Managed IT · Custom software · Since the SaaS bill got silly
Stoprentingthesoftwareyourbusinessrunson.
We audit what you are paying for, rebuild the expensive parts as systems you own, and then run them for you. Built with AI so it costs what it should. Answered by people because that part still matters.
- Every subscription mapped, with its real owner and its real cost.
- Replacements built on open source, deployed to hardware in your name.
- Month to month, with a written exit plan from the first week.
Built on foundations you can read
- PostgreSQL
- Proxmox
- Docker
- Nextcloud
- Authentik
- Grafana
- n8n
- Mattermost
- Traefik
- Zammad
01 / The situation
None of this started as a bad decision.
Every subscription solved a real problem on the day it was signed. It is the tenth one that hurts, and the fact that leaving any of them now costs more than staying.
- Nobody can say what the total subscription bill actually is.finance
- Two tools do the same job because two teams bought separately.duplication
- The seat count went up. The value did not.per-seat pricing
- A vendor changed their pricing and there was nothing to say back.leverage
- Getting your own data out means a support ticket and a wait.lock-in
- Half the workflow is a person copying between two systems.glue work
Subscriptions the platform displaces
- Zendesk
- Jira
- Miro
- Slack
- Notion
- Zoom
- HubSpot
- 1Password
- Dropbox
- DocuSign
- Confluence
- Tailscale
- Asana
- PagerDuty
- Zapier
- Linear
- Harvest
- Box
02 / How it works · 01
Audit
We inventory every subscription, every login, and every workaround your team invented to survive them. You get the map whether or not you hire us.
- Every recurring line item, with its real owner
- What each tool is actually used for, by whom
- Data you cannot export, flagged early
- A replacement cost model, per system
- Ticketing SaaS$412 / mo
- Document storage$286 / mo
- Form builder$149 / mo
- Scheduling$118 / mo
- Reporting add-on$240 / mo
A worked example. Your line items and totals will differ.
03 / What replaces it
One platform where twenty subscriptions used to be.
Not a bundle of separate apps behind one login. One data model, so a ticket already knows the company, the contract, the invoice, the device and the outage it belongs to. Named tools below are the category each surface displaces.
One work graph. A ticket can carry the company, the deal, the contract, the invoice, the device and the outage it belongs to, instead of five tools each holding a quarter of the story.
- Tickets, queues, SLAsZendesk · Freshdesk · ConnectWise
- CRM, companies, dealsHubSpot · Pipedrive
- Estimates, invoices, retainersHarvest · billing add-ons
- Contracts & e-signatureDocuSign · PandaDoc
- Client portalone portal per vendor
04 / What we do
Six lines of work. One bill. One phone number.
Most shops do one of these and subcontract the rest. We do all six because the seams between them are exactly where small businesses lose money and time.
Managed IT
A real person answers, and an AI has already read every ticket you have ever filed before they pick up.
- Helpdesk with named response times
- Device enrolment, patching, inventory
- Identity, email, backup, recovery
- Quarterly stack and spend review
Custom software
The tool you keep paying $900 a month to rent, built once, deployed to hardware you own.
- Discovery against your actual workflow
- Built on open-source foundations
- Deployed to your infrastructure
- Source, schema and runbook handed over
Security
Fewer vendors holding your data means fewer ways to lose it.
- Access review and least-privilege rollout
- Endpoint and network hardening
- Backup restore drills, not backup reports
- Incident runbook your team can execute
Infrastructure & DevOps
Self-hosted does not mean fragile. It means the pager reaches us, not you.
- Proxmox, Docker or cloud, your call
- Monitoring, alerting, and log retention
- Reproducible deploys from git
- Capacity and cost modelling
Automation
The copy-paste work between your systems, deleted rather than documented.
- Process mapping before any building
- Workflow automation across your tools
- Document and reporting pipelines
- Failure alerting with a human fallback
AI enablement
Your team using AI on your data, with the boundaries drawn before anyone types.
- Private model deployment options
- Role-specific training, not a webinar
- Usage policy and data boundaries
- Ongoing office hours
05 / The method
Old-fashioned support. Modern build costs.
Your stack, on your silicon
- Why open source underneath
- Because when the component is public, the thing you depend on outlives us. You can hire anyone to work on Postgres. Nobody else can work on a vendor's closed platform.
- Why AI in the build
- It collapses the cost of the unglamorous 80%: schemas, migrations, admin screens, tests. That is the difference between custom software being a fantasy and being cheaper than the subscription.
- Why people on the phone
- Because the moment something is actually broken, nobody wants a chat widget. The AI has already read the history by the time a technician picks up.
06 / How to fire us
A provider you cannot leave is just another subscription.
We would be a poor advertisement for escaping lock-in if we quietly became the lock-in. So the exit is written down before the work starts, and these terms are in every agreement.
- You own the code
- Every line we write for you lands in a repository you control, under a licence that does not expire when the invoice does.
- You own the data
- Your databases run on your infrastructure. Export is a documented command, not a support ticket and a 30-day wait.
- You own the servers
- Cloud, colo, or the closet down the hall. Billed to your account, in your name, with your credentials.
- Month to month
- No multi-year lock-in and no early-termination fee. If we stop earning it, you stop paying for it.
- A written exit plan
- Maintained from the first week: credentials, runbooks, architecture notes, and a handover call with whoever comes next.
- No margin on your vendors
- We do not resell your licences back to you at a markup. What the vendor charges is what you pay.
07 / How we charge
No price list, because it would be wrong for everyone.
A twelve person firm running four subscriptions and a forty person firm running thirty are not the same job, and any number published here would only mislead one of them. What does not change is how the billing works, so that is what we publish instead. You get a real number after the audit, before any build starts.
- Retainer
- A monthly block of covered work, agreed before it starts and drawn down as it is used. What is left at the end of the month is visible to you, not quietly kept.
- Hourly
- For work that falls outside the retainer. Every hour is logged against a ticket, so the line on the invoice points at something you can read.
- Fixed contract
- A defined scope for a defined number. Used for replacement builds, where both sides can say in advance what finished looks like.
- Project
- Larger builds split into stages, each with its own sign off. You can stop between stages without owing the rest.
- Hosted
- Where we run infrastructure on your behalf, billed on what it consumes. The underlying hardware or cloud account stays in your name.
- Included
- Onboarding, quarterly stack reviews and keeping the exit plan current. These are never a separate line.
Vendor licences pass through at cost. We take no margin on anything we resell you, because the moment we do we start recommending the wrong tools.
08 / Straight answers
The questions you were going to ask anyway.
The opposite. AI is how we keep the cost of building and supporting custom systems low enough that replacing a SaaS subscription actually pencils out. What you buy is the working system and the support behind it. The tooling is our problem.
09 / Start here
Book the stack audit.
Two to three weeks. You get the full map of what you are paying for and what it would cost to own instead, as a document you keep whether or not you hire us.
No newsletter. No sequence.