsoft said the quiet part out loud: no extension, no migration tool — their guidance is to rebuild in Power Apps by hand. Their assessment tools (which we use gladly) inventory what you have; nothing Microsoft ships converts it. That rebuild-by-hand cost is precisely what Marcus collapses: the reading, specification, and rebuild are automated, and your people spend their hours on approvals instead of re-keying rules.
Three ways. First, the spec is written in plain English and reviewed by your process owner before build — extraction errors die in review, not in production. Second, every rebuilt app is driven end-to-end per role with test evidence published. Third, old and new can run in parallel against the same data during acceptance, so discrepancies surface while the legacy app is still there to compare against. We don't ask you to trust the AI; we ask you to check its work — the whole system is built to make that easy.
Data is the most readable layer of all of these apps, and the migration is the least speculative part: schema and data extract via the standard engine interfaces, Microsoft's own SQL Server Migration Assistant is a proven de-risking step, and the destination is a real server database with backups, concurrency, and no 2 GB ceiling. Where a VB6 app already runs on SQL Server, the data often doesn't move at all — the app around it does.
It's the normal case, and it's why reading the source matters. The artifact is the documentation — the form definition holds the rules, the VBA holds the logic, the schema holds the model. Marcus extracts all of it whether or not anyone remembers writing it, and the assessment gives every app an owner decision: rebuild, retire, or archive. Orphaned apps usually land in the last two — which is the cheapest outcome anyone's offered you for them.
Your current position is lock-in to abandoned platforms — InfoPath has no vendor path at all. The rebuilt apps are standard Mendix in a repository you own, on data in a standard SQL database, with 100% in-model documentation and a full test suite. If you leave Mendix someday, you leave with a documented, tested model and your data in plain tables — a categorically better exit than the .xsn files you're holding now.
10 · Next step
Choose one InfoPath form with a workflow behind it, one Access database with real users, and one thing nobody wants to touch. The assessment reads them, specs them, and quotes them fixed — and you judge the evidence, not the pitch.
Marc Lehane · Solutions Architect / CTO, Kinetech · [email protected]
Kinetech Cloud · San Antonio, TX · How we deliver · Live Build Console
You are reading Your platform — deeper detail on step 4, Why now?.
That is the pressure. What it costs to answer it is not one number — three things move it.