The platform
Developed at Vanderbilt University
open source (MIT license)
supported in partnership with AWS
deployed at more than 50 institutions
runs in your own cloud account
model choice
approved sources
usage controls
Start with a no-fee scope assessment. It produces a short implementation brief your CIO, counsel, security team, and procurement office can review, before you commit to an implementation.
A generative-AI platform you deploy and control, not another subscription you can’t govern
Amplify GenAI is an open, vendor-independent generative-AI platform. (It is not AWS Amplify, the app-development service; that is an entirely different product with a similar-sounding name.) We implement and customize it so your institution gets the capability of a modern AI assistant while preserving the controls consumer accounts often lack:
You control the deployment.
It runs in your own cloud account, so your data and your AI usage sit in an environment your institution administers.
You approve the sources.
It answers from the documents and data you designate, with retrieval over your own content, so answers trace back to material you trust.
You choose the models.
It is model-independent, so you are not locked to one provider’s roadmap, pricing, or terms.
A person stays in the loop.
Usage controls, per-user limits, cost accounting, and review are part of how the platform is built to be operated.
A proven open-source platform from higher education
Amplify GenAI was developed at Vanderbilt University and is open source under the MIT license, supported in partnership with Amazon Web Services. Vanderbilt reports the platform in production at more than 50 institutions and organizations, including the University of Texas at Dallas, the University of Missouri, and the University of Montana, alongside K–12 districts and other organizations. The software you would adopt is established and actively developed; what Watkyn adds is the work of implementing, customizing, and supporting it for your institution. → Learn about the platform at amplifygenai.org (opens in a new tab) and see the open-source code (opens in a new tab).
AI is already in your organization. The question is whether you can see it.
People across most organizations are already pasting documents into consumer AI tools to draft, summarize, and answer questions. It is useful, and it is almost always ungoverned. Sensitive material leaves through accounts nobody administers, answers come from sources nobody approved, and there is no record of what was asked or sent. For an institution with confidentiality, FERPA, or contractual obligations, that exposure grows every month it goes unaddressed, whether or not anyone has noticed it yet.
Banning the tools rarely works, because the need is real. The durable answer is to give people a capable AI platform that runs in an environment you administer, answers from sources you approve, and leaves a record you can review. The goal is not only to govern AI use. It is to give people a tool good enough that they prefer it to the ungoverned workaround, so the capability and the control arrive together. Getting that right is an engineering and governance problem more than a software purchase.
Especially relevant for higher education
On a campus, generative AI is already in use across advising, admissions, teaching support, research, HR, policy review, and grant administration. The real question is no longer whether people have access to AI. It is whether the institution controls the sources those tools draw on, handles student and personnel data in a FERPA-aware way, sets the model-provider terms, governs usage, and keeps visibility into what is happening. Amplify GenAI gives an institution one platform to put those controls in place, and our implementation work turns the open-source platform into a deployment your IT and governance teams can actually run. → More on how we work with colleges and universities: see our software for higher education.
Where it tends to land first
Advising and student services
Admissions
Teaching and course support
Research workflows
HR and policy
Sponsored programs and grant administration
Where institutions often start
Common first projects for a campus Amplify GenAI deployment. Final use cases are selected during the scope assessment.
Campus policy and procedure assistant
Answers staff and student questions from your approved handbooks, policies, and procedures.
Advising and student-services assistant
Helps advisors and front-line staff find accurate answers from approved program and services material.
Admissions operations assistant
Supports admissions teams with consistent, source-grounded responses to common applicant questions.
Sponsored-programs and grant-administration assistant
Helps research administrators navigate sponsor rules, internal procedures, and deadlines.
HR and internal-policy assistant
Gives employees governed answers about benefits, policies, and processes from approved HR sources.
Research-administration document assistant
Drafts and summarizes from approved templates and documentation, with a person reviewing.
A controlled platform only delivers value if people actually use it. We treat administrator training, source-owner workflows, and user rollout as part of the implementation, not an afterthought, so the platform people adopt is the same one your institution governs.
Start with a bounded first deployment
Most institutions should not begin with a campus-wide rollout. A safer first step is a focused deployment: one or two high-value use cases, a defined set of approved sources, a small group of administrators, single sign-on, usage controls, and logging, with a clear plan for handing the platform to your internal team. That lets IT, academic leadership, information security, and the administrative owners see how Amplify GenAI works in practice, on your own data and in your own environment, before deciding whether and how to expand. You prove the value and the controls together, at a scale you can manage.
What a first deployment leaves you with
A focused first deployment should leave your institution with:
A working Amplify GenAI environment running in your own AWS account
One or two configured campus use cases, on your approved sources
Single sign-on and role configuration
Usage and cost controls, with logging you can review
Documentation of the build and its dependencies
Administrator and source-owner training
A rollout plan for deciding whether, and how, to expand
A new Amplify GenAI practice, built on work we have already done
We have not yet completed an Amplify GenAI implementation for a client. The capabilities an implementation demands are not new to us: hands-on command of the platform itself, secure cloud deployment, controlled data pipelines, and production AI assistants that draw only on approved sources. Here is the foundation.
Our own hands-on Amplify GenAI work
We run Amplify GenAI ourselves, and we have worked inside its codebase.
We know this platform from the inside. Our team has deployed Amplify GenAI internally and put it through thorough, hands-on evaluation: standing it up in a cloud environment, connecting approved sources, testing model options across providers, and exercising the usage and review controls in practice. The work has gone beyond the configuration layer. We maintain our own fork of the open-source codebase with improvements we have made on top of the official version, and we intend to contribute those improvements back to the project. We work at the level where a real implementation succeeds or fails.
Why it matters here: An institution-controlled AI platform is only as good as the engineering behind its deployment. On a fit call we can discuss the platform candidly and, where appropriate, walk through what we have learned from running our own build in a test environment.
The delivered work an implementation rests on
Production AI assistants, on cloud infrastructure we built and operate.
We have built and operate production AI assistants on AWS for a national nonprofit, drawing only on approved sources, on a controlled set of models, with a person in the review loop. That is the same shape an Amplify GenAI implementation takes: governed sources, controlled models, human oversight, on secure infrastructure. Alongside it sits years of cloud-architecture, data-engineering, and systems-integration delivery under formal contracts. → See our software delivery methodology and government delivery record.
How we think about governance
We design implementations against recognized AI-governance frameworks.
This includes the ISO/IEC 42001 AI management system standard and the NIST AI Risk Management Framework. This means source approval, model selection, oversight, documentation, and review are structured to a known reference rather than improvised. We design to these frameworks; we do not represent the institution or Watkyn as certified to them, and we scope the specific controls with you in discovery.
We are now applying that hands-on platform knowledge and delivery discipline to Amplify GenAI implementation, for colleges and universities that want their own governed generative-AI capability. As we complete Amplify GenAI engagements, their case studies appear here first.
How we implement Amplify GenAI
An implementation follows the same disciplined method we use everywhere: understand your governance obligations and use cases first, design the deployment and controls deliberately, then build, test, deploy, and support. The aim is a platform your team can run and govern after handover, not one that depends on us.
1
Discovery
2
Design
3
Build
4
Test
5
Deploy
6
Support
What an implementation typically includes:
Cloud deployment architecture in your own AWS account (Terraform / Serverless Framework)
Identity and single sign-on integration (SSO / SAML / your identity provider)
Roles, permissions, and per-user usage and cost controls
Model-provider setup, with provider terms identified and documented for your privacy, security, and counsel review
Approved-source ingestion and retrieval (RAG) design over your own documents and data
Custom assistants configured for your workflows
Logging, usage accounting, and cost monitoring
Security review aligned to your obligations
Documentation, admin training, and user rollout
Ongoing support and the handoff to your internal team
→ The full method, including documentation and training at handover: see our software delivery methodology.
How an engagement begins
1
Amplify GenAI fit call
A short, no-obligation conversation: your situation, your obligations, and whether an institution-controlled platform is the right answer. Where useful, we walk through what we have learned running our own build of Amplify GenAI in a test environment.
2
Scope assessment (complimentary)
We work through your use cases, data sources, identity and security requirements, model and cost considerations, and governance needs, and produce a short implementation brief: recommended first use cases, a data-source inventory, identity and single sign-on requirements, privacy and model-provider questions, AWS deployment assumptions, governance responsibilities, the main cost drivers, and a recommended implementation path. It is an artifact you can take to your CIO, counsel, and procurement. There is no fee for this step.
3
Implementation proposal
A defined plan and scope you can act on, with the deliverables above mapped to your environment.
Who should be in the room?
A useful first conversation usually includes IT or cloud infrastructure, information security or privacy, an academic or administrative sponsor, and one or two owners of the first candidate use cases. Counsel and procurement can join later, as the project moves toward implementation.
What an implementation needs from you
Implementations move fastest when these are in place. We help you assemble them if they are not.
An AWS account your institution controls (or willingness to set one up)
Access to your identity provider for single sign-on
A few candidate use cases and the people who own them
Owners for the data sources you want the platform to draw on
Your model-provider preferences and any required terms
A governance sponsor and an IT/security contact
An internal administrator who will own the platform after handover
Watkyn LLC is a software development and consulting firm, operating since 2015, that offers Amplify GenAI implementation: the deployment and customization of Amplify GenAI, an open-source, vendor-independent generative-AI platform originally developed at Vanderbilt University that an organization runs in its own cloud environment. An Amplify GenAI implementation lets an organization use generative AI on infrastructure it controls, drawing on approved sources, across a chosen set of model providers, with usage controls and human review built in; data-privacy outcomes depend on the organization’s agreements with its cloud and model providers, which the implementation is configured to support. Watkyn’s team runs Amplify GenAI internally and maintains its own fork of the open-source codebase with improvements, and Watkyn’s adjacent delivered work includes production AI assistants built and operated on AWS for a national nonprofit, alongside years of cloud-architecture, data-engineering, and systems-integration delivery under formal contracts. Watkyn has not yet completed an Amplify GenAI implementation for a client and frames the service accordingly. Watkyn is a Quickbase partner and a Caspio Gold Partner and works remotely with organizations across the United States.
Is an institution-controlled AI platform the right answer for you?
We would rather point you to the right fit than sell you a platform you do not need.
Amplify GenAI is likely right when your organization handles sensitive or confidential material, has real governance or compliance obligations (FERPA, contractual confidentiality, public-records duties), and wants generative-AI capability that runs in an environment you control, with answers traceable to sources you approve.
A lighter approach may serve you better when your needs are already met by a governed enterprise subscription with adequate controls, or when a single workflow rather than a general AI capability is the real requirement. We will tell you plainly which situation you are in. → If a specific application is closer to what you need, see our custom software development. We also work with nonprofits and associations on governed AI.
Common questions about Amplify GenAI implementation
Amplify GenAI is an open-source, vendor-independent enterprise generative-AI platform an organization deploys and runs in its own cloud environment. It answers from sources you approve, works across model providers you choose rather than locking you to one, includes usage controls and per-user cost accounting, and is built to keep a person in the review loop. It was developed at Vanderbilt University and is in use at more than 50 institutions and organizations. This is not AWS Amplify, the application-development service; that is an entirely different product with a similar-sounding name. We implement and customize Amplify GenAI for your organization.
You can. Amplify GenAI is open source, and a strong internal cloud team can deploy it directly. The harder problem is not getting the software running; it is running it as institutional infrastructure: AWS architecture, identity and single sign-on, approved-source design, model-provider configuration, privacy and security documentation, usage and cost controls, administrator training, and a handoff your team can sustain after the person who set it up moves on. Internal developers or student workers can often stand up a prototype. Making the platform governable, reviewable, documented, and safe to expand is where most of the real work and most of the risk live. We build it to be yours to run, not dependent on us to operate.
When you already have cloud staff with available time, AWS, Terraform, and Serverless experience, identity-provider access, security-review capacity, and someone who will own documentation, rollout, and ongoing administration, self-deployment can be the right path, and we will say so. Watkyn is the better fit when you want to preserve your internal team’s capacity, lower implementation risk, and move from evaluation to a governed first deployment without it becoming an open-ended internal project.
That is part of the implementation problem. We start with use cases where the institution has approved sources and people have a real daily need, configure assistants around those workflows, train administrators and source owners, and plan a rollout so the governed platform is genuinely useful, not just compliant. A platform people ignore solves nothing; the aim is one good enough that it replaces the ungoverned workaround on its own merits.
Not yet. What is proven is the work an implementation rests on: our own team runs Amplify GenAI internally and maintains its own fork of the codebase with improvements, we have built and operate production AI assistants on AWS for a national nonprofit (approved sources, controlled models, a person in the review loop), and we have years of cloud-infrastructure and systems-integration delivery behind us. If your organization wants its own governed generative-AI capability, we will give you an honest read on scope and fit first.
We design implementations against the ISO/IEC 42001 AI management system standard and the NIST AI Risk Management Framework. In practice that means source approval, model selection, oversight, documentation, and review are structured to a recognized reference rather than improvised. To be precise about what that does and does not mean: we design to these frameworks, and we do not claim that Watkyn or your institution is certified to them. Certification is a separate process through an external auditor. If formal certification is a goal for your institution, we will be clear about where implementation work ends and a certification effort would begin.
Amplify GenAI is open source, so there is no platform license fee. The cost structure is implementation work, your cloud (AWS) hosting, the usage cost of the AI models you choose, and ongoing support. Because you pay for the models and capacity you actually use, the economics depend heavily on how many people use it and how. We estimate those costs with you during discovery based on expected users, data volume, model providers, security requirements, and support expectations. The complimentary scope assessment can include a preliminary cost model showing the main cost drivers, so you have something concrete to take to finance before deciding whether to proceed.
Because the platform runs in a cloud environment you control and connects to models on terms you set, your data is not handed to a consumer product by default. We design approved-source access, retrieval, and model-provider configuration so the platform draws on the material you intend and nothing else, and we help you put the agreements and controls in place that your obligations require. We do not set your FERPA policy or legal position; your institution and its counsel decide those. We implement the technical controls, documentation, approved-source design, and provider configuration that support them. Data-privacy guarantees ultimately rest on your agreements with your cloud and model providers; most organizations can obtain terms that include not training on their data, and we configure the deployment to support that.
They are separate. Amplify GenAI implementation is a service we offer today: we deploy and customize an existing open-source platform so your organization has governed AI capability. The AI governance platform is a separate software product Watkyn is developing; it is in development and not yet available. An Amplify GenAI implementation does not depend on it.
You control the deployment: Amplify GenAI runs in your own cloud account, and your institution administers the environment. Data handling depends on the sources you connect, the access controls you configure, and the cloud and model-provider agreements you put in place. We document those dependencies during implementation so your team understands exactly what determines the privacy and handling of your data, and we configure the deployment to support the terms you require.
Yes. We can provide security and compliance documentation, describe the implementation architecture, identify subprocessors and model-provider dependencies, and work with your IT, privacy, and procurement teams. Where regulated data is involved, your governing agreements and institutional policies set the requirements; we implement the technical and documentation controls that support them. If your institution requires a formal security review before adopting a new platform, we plan for it as part of the engagement.
Amplify GenAI is designed for model choice across supported providers, so you are not tied to one. We confirm the current provider options during discovery and configure the models that fit your needs, cost constraints, and policy requirements, and you can change that selection as the landscape moves.
No. Colleges, universities, and agencies have the clearest governance obligations, so they are a natural fit, but any organization that handles sensitive material and wants governed AI capability is a fit. We work with nonprofit and commercial organizations as well.
Let’s talk about governed AI for your institution.
Tell us how AI is being used across your organization today and what you need to keep control of. We will give you an honest read on whether an institution-controlled Amplify GenAI platform is the right fit, and what an implementation would involve.
Prefer to talk?
Want more first?
See our software delivery methodology and our security and compliance practices.
