We build the department.You hold it.
The delivery service is how a client gets agents built for them. We design the agents against your operating model, deploy them into an APG appliance you control, and keep them running. You hold what we make: the agents, the workflows, the runbooks and the deployment they sit in. The delivery service is the route to the personal assistant: we build it as an engagement, an agent operating system connected to your own tools, with the Skáli dashboard as the example.
A delivery
We design it, install it and keep it healthy.
The delivery service is how a client gets agents built for them. We design them around how you work, run them on your own machine and keep them healthy on a retainer.
- We listen first
- We learn how the work goes before we propose anything.
- We build around you
- The agents are shaped to your process, not the other way round.
- It runs on your side
- On your own machine, with your data staying there.
- We keep it healthy
- A retainer covers watching the runs and fixing faults.
Done: handed over, and we keep it healthy
Done- Talk and scope
- Design the agents
- Install on your machine
- You approve the preview
- Hand over
Our own delivery mechanism, pointed at your operation.
Solutions Factory is a company and an internal delivery mechanism. We use it to build the agents behind our own products. The delivery service is that same mechanism turned outward.
This is where the suite stops being software you install and starts being work that gets done. A CRM with four agents is useful on its own. A CRM with four agents plus six more built for the way your business runs is a department. The delivery service is how the second one happens.
There is no Solutions Factory dashboard for your team to log into, because the deployment lives on your infrastructure. We operate it with you, and everything we build stays with you.
We find the work, build the agents, deploy them and run them.
- 1
We start by mapping your operating model and finding the work an agent should carry. That means the manual, repetitive, error-prone parts that are worth automating. Some of what you do should not be automated, and we will say so.
- 2
Then we build on APG against that model. The runtime discipline is already in place: validated output, cost caps and an operator-confirm gate on anything that leaves the building.
- 3
We deploy into your own APG appliance, on your inference tier. Nothing routes through us unless you choose it.
- 4
After that, the retainer keeps the fleet healthy and lets it grow as the operation changes. You hold the runbook the whole time, including the day you decide to run it without us.
From one agent to a whole department.
Every engagement is a setup fee plus a monthly retainer, with inference billed separately on the tier you pick. The tier names describe scope and do not map to fixed prices.
Light
A single agent deployed into your APG, with a retainer to keep it healthy.
Standard
A small fleet covering one workflow end to end.
Heavy
Several workflows with multi-agent orchestration, operated under the terms agreed for the engagement.
Custom
A bespoke fleet built against an operating model that does not resemble anyone else's.
We scope it on the discovery call, because the answer depends on what the work is. See Infrastructure for how inference is metered and Pricing for the fee shapes across the suite.
Tier 3 runs on hardware you own.
Tier 3 Sovereign runs the model on a machine you own, and the sale is software only. There is no inference bill from us and none from a model provider. You pay for the hardware and the electricity, and nothing else.
Getting there is part of an engagement. We do not sell the machine as a product. We specify the machine against the agents you plan to run, size the model and its quantization to it, install it and tune it. That is a project with a fee, and the license and support line is the same one every other engagement carries.
We do not publish a hardware specification on this page, because the right machine depends on which agents you run and on what the model needs to hold in memory.
This is a build. You cannot sign up for it.
There are products that give you a working assistant this afternoon for a subscription. If what you need is a generic helper and you do not mind it living on someone else’s platform, those products are a reasonable buy, and we will tell you so on the call.
The delivery service is the other thing. It takes discovery, it takes a build, and it takes an operator on our side who stays with it. What you get is a department shaped like your business, running on hardware you control, that keeps working when a vendor changes its terms.
Tell us what the work is.
A discovery call is the fastest way to find out whether an agent should carry a piece of your operation, and what it would take to run it on your own APG.