Virtual Software Engineering Teams, as a Service · Open for pilots

Virtual engineering teams, led by a human.

Add dedicated agent fleets by role: Dev, QA, Cloud, IoT, Support. A flat monthly price, KPIs and SLAs agreed up front, and an audit trail. Move from human teams to virtual FTEs at your own pace.

Build your team Talk about a pilot
One agent fleet
€3,500to €5,000 a month
Product team, 6 fleets + 2 light
€18,000a month, lead included
First 3 months
−50%half the hours, if eligible

The upgrade

From human teams to virtual FTEs, at your pace.

Go as far as it makes sense for you. Hybrid is a fine place to stay.

Today

Start with the team you have.

Your people, your tools, your process. We map the backlog and agree the KPIs. Nothing changes until you say so.

Pilot

Add a few fleets next to your people.

Three months at half price and half the hours, for eligible clients. Your engineers review what the fleets propose, and the KPIs show if it works.

Your call

Grow what the numbers prove.

Add fleets where the KPIs hold. Stop adding where they don't. Whatever the mix, one rule stays: a human leads every team.

Agile, not change requests

Your backlog moves. Your price doesn’t.

Fixed price for a measured outcome is a great deal when the scope holds still. When a team runs agile, the scope can move every sprint, and a move can become a change request. A VSETS team takes your next ticket instead. Re-prioritise any day, with no new quote, within your hours cap. You still measure the results: KPIs, SLAs and hours, every month.

Some work has no end date. It needs a standing team.

Continuous deliveryDev teams that take the next ticket, sprint after sprint. 24/7 support and maintenanceAlerts and tickets handled at any hour, within your active hours. Security guardsChanges, dependencies and secrets checked, in the repositories you choose.

The team builder

Compose your team from the menu.

Each fleet is a dedicated group of agents, priced per month and capped in active hours. We count it as one virtual FTE for budgeting, not as a measure of output. The ticket shows the list price, and the lead comes with the team.

8 fleetslead included€18,000a month

What's inside

An engineering org, not a chatbot.

Each fleet is built to run on the Digital One stack: orchestrated, controlled, informed, audited and measured. RiverAct and LiftWard are in beta, and your pilot plan says what is switched on from the first day.

  1. Orchestrates

    builds.one Enterprise Ultra

    End-to-end orchestration of the fleets. For enterprises only.

    Beta

  2. Controls

    Gateward

    The warded gateway for your models. One org key, daily caps and an evidence chain.

    Production

  3. Chooses

    FrontierScore

    Know which model is actually good right now. Model choices rest on live evidence.

    Production

  4. Knows

    SkilledMind

    The mind your models share. Team, organisation, market and industry knowledge.

    Production

  5. Proves

    RiverAct

    Govern the flow. RiverAct takes custody of the evidence chain, so you can audit the work. With SignWard (beta). Every agent carries papers.

    Beta

  6. Measures

    LiftWard

    The auditable AI P&L. What the team costs against what it delivers.

    Beta

Data and IP security

Your code, under your controls.

Two modes planned for enterprises that cannot send everything to a model. Both are on the builds.one Enterprise Ultra roadmap. Neither is in the beta yet.

Chameleon mode

Roadmap · in progress

Your organisation's policy reroutes sensitive work to local models. The fleet keeps working.

Camouflage mode

Roadmap · in progress

No local model available? Personal data, client names and IP are swapped for realistic look-alikes before a prompt leaves, then swapped back in the answer.

In the builds.one closed beta today

  • Each helper in its own worktree, on its own branch
  • Build and review helpers sandboxed, with the network off
  • Nothing pushed or merged: a human merges
  • Budgets, time limits and a kill switch
  • Every helper shown live, and the day replayable
  • Approvals as cards, with the diff

These guards work on a person's own plans today. Your pilot plan says which apply to your fleets, on which model route, and how Cloud and Support fleets connect.

Where it runs: on your machines, or in a hosted EU residence. On-prem is agreed per contract, and the contract names the models. The rest we show in the room, under NDA.

The business case

Standing teams, in plain numbers.

List prices. Your pilot KPIs replace these examples within three months.

Continuous delivery

A product team that keeps shipping.

€18,000a month

  • 6 fleets, 2 light fleets and the lead, sharing 1,500 active hours
  • First 3 months: €9,000 a month with half the hours, if eligible
  • Year one: €189,000 with the pilot, €216,000 without
  • At list, €216,000 a year pays salary and employer cost for 2 to 3 senior engineers. Cost only, not output
Load this team

24/7 support and maintenance

Nights and weekends, covered.

€10,500a month

  • A week has 168 hours. One human seat around the clock takes more than 4 full-time people. Cost only, not output
  • 2 Support & Monitoring fleets, plus Maintenance
  • 750 active hours. Waiting between alerts does not count. At the cap, agents stop until your budget owner approves more
  • Your lead reviews in business hours. Human night on-call is quoted apart
Load this team

Security guards

Changes checked, day and night.

€9,000a month

  • 2 Security Guard fleets on your repositories
  • 500 active hours, day or night. Findings go to your people
  • An AI Engineering Lead comes with the team
  • Findings go to your people. Nothing merges by itself
Load this team

The specialist gap

A specialist fleet, without a hiring search.

€12,000a month

  • IoT & Embedded with Software and QA fleets
  • An AI Engineering Lead comes with the team
  • No search and no recruiter fee
Load this team

No speed promise. Independent evidence on AI and developer speed is mixed. METR measured tasks taking 19% longer in early 2025. Its late-2025 data points to a speedup, which METR calls “only very weak evidence”. DORA 2025 calls AI “an amplifier, magnifying an organization’s existing strengths and weaknesses”. So every pilot starts from your own baseline.

Cost comparison. Our calculation from 2026 salary data and employer contributions in the UK, Germany, Ireland, the Netherlands and Hungary. For example, the UK: £75,000 median advertised salary, ITJobsWatch, 6 months to 2 November 2026, and 15% employer contribution above £5,000, PwC. It compares cost only. It does not say a fleet does the work of a person.

The deal, plainly

What you get, in plain words.

A human, always

Every team has a fractional AI Delivery Lead: 2.5 days a month per paid fleet on Standard, 3.5 on Extended. Teams of only QA, only Cloud or only Security, and any team with an IoT or Game fleet, get an AI Engineering Lead. By default nothing merges without human review, and your contract names who reviews. On very large deals, you can bring your own lead after our training.

Hours, not 9 to 5

Agents run at any hour, so we cap the work, not the clock. An active hour is one agent on a task from your backlog for 60 minutes. Waiting does not count. Standard has 250 per paid fleet each month, Extended 500, shared across the team.

Costs that stop

Every team has an hours cap and a monthly model budget. At either one, agents stop and the lead re-plans. Nothing runs past them unless your budget owner approves: €8 an hour, with model use on the route your contract names.

Easy to pilot

Three months at half price and half the hours for eligible clients, with the lead's time in full. Eligible means tests and CI in place, and one pilot per client. We agree a baseline, KPIs and SLAs first. If we miss the written criteria, you can leave at month three. Your agreement sets the term.

A monthly service fee, for finance

Usually recorded as an operating expense. Some spend on building software you own may qualify for capitalisation under IAS 38 or ASC 350-40, depending on your facts. We plan a monthly statement of hours and model use by project and activity; your pilot plan says what is ready. You and your auditor decide. This is not accounting advice.

Your place or ours. Your models or ours.

On your machines, or in a hosted EU residence. On-prem is possible under certain agreements. Local models are on the roadmap. We bring the models inside one flat price, or your company's API or cloud account pays the provider directly and the price is lower. Your contract names the models. Built around builds.one Enterprise Ultra, which is for enterprises only.

Start with one squad.

Pilots are open for eligible clients. Tell us about your backlog, and we come back with a team, KPIs and a pilot plan.

Talk about a pilot Build your team