Off Manus, onto our own stack

01The production cutover

Everything below runs on staging addresses. The real addresses — the olympustentsandevents.com subdomains customers actually visit — still point at Manus and will die with it. Moving them is a DNS change, made by hand in Squarespace, which is where that domain's records live. Cloudflare hosts the staging domain only; it has no control over the live one.

Nine records change, and every one gets the same new value. One record is deleted.

Leave every other record alone. The appmail records, both mail exchangers, the SendGrid and Mailgun entries and www have nothing to do with Manus, and changing them breaks email.

The Vantage is a separate job

thevantageatolympus.com is also still on Manus, but it is not in the Squarespace panel — it is on Cloudflare, in an account separate from the one holding the staging domain. It needs its two addresses registered against the Vantage application first, then its own DNS records changed, with Cloudflare's proxy switched off so certificates can be issued. Full instructions were provided separately.

02The sites

All eleven run on *.staging.aetenum.com. Staging hosts are marked noindex so Google will not index them alongside the live sites; the production domains are served indexable.

03What we were handed

Manus is shutting down. What survived was seven private GitHub repositories and nothing else — no running servers, no database, no environment settings, no admin console. The job was to make all seven run again on infrastructure we control, and to prove each one actually works rather than merely loads.

The repos turned out to be seven copies of one Manus scaffold: a React front end, an Express back end, and a folder called server/_core holding Manus's own platform code. That sameness was the single biggest advantage in the whole project — nearly every fix, once found, applied to all seven.

An eighth application surfaced afterwards — the internal document library — delivered as a full export rather than a repository: source code, a database dump, and the uploaded files downloaded out of Manus's storage. It runs on the same scaffold, so the work already done carried straight over; what it added was a second set of real business data to move, and a genuinely new problem described in section 07.

Three more surfaced when the live DNS records were finally read: the privacy policy, the terms and conditions and the thank you page were each a separate Manus site on its own subdomain, not pages inside another site as everyone assumed. They arrived as three more exports and were rebuilt the same way, bringing the total to eleven. Two of the three mattered more than their size suggests:

• The thank you page is where every lead form on every site sends the customer after they submit, and it is where the Google Ads conversion is recorded. Had it been left behind, forms would still have reached GoHighLevel but no advertising spend would have been attributable to any of it.
• The privacy policy is the policy of record for text-message carrier registration. That registration requires a publicly reachable address; if the page had gone dark, business texting would have been at risk rather than merely a broken link.

A fourth subdomain, invitation, held the RSVP site for the Vantage launch event in May. The event has passed, so it is being retired rather than rebuilt.

The scaffold was hiding how simple most of them were

On the surface, six of the seven looked database-backed. They shipped a database schema, a database driver, and login plumbing. Looking closer, the schema in five of them was untouched boilerplate that still ended with the generated comment // TODO: Add your tables here, and no page in any of those apps had a login screen.

So the databases were decoration. Removing that dead weight meant five apps needed no database at all — no hosting bill, no credentials, no backups, nothing to break. Of the original seven, only the time clock was a real application; the document library that arrived later is the second.

04The awkward parts of the Manus export

The exports were complete in the sense that the source code was all there. They were incomplete in every way that matters for actually running the software. Each of the following would have produced a site that looked fine in testing and then quietly broke.

05The new setup

Everything was deliberately created fresh under aaron@olympustentsandevents.com, kept separate from any personal accounts, so ownership and billing sit with the business.

Roughly $30 a month for all eleven — the three sites added last are static pages and cost nothing extra to host. The Vercel paid tier is not optional: its free plan forbids commercial use, and the time clock's automatic calendar sync needs scheduling that only the paid tier allows.

What changed inside the code

• The three static sites no longer run a server at all. Their pages are served directly from a global network, which is faster and cheaper than keeping a machine running to hand out files.
• The three one-endpoint sites keep their single API, but it now runs on demand instead of on an always-on server. Their dead login and database code was removed entirely.
• The time clock and the document library needed the most work, covered below.
• Manus's platform code was deleted — file storage, notifications, AI helpers, voice transcription, maps and the rest, none of it reachable from these apps and all of it pointed at services that are going away. server/_core went from 17 files to 4 in most projects, and about 20,900 lines of dead code were removed across the original eight. The three sites added later were cleaned the same way.

06Rebuilding the time clock

This was the only genuinely difficult application, and the only one with real business data at stake.

Logging in

Staff signed in through Manus's own login service. There was no way to keep that, so it was replaced with Clerk.

The risk in swapping a login system is losing the connection between a person and their history. Here the numbers made it manageable: the database held 48 employees but only 2 user accounts — only two people had ever actually logged in; the rest exist as employee records. The new system matches people by email address, so those two accounts keep their identity, their admin rights, and their attached time entries.

Helpfully, the whole application asks "who is this?" in exactly one place in the code, so the swap touched one file rather than being spread across the app.

The automatic calendar sync

The app pulled jobs from Google Calendar every 15 minutes using a timer inside the always-on server. On-demand hosting has no always-on process for a timer to live in, so the sync now runs as a scheduled task that calls the application on the same 15-minute cadence. The sync logic itself is untouched — only what starts it changed. The endpoint is protected by a secret, and refuses unauthorized calls.

The data

Restored into the new database and verified row by row. One deliberate detail: the export file contains employee names, hours and location data, so it is explicitly excluded from anything deployed to the web. It stays in the repository for recovery only.

07The document library

The eighth application is where staff read handbooks and SOPs, sign to say they have read them, and take a short quiz to confirm it. It carried 287 rows of real records: 12 staff, 12 signed acknowledgements, and 33 quiz attempts across 40 questions. All of it restored and verified row by row.

Three Manus services were doing real work inside it, and each needed a genuine replacement rather than deletion.

The signatures were the thing most at risk

Every handbook, every uploaded video and all twelve signed acknowledgements lived on Manus's file storage. The database recorded where each one was, and every one of those addresses pointed at a Manus server. They work today; at shutdown all twelve signatures — the actual evidence that a named employee accepted a policy on a given date — would have become dead links, with the records still claiming they existed.

All 16 files were re-uploaded to our own storage under the exact same filenames the database already referenced, so only the addresses changed. A check afterwards confirmed no record anywhere still points at Manus, and that every file is a real PDF or image rather than an error page wearing one's clothes.

One login for two apps, which needed a door on it

Sharing the time clock's login is convenient for staff, but it has a consequence worth being explicit about: it means anyone who can clock in can also reach the document library. And the app's own default was to give a new arrival the "sales" department, which would have shown them every document marked as company-wide — which is the default setting, and is set on both handbooks.

So signing in and being allowed in were separated. A person who has never been invited now gets an account with no access at all and a plain message telling them to ask HR for an invite link. The check sits at the lowest level of the app rather than on each page, so any screen added later is closed by default instead of relying on someone remembering to lock it. The twelve existing staff were admitted automatically and notice nothing.

What it means for the bill

The document library runs on its own database, separate from the time clock, so an incident with one cannot take out the other. That is the $5 difference between the old figure and the new one.

08Problems hit along the way

Five things broke that no amount of local testing would have caught, because each only appears in real hosting.

09How each one was checked

"It loads" is not proof. Every claim above was tested against the live addresses:

• Every page returns a real page, over valid HTTPS.
• Every image was confirmed to be an actual image file, not a disguised error.
• Sub-pages work when opened directly or refreshed, not only when navigated to.
• The APIs were exercised for real — but deliberately with invalid input, so their validation could be proven without pushing a fake customer into the CRM.
• The time clock's protected pages were confirmed to reject anyone not signed in.
• The scheduled sync was confirmed to reject calls without the correct secret.
• The database connection was proven indirectly but conclusively: the sync reached Google using credentials it had just read out of the restored database.
• For the document library, every one of the eight ways to reach staff records was tried without signing in, and every one refused. Its database was proven live from the hosting by reading the seven document categories back out of it over the public internet.
• All twelve signatures and both handbooks were fetched from their new addresses and confirmed to be genuine PDFs and images of the exact expected size — not error pages returning a misleading "OK".
• Anything not yet configured fails loudly rather than silently — the tents form says its webhook is missing instead of pretending a lead was sent.
• The search-engine instruction was proven both ways on a real address, not just read in the configuration: staging receives it, a non-staging address does not, and the two do not contaminate each other through caching.
• All eleven staging addresses were re-checked together at the end and every one returned its own correct page.

10What is left