Who we build for
You do not need a spec. You need someone who has done this before.
Most people land here from the footer of something we built for somebody else. Whichever of these sounds like you, the first step is the same: a short email and a proper conversation.
You have the idea
No technical co-founder, no engineering team.
You know the problem better than anyone and you can describe the product in a paragraph. What you are missing is the person who turns that into software people can pay for.
We become that person, and you keep every share.
You run the business
The spreadsheet stopped coping about a year ago.
Quotes, jobs, stock, staff, invoices, all held together by files, group chats and one person who remembers how it works. It is slowing you down and you know it.
We turn how you already work into a proper system.
You are technical
You could build it. You do not have the hours.
You want people who read a schema before opinions, who write tests, who can talk about idempotency and blue green releases without a glossary, and who leave the repo tidy.
Skip the sales call. Send us the repo and the problem.
You already have a product
It works, but every change now feels risky.
Maybe an agency built it and moved on. Maybe it grew fast and nobody had time to tidy up. Deploys are scary, the bill keeps rising and nobody is quite sure what runs where.
We audit it, stabilise it, then make it cheap to change.
What we build
Name the thing. We have either built it or built its awkward cousin.
Subscription software is a business model, not a category. Underneath, it is the same craft every time: screens people can use, an API that keeps its promises, data that stays correct, and infrastructure that does not fall over on a Friday.
Customer portals
The account your customers log into. Sign up, teams and permissions, documents, messages, notifications, an audit trail of who did what.
client areas · member portals · tenant dashboards
Operations software
The system your staff live in all day. Jobs, scheduling, stock, approvals, forms that used to be paper, and the reports your accountant keeps asking for.
field service · warehouse · clinics · back office
Money and financial products
Anything that has to add up. Ledgers, reconciliation, quotes and invoices, subscription and metered billing, VAT, refunds, dunning, payouts.
Stripe · Open Banking · ledgers · payouts
Marketplaces and platforms
Two sides that need each other, plus the awkward middle. Listings, matching, escrow, reviews, disputes, commission, and the admin tools to police it.
bookings · rentals · brokerage · supply chains
Data products and reporting
Pulling numbers out of the systems you already pay for and making them mean something. Pipelines, warehouses, dashboards, alerts, exports people trust.
BI dashboards · pipelines · exports · alerting
AI products
Software where a model does real work rather than decorating the page. Assistants grounded in your documents, agents that take actions, review queues for humans.
agents · retrieval · classification · copilots
APIs sold as a product
When the product is the endpoint. Keys, quotas, versioning, webhooks, sandbox environments, docs a developer can follow without emailing you.
developer platforms · integrations · webhooks
Mobile and field apps
The same product in a pocket, working when the signal does not. Offline first, camera and scanning, push, background sync, on both app stores.
Flutter · React Native · offline sync · scanning
Sectors we have shipped into
Not on the list? That has never once been the reason a project did not work. Tell us the industry and we will tell you honestly what we already know and what we would need to learn from you.
Still reading? That usually means you have one in mind.
Describe it in three sentences and send it over. You will get a straight answer about whether it is worth building, what it would take, and roughly what it would cost.
How it works
Six weeks from first email to something your customers can log into.
That is the usual shape for a first release. Bigger products take longer and we will tell you that upfront rather than at the end. What never changes is the order: understand it, shape it, build it where you can see it, then run it.
01
Day 1
A conversation, not a pitch
Half an hour on a call. What the product is for, who pays, what happens today without it, and what would count as it working. If we are the wrong people we say so on that call and point you somewhere better.
02
Week 1
Shape it and price it
You get the screens sketched, the data model written down in plain English, the risky parts named, and a fixed price for the first release. No hourly meter running while we work out what you meant.
03
Weeks 2 to 5
Build where you can watch
A real link you can open, updated every few days, running the same way it will in production. You use it, you tell us what feels wrong, we change it. Nobody disappears for two months and returns with a surprise.
04
Week 6
Live, on your own account
Your cloud account, your domain, your repository, your database. We set up backups, monitoring, alerts and the deploy pipeline, then hand over the keys and walk you through all of it.
05
After
Someone still answers
Software is never finished. We keep it patched and watched, ship changes as your business changes, and stay on the other end of an email. Or we hand it cleanly to your own team, which is a perfectly good ending too.
What you get either way
You own the code, the accounts and the data from day one
Fixed price for the first release, agreed before we start
A working link in weeks, not a slide deck in months
Plain English updates, no jargon used as a shield
No lock in. If you leave, everything leaves with you
The fastest way to find out what your idea would cost is to describe it in a few sentences and send it over. You will get a real answer, not a form.
The layer under the product
We build the software and we build the ground it stands on.
Plenty of studios can produce screens. Fewer can tell you what happens when your database fails over at 3am, or why your cloud bill tripled after a launch. Eight Mile started in networks and infrastructure, then grew upwards into products, so the whole stack is one conversation with one team.
Languages and frameworks
Cloud and hosting
Infrastructure as code
Data and storage
Shipping changes
Knowing it is healthy
Security and access
Networks and edge
Architecture questions get answered properly, whether or not you hire us.
AI and automation
AI in your product, built by people who have shipped it and run it.
Two of our open source projects are exactly this: an agent that routes conversations to the right specialist and grounds its answers in your documents, and a cache that stops you paying a model twice for the same question. Read the source before you believe a word of it.
Assistants that know your business
Grounded in your own documents, tickets and policies rather than the open internet. They quote the source they used, and they say “I do not know” instead of inventing a refund policy.
Agents that actually do things
Software that reads a request, decides which tool to reach for, calls your API, and hands off to a human when it is out of its depth. Every step is logged and replayable.
Documents into data
Invoices, contracts, forms and PDFs turned into fields you can query, with a review queue for anything the model is unsure about. Nothing silently guessed into your database.
The boring work, automated
The daily export, the reconciliation, the chasing email, the report someone rebuilds by hand every Monday. Scheduled, monitored, and loud when it fails.
AI dropped into what you already run
Search that understands intent, drafting inside your existing tools, classification and routing, summaries that save a person twenty minutes. No rebuild required.
The unglamorous half
Prompt injection defences, PII masked before the model sees it, caching so you stop paying twice for the same answer, evaluations so you can prove a change made it better.
Proof, not promises
ChatAgent
An LLM decides whether a conversation belongs to sales, billing or support, then a specialist answers from your own documents, behind guardrails on the way in and out.
Read how it worksllm-cache
Exact matching first, then vector similarity when the wording changes. The second person to ask the same question costs you nothing and answers instantly.
Read how it worksBoth are open source, both run in production, and both were written before anyone paid us to say we do AI.
Talk about an AI featureBring us the part you are dreading.
The integration nobody wants to touch, the migration that keeps slipping, the feature your team has rewritten twice. Those are the emails we enjoy most.
Already have something
Moving it, fixing it and taking it over are all normal jobs here.
Not every project starts on a blank page. A good share of our work is inherited: something that runs, mostly, but nobody enjoys touching. We are comfortable walking into that.
Off the server under the desk
That machine in the cupboard, or the one VPS somebody set up in 2016. We map what it actually does, rebuild it as code, move it, and prove the new one works before the old one is switched off.
Cloud to cloud, or off cloud entirely
AWS to GCP, either of them to your own hardware, or the other way. We have written our own Terraform providers, so we are comfortable in the corners where the tooling runs out.
Database moves without the drama
Version upgrades, engine changes, splitting one database into several. Rehearsed on a copy, cut over with replication, with a rollback that we have tested rather than hoped for.
Codebases somebody else left behind
No documentation, no pipeline, no original developer. We read it, write down how it works, get it building again, then make the first change safely so everyone can breathe.
Bills that stopped making sense
Rightsizing, killing infrastructure nobody remembers creating, storage tiers, committed use, and honest advice about the managed service you are paying four times over for.
Getting ready to be audited
Backups that have been restored at least once, access logs, least privilege, secrets out of the repo, disaster recovery you could actually perform on a bad day.
Sound familiar?
Only one person knows how to deploy it
Nobody has restored a backup to check it works
A small change takes three weeks and a prayer
Servers were configured by hand and never again
There is no staging, so production is the test
The cloud bill goes up and nobody can say why
Three or more of those and it is costing you money every month. A short audit usually pays for itself before the work even starts.
Things we have already built
14 products running today. 9 of them you can read the source of.
Access control on card readers, high availability PostgreSQL, an inbox that triages itself, a PDF editor that never uploads your file, Terraform providers published to the public registry. Different worlds, same team.
Ways to work together
Hire us for the whole product, one slice of it, or just a straight answer.
Every arrangement below starts with the same email. Tell us roughly what you need and we will say which of these fits, including when the honest answer is that you do not need us yet.
Second opinion
A week of honesty
We read what you have, or what you plan to have, and write down what we would do differently. Useful before you spend real money, and cheap compared with finding out later.
Architecture and code review
Risks and costs written plainly
A short plan you can hand to anyone
Build
Most people
We build the whole thing
The usual arrangement. Design, product, front end, API, database, infrastructure and launch, delivered as one fixed price release and then the next one after that.
Fixed price per release
A working link from week two
Live on your own accounts
Handover documents and a walkthrough
Alongside
We join your team
You have developers and a backlog they will never reach. We take a slice of it, work in your repositories and your process, and leave your standards better than we found them.
Your tools, your board, your review process
Pairing and knowledge left behind
Cover for the parts nobody enjoys
Run it
We keep it alive
Updates, patches, backups tested for real, monitoring that wakes someone up, and a person who answers when something looks wrong at an inconvenient hour.
Monitoring and alerting
Patching and dependency upgrades
Backup and restore drills
A monthly change budget
About money
No price list, because we have never met two identical products.
What we will do is give you a real number quickly. A few sentences by email usually gets you a range the same day, and a fixed price for the first release within a week of talking properly. If the budget you have and the product you want do not meet, we will say so early rather than build you two thirds of something.