One assistant, one job, connected to your systems, with staff trained to check it
Buy now
From$3,799USD one-off
One payment. Work starts once we have confirmed your requirements.
The four questions this page answers
1
What you buy
A single AI assistant scoped to one defined job, such as answering questions from your own documents, connected to the systems it needs, with your staff trained to supervise its output.
2
What you get
6 deliverables, listed in full below
A written scope document defining the one job and the refusal boundary
A working connection to your model provider account, from Third-Party API Integration
A cleaned, indexed document set, from Document Workflow Automation
Carried out remotely by our team. Nothing is posted and no one visits.
REMOTE SERVICE
4
How long it takes
9–12 weeks from receipt of the document set
Measured from the point everything under “what we need from you” is with us, not from the moment you pay.
An assistant panel answering from your own content, with the sources it used beside it. Illustrative interface concept — not a screenshot of a client deployment.
What is this, in plain English?
A single AI assistant scoped to one defined job, such as answering questions from your own documents, connected to the systems it needs, with your staff trained to supervise its output.
A deliberately narrow first AI deployment, sold against a defined task rather than a promise. It combines Third-Party API Integration for the model provider, Document Workflow Automation to prepare the source documents, a Workflow Automation Solution for the process the assistant sits inside, Notification Automation, and three months of the Software Support Plan. Those list at $4,293 in total, so the package price of $3,799 saves $494.
The problem it solves
Your team answers the same questions from the same twenty documents every week, and the answers vary depending on who replies and whether they found the current version of the file. Meanwhile you are being told to adopt AI without anyone specifying what it would actually do.
If that does not describe you, this is probably the wrong product. Tell us and we will point you at a better fit rather than sell you the nearest one.
What you get
The concrete things that exist at the end and that you own.
6 deliverables
A written scope document defining the one job and the refusal boundary
A working connection to your model provider account, from Third-Party API Integration
A cleaned, indexed document set, from Document Workflow Automation
The assistant running inside a defined process, from the Workflow Automation Solution
Failure and rate-limit alerting, from Notification Automation
A test log showing where the assistant was right and where it was wrong, plus recorded training
What's included
The work covered by the figure in the buy box, at no extra cost.
Third-Party API Integration, listed separately from $899
Document Workflow Automation, listed separately from $999
Workflow Automation Solution, listed separately from $999
Notification Automation, listed separately at $499
Three months of the Software Support Plan, listed separately at $299 per month
The assistant configuration and the staff training session, which are produced only inside this package
Everything listed here is delivered as part of this product. Nothing on this list is aspirational.
One assistant scoped to a single named task, agreed in writing before build
Connection to a model provider through an API account held in your own name
Up to two hundred of your documents prepared, cleaned and indexed as the source material
Answers returned with a reference to the source document, so staff can verify them
A human review step for anything the assistant is not confident about
Notifications when the assistant fails, is rate-limited or returns nothing usable
A recorded training session on supervising and correcting output
A refusal boundary, so the assistant declines questions outside its defined job
How it works
The delivery sequence from order to handover.
1
Define one job
We agree the single task, the questions in scope and the questions the assistant must refuse. This document is what the build is judged against.
2
Prepare the documents
Your source material is collected, deduplicated and cleaned. Contradictory and outdated versions are found here, and that is often worth the exercise on its own.
3
Connect and configure
The model provider API is integrated against your own account, and the assistant is configured with its scope and refusal rules.
4
Test against real questions
We run a set of your genuine past questions through it and record where the answers are wrong, incomplete or invented.
5
Train the supervisors
Staff are trained on what the assistant is good at, where it fails, and how to correct it. Nobody is asked to trust it blindly.
6
Launch with review on
It goes live with human review enabled and notifications configured, and the review threshold is relaxed only once the results justify it.
Delivery & fulfilment
Stream Edge Solutions ships nothing. Every package on this site is carried out and handed over remotely — this is exactly what happens after you order.
REMOTE SERVICEHow this is delivered
1We confirm requirementsWe read what you sent with the order, come back in writing with anything missing, and agree the scope before any chargeable work starts.
2We build and configure remotelySetup, configuration and testing are done by our team over a remote connection — on your systems or on ours, whichever the package specifies.
3We hand it overYou receive the working setup, the credentials, the documentation and a walkthrough. Everything built for you is yours at handover.
Carried out remotely by our team. Nothing is posted and no one visits. Typical turnaround: 9–12 weeks from receipt of the document set.
Delivery method
Remote Setup
Expected delivery time
9–12 weeks from receipt of the document set
Measured from the point everything under “what we need from you” is with us — not from the moment you pay.
Full detail of how work is handed over, what happens if requirements are outstanding and how order records are kept is in the Delivery & Fulfilment Policy.
What we need from you
Nothing starts until these are in place, and the turnaround above is measured from that point — not from the moment you pay.
The document set the assistant will answer from, and a person who can say which version is authoritative
An account with a model provider in your own business name, or agreement that we help you open one
A sample of real questions the assistant should handle, including ones it should refuse
A named member of staff who will own supervision, since an unsupervised assistant is not something we will hand over
Who this is for
Businesses with a large body of internal documents staff must search repeatedly
Support and admin teams answering the same questions in different words
Managers asked to trial AI who need a defined scope rather than a slogan
Organisations that need a human review step built in from the start
Just as plainly as the list above: the things this package is not.
Model usage fees, which are metered by the AI provider and billed to your account directly, not by us
Training or fine-tuning a model on your data, which is different work and is not included
A second assistant or a second use case, quoted separately
Any commitment that the assistant will be correct, since no supplier can give one
Legal, medical, financial or regulatory advice generated by the assistant, which we configure it to refuse
Ongoing content curation of your document set after handover
Anything outside the scope above is quoted separately, in writing, before it is carried out. Nothing is added to your bill without your agreement.
FAQs
Will the assistant ever give a wrong answer?
Yes, and any supplier who tells you otherwise is misleading you. Language models produce fluent text that can be confidently incorrect, which is precisely why answers carry a source reference, why a human review step is built in, and why we train your staff to check rather than trust. The test log we deliver shows you the actual error rate on your own questions rather than an industry figure.
What are the ongoing costs after the package?
Two things. The model provider charges per use, metered against your own account and paid to them directly, which varies with how much staff use the assistant. The Software Support Plan is included for three months at no extra cost and then stops unless you choose to continue at the listed $299 per month.
What if the model provider changes something?
They will. Providers deprecate models, change pricing, alter rate limits and adjust behaviour with limited notice, and an assistant that worked in March can behave differently in September. The included support plan covers exactly this for the first three months, and the notification setup tells you when something has broken rather than leaving you to discover it from a customer.
Can I cancel once the documents are prepared?
Yes. If we have completed the document preparation and the API connection, those are charged at their individual list prices of $999 and $899, the package saving is withdrawn, and the remaining $1,901 of the $3,799 is refunded. You keep the cleaned document set, which is useful whether or not you proceed with an assistant.
Refund & cancellation
What happens if you change your mind, and what happens to money already paid.
Cooling-off before work starts
You have 7 days after ordering to cancel while work has not yet begun. Up to 25% may be retained to cover scoping and scheduling already carried out.
Work already done
Once delivery has started, refunds are assessed against the work completed and the deliverables already handed over — the list under “what you get” above is what that assessment is made against. Approved refunds are processed within 14 days.
If we cannot deliver what is described here
You are refunded. The scope on this page is the scope we are held to, which is exactly why the exclusions above are stated before you buy rather than after.