The questions you would ask on the second call.
Answered here instead, in the order they usually arrive: what the platform is, what the agents actually do and where they stop, how a rollout runs, what it costs, and what a security review will find. Nothing below needs a form first.
43 questions·7 topics·No form to read them
Seven topics. Jump to the one you came for.
Every answer is open in the page rather than behind a search box, so ⌘F works on the lot of them.
What VEYQON is, and what it replaces.
What is VEYQON, in one sentence?
One system of record and one system of intelligence: an open ERP that runs finance, supply chain, people, and projects, with AI agents that read the same ledger and act on it inside limits you set.
Is this an ERP with AI bolted on, or an AI product with an ERP attached?
Neither. The agents are not a separate product with a connector between them and your data — they hold a role and a toolset on the same records, the same permissions, and the same audit trail as everybody else. That is the whole reason an agent here can check a credit limit or post a journal, and a chat assistant over an export cannot.
Which modules are included?
All twenty, in every plan: accounting, CRM, sales, procurement, inventory, manufacturing, projects, HR and payroll, assets, quality, support, e-commerce, POS, and the rest. There is no accounting edition and no manufacturing add-on. New modules arrive in the release rather than on a price list.
Do we have to replace everything we run today?
No, and most rollouts do not. One real process goes live first, the systems you are keeping stay connected, and the rest of the business is added around something that already works. Modernisation without a rewrite is a supported route, not a compromise.
Is it built for our industry?
There are twenty-four industry configurations — manufacturing, healthcare, construction, facilities management, retail, logistics, education, and others — each of which is the same platform with the module stack, terminology, and workflows a sector actually uses. Nothing about it is a fork, so you still get every platform release.
How many people does it take to run?
The platform is administered from the interface, not from a console: roles, workflows, fields, and reports are configuration. Most customers run it with the team they already have, plus an implementation partner for the first quarter.
What the agents do, and where they stop.
What does an agent actually do?
It holds a role, a toolset, and a limit. An invoice agent extracts lines, resolves the supplier, runs a three-way match against the purchase order and goods receipt, and posts when the variance is zero. A procurement agent issues RFQs, normalises quotes, and drafts a purchase order against the best landed cost — and never sends it, because a buyer signs.
What stops an agent doing something it should not?
Agents sit inside the same permission model as people, with row- and field-level rules applied before inference rather than after. Each one has an explicit list of permitted actions and hard ceilings on value, period, document type, and volume that cannot be crossed however it is prompted.
Can we see why an agent did something?
Every agent action is on the audit trail with its inputs, its reasoning, and the specific limit that permitted it. That record is the same one your auditor reads for human actions — there is not a second, softer log for the AI.
Do you train models on our data?
No. That is a contractual term rather than a setting, and it covers every model option we offer.
What happens when an agent is not confident, or is simply wrong?
Anything outside its authority is routed to a person with the exception attached, which is the normal operating mode rather than a failure state. If an agent posts something incorrectly it is reversed the way any other posting is reversed, and the trail shows both entries.
Can we bring our own model, or run inference in our own region?
Inference can be pinned to a region alongside your data and backups. On Enterprise deployments, model choice — including self-hosted open-weight models — is part of the architecture conversation rather than a fixed answer.
How an implementation actually runs.
How long does it take?
Weeks, not financial years. One real process runs in the first fortnight on your own data, and the remaining scope is sequenced around it. A mid-sized rollout across finance and one operational area typically reaches steady state in a quarter.
Is there a big-bang cutover?
Not by default. Processes move one at a time, with the old system still standing until the new one has run a full cycle. Where a hard cutover is genuinely simpler — a single-entity finance migration at year end, for instance — we will say so.
What do you need from our team?
A decision-maker per process area who can answer questions the same week, and someone who knows where the data actually lives. That second person matters more than most people expect.
How does our data get in?
Masters, open balances, and history come in through mapped imports that are run repeatedly against a sandbox tenant until the reconciliation is clean, then run once more for the real thing. You see the reconciliation, not a claim about it.
What about training?
Role-based sessions close to go-live rather than months before, plus in-product guidance for the paths people take daily. The people who will use the platform have usually already seen their own process running in the sandbox by then.
What happens the day after go-live?
Hypercare for the first cycle — the people who built the configuration are the people watching it run, not a handover to a queue. Once a full month has closed cleanly, the account moves to normal support.
Two lines on the invoice: people, and agent capacity.
Who counts as a user?
Anyone who signs in and changes something — posts, approves, dispatches, or edits a record. External auditors, inspectors, and anyone with read-only access are free. Customer and supplier portal accounts are free and unlimited.
Do you charge per module?
No. All twenty modules ship in every plan. There is no accounting edition, no manufacturing add-on, and no unlock fee for the twenty-first module when it ships — you get it in the release.
What happens if we run out of agent capacity?
Agents stop proposing work and the queue goes back to your team exactly as it was before. Nothing is dropped and nothing is lost. You will have had a warning at 80%, and you can add a capacity pack at any point in the month.
Is there a minimum term or seat count?
Ten users, and a twelve-month term on annual billing. Monthly billing runs month to month with thirty days' notice. Enterprise contracts are annual by default and multi-year by negotiation.
Can different parts of the business be on different plans?
Yes. Seats are assigned per user, so a finance team can sit on Intelligence while a depot runs on Standard, on the same tenant and the same ledger. Agent capacity pools across everyone on Intelligence.
What does an implementation actually cost?
Most mid-sized rollouts land between €6,000 and €25,000 of services, depending on how much data has to move and how many processes go live at once. We quote a fixed scope before you sign, and the quote survives the project.
Can we self-host?
On Enterprise, yes — your own hardware, your own cloud account, or a single-tenant environment we run for you. The platform is the same build in all three cases, and the exit licence is written into the contract.
What happens to our price at renewal?
Your seat rate is fixed for the term. At renewal it moves by no more than the published list increase, and never because you filed more documents, added a module, or connected another system.
Do you discount for education and non-profits?
Yes — a standing discount on seats for registered charities, schools, and public healthcare. Capacity is priced the same, because the compute costs the same.
Can we try it before we commit?
Bring one process to a 45-minute session and we will build it live on the platform with your own document as the input. If it survives that, we set up a sandbox tenant with your data in it.
The questions a security review opens with.
Where is our data held?
In a region you choose, with data, backups, and inference pinned to it. The current sub-processor list is published, and it changes with notice so your own DPIA stays accurate.
What certifications do you hold?
SOC 2 Type II, reported annually and available under NDA, and ISO 27001 with the statement of applicability. Independent penetration testing runs at least annually and after significant architectural change, with a shareable summary letter.
How does it fit our identity provider?
SAML 2.0 and OpenID Connect for sign-in, SCIM for joiners, movers, and leavers, and MFA enforced by your policy rather than ours. Provisioning through SCIM is what makes access review tractable at audit time.
Can your staff see our data?
Support access to your instance is opt-in, time-bound, logged, and visible to you. It is not a standing privilege, and you can see every time it has been used.
What happens if we leave?
Open-format export of your records and documents, a documented deletion process with a certificate, and — on Enterprise — an exit licence written into the contract. Leaving is a procedure, not a negotiation.
Can we run our own penetration test?
Yes. Testing your own instance is permitted through a notification process rather than prohibited by the terms.
How it meets the systems you are keeping.
What can it connect to?
Over 300 connectors covering bank feeds, e-invoicing, carriers, payment providers, and the tools your teams already use. A connector is configuration rather than a project.
What if the system we need is not on the list?
REST APIs, webhooks, and typed SDKs cover everything the interface can do, because the interface uses the same APIs. Where a system only speaks files, scheduled import and export pipelines are part of the platform rather than a script somebody has to maintain.
Can we extend the data model?
Yes — custom fields, record types, and relationships, with permissions and the audit trail applying to them automatically. Extensions survive upgrades; that is the point of doing them in the platform rather than beside it.
Is any of it open?
The foundation is open and documented, so the data model is yours to shape and nothing about your records is a proprietary format you cannot read without us.
Does it work on a phone?
There are offline-first mobile apps for ERP, HR, and Agent Q. Crews on site capture work without a signal and the records reconcile when they get one.
Who you reach, and what happens when something breaks.
What support is included?
Every plan includes support from people who work on the platform. Enterprise adds a named engineer who knows your deployment rather than a queue position.
How do upgrades work?
Releases are applied without a re-implementation, and your configuration, extensions, and integrations carry across. There is no version you have to be talked into paying to leave.
What is your uptime, and where can we see it?
A public status page carries current state and incident history, including the incidents that were our fault. Recovery objectives are stated as numbers and tested, with the results available to you.
Who do we call during an incident?
A documented incident response plan with defined severities and named roles, plus breach notification timelines committed in the contract rather than left to best endeavours.
Next step
Bring one process. We will run it live.
Pick the process that costs you the most time this month. In a 45-minute session we will build it on the platform with your own document as the input.
Or start reading — docs.veyqon.com