Eight enterprise AI products — a gateway, live model intelligence, shared memory, governance, trusted evidence, agent identity, liability evidence and an auditable AI P&L. Each supported on its own; connected through Gateward, one governed backbone.
Builds runs your coding agents — Claude Code and Codex — where your code lives: on your own Mac or Linux machine, or in a hosted EU residence. Every session keeps running on that machine; your phone, tablet, laptop and browser are live views of the same work. Agents ask, you answer from anywhere. Bring your own plans and keys — Builds sells no model access.
Close the laptop or lose signal: the turn keeps going, and every screen catches up.
An agent's question arrives as a card with its diff — allow once, always, or reject. Answer from the iPhone lock screen; Face ID before risky actions.
One conversation per desk, several desks working at once, sub-agents in view. Switch model, effort or provider mid-conversation — the context comes along.
Claude Code on your Claude plan, where your provider allows it; Codex on your ChatGPT plan; or API keys sealed on the device, with spend caps per turn, per day and in total.
Web and iOS push. Documents an agent writes are kept and badged until you open them; every desk has a to-do list an auto-worker can work by itself.
The suite's roles and invites, sign-in with Digital One ID, and an org residence that gives each member an isolated seat.
Your phone, laptop and browser are live views of the same work. The agents run on a residence: your own machine, or a hosted one.
Web · iPhone & iPad · Mac in beta — Windows & Android in preview — runs on your Mac or Linux server.
A residence is a machine Builds runs on; a desk is a project folder on it, with its own agent and conversation.
Installed with one line. Your code, sessions, plans and keys stay on that machine; it connects out over HTTPS only.
$ curl -fsSL https://app.builds.one/get | sh -s -- <code>“Cozy cloud homes for your AI agents.” Always on, nothing to install, in the EU — so your laptop can finally close, and the work keeps brewing. DispatChai is where Builds' agents live; Builds is how you talk to them. It pairs with Builds automatically, and a Builds invite can come with one. Beta, by invitation.
dispatchai.cloud →The Builds hub that links your screens is run by Digital One in the EU.
Builds works on its own. Each connection below is optional — switch it on when your organisation is ready.
An org key with a daily cap per member, and Gateward's SmartFrontier router as a model choice.
gateward.ai →A card that ranks the models you actually have.
frontierscore.ai →“Ask the org”: your organisation's shared memory, inside Claude desks.
skilledmind.ai →Each product is enterprise-grade and supported on its own — and clicks into the others through Gateward, the spine of the suite, for one centrally managed backbone that learns across products.
The platform: the gateway, live model intelligence and shared memory.
The warded gateway for your models.
Sits in front of every LLM — policy-as-code on every call, PII guardrails, budget caps and cost-optimised routing, local or frontier. Every decision is sealed into a signed evidence chain; a governed MCP gateway does the same for your agents' tool calls.
Know which model is actually good right now.
Continuously tests and scores live model quality, speed and reasoning across every major provider — so you and your routers pick the best one for the task, today. The live dashboard is public; agents connect natively via MCP.
The mind your models share.
A memory & skill layer beneath your AI: it ingests what your organisation already knows, remembers it across every model and session, and grows into a specialist for your industry — with governed team & project scopes.
Governance and applied products — by invitation, each with a public demo you can try today.
Govern the flow.
Sees every AI system you run, classifies its risk, maps it to concrete EU AI Act obligations and produces regulator- and insurer-ready evidence — with cryptographic custody of Gateward's evidence chain. Start with the free public AI Act readiness check.
Evidence that AI fraud can't fake.
Turns any phone camera into tamper-evident evidence for insurance claims — a guided capture sealed into a chain of custody the moment it is taken, on your own domain. The first applied product built on the suite: its AI checks run through Gateward, under RiverAct.
Every agent carries papers.
A verifiable identity, a scoped mandate and an authority that can be withdrawn — anchored to your legal entity under European law. Know-Your-Agent for the EU.
Strict liability rewards the prepared.
Tamper-evident records of how your AI behaved, PLD defence dossiers, and the governance telemetry insurers need to price AI risk. A module of RiverAct.
The auditable AI P&L.
Built from the instrumentation you already run — counterfactual-aware baselines, evidence-linked deltas, and scale/fix/kill answers your CFO can audit. Pairs with SureWard.
Put Gateward in front of your models and the suite comes alive: FrontierScore feeds it live routing intelligence, SkilledMind serves shared memory through it, RiverAct governs and takes custody of its evidence chain, and SignWard gives every agent passing through a verifiable identity. On top, builds.one can send your coding agents through an org Gateward key. Or run any one of them on its own — they're built to drop into almost any standard enterprise AI environment.
On the same evidence spine: ClaimWard runs its AI checks through Gateward, under RiverAct · SureWard launches as a RiverAct module · LiftWard reads the spine — captured once, used twice.
All five run in your browser right now — nothing to install, no account. The products are in beta, by invitation; the demonstrations are live today. You are invited to try to break them.
Regulators and legislatures on both sides of the Atlantic are converging on the same question, in different words. The EU AI Act — Regulation (EU) 2024/1689 — requires at Article 14(4)(e) that a provider enable a human overseer, as appropriate and proportionate, to intervene in a high-risk system's operation or interrupt it through a stop button or a similar procedure that brings it to a halt in a safe state; and at Article 26(5) that a deployer with reason to consider the system presents a risk suspend its use. Both apply from 2 December 2027 for Annex III systems and 2 August 2028 for those embedded in regulated products. In the United States two bills introduced in Congress on 23 July 2026 point the same way — neither has had a vote in either chamber, and both lapse on 3 January 2027 unless they move.
Three products in the suite answer different parts of that question. Gateward refuses a call at the gate — where a rule you have written says so — before it reaches any provider. SignWard withdraws an agent's authority to act, at one mandate, one agent version or an entire organisation, and returns a proof that the withdrawal happened; it binds wherever an enforcement point checks. RiverAct sits in no request path and refuses nothing: it names which obligation applies to which of your systems and on what date, records the decision, and computes the reporting deadline if something has already gone wrong. None of them turns a model off. That duty sits with the provider who builds the system and the deployer who runs it; none of these instruments asks a governance layer to reach into somebody else's model — which is why these three work at the gate, at the authority and at the record instead.
The category is full of open-source point tools you host and hope. The Digital One suite is different on the one axis that matters to an enterprise.
Every product is enterprise-grade and supported on its own, running solo in almost any standard AI environment — no rip-and-replace, vendor-neutral by design.
Connect through Gateward into one centrally managed, supported backbone — shared control, memory, model intelligence and governance, with deeper self-learning across products and solutions tailored to your organisation. One identity, one bill, one place.
Designed, owned and supported by Digital One — enterprise SLAs, EU-sovereign deployment, real engineers behind every product. Not a project you maintain alone.
Open standards, evaluation methodologies & transparency infrastructure — to be stewarded independently by the Digital One Foundation · Amsterdam · in formation
Digital One Technology Consulting is a signatory to the EU Code of Practice on Transparency of AI-Generated Content.
Connect Claude, Cursor or any MCP-capable assistant and it can set up products, issue API keys and even join you to the FrontierScore fleet — on your behalf, with your approval, and only what you allow.
Your agent gets a scoped, on-behalf-of token through a standard OAuth 2.1 device flow. It never sees your password — you approve it once, on screen.
You choose exactly which permissions to grant — read, keys, entitlements, workers. Every action the agent takes is logged to your audit trail.
See every connected agent in your account and cut one off with a single tap. The agent's next call is refused.
Three early-stage products beyond the enterprise suite — simpler, and each made for one job. None is open yet — ask for early access on each site.
Every client's balances. In one calm place.
For accountants and accounting firms: it sorts what clients send — invoices, receipts, bank statements — checks it and prepares it for your accounting software. You approve every entry.
balances.oneWhere does your company really stand with AI? Five minutes to find out.
A short, smart AI-readiness questionnaire for leaders and their teams — each person only gets the questions that apply. Try the interactive demo on the site.
selfaudit.meMake this the last Monday you don't like.
Connects your inbox, notes and to-dos to your own AI. It prepares replies, meetings and follow-ups — then waits for your yes.
lastmonday.xyzCreate your organisation once and reach every product — or sign in if you're already in. Membership is rolling out to enterprise design partners now.