Keep data where it belongs
It has to run on hardware inside the building, because customer data does not leave.
About ToggleLogic
ToggleLogic decides what runs where, under whose policy, and leaves a record of why.

It ran in production before it was a company.
Access was never the hard part
Once a company uses more than one model, getting to the models stops being the problem.
Most AI stacks cannot answer those questions. None of them answer before execution, which is the only moment an answer is worth anything.
ToggleLogic answers all four. It evaluates each request against policy the company sets, selects the model and the execution environment, and writes what it did to a record: the request, the policy that applied, the model chosen, the cost, and the human override if there was one.
It started in a business, not a lab
ToggleLogic did not begin as a product idea. It began inside an IT services company that has operated on Main Street in Chagrin Falls, Ohio since 2012, at the point where that company started putting AI into its own operations and into the businesses it serves.
The audit record was not a roadmap item. It existed because someone had to answer for the bill, and for where a customer's data went, in a working business, with real customers, every day.
That origin set the constraints that still define the product.
It has to run on hardware inside the building, because customer data does not leave.
It has to be operable by someone who is not an engineer, because the person responsible is usually not one.
It has to prove what it did, because a company that serves other companies gets asked.
Where it runs today
ToggleLogic Free is published on the ClawHub registry with public source on GitHub. Every release is listed on the changelog.
ToggleLogic Intelligence routes live in our own operations. It selects the model for real work every day and records why. That is daily use in a working business, not a pilot and not a staged demonstration.
The routing itself is no longer the hard part. Basic model gateways are becoming a commodity. The hard part is the policy layer above them, the evidence underneath them, and fitting both into the AI stack a company already runs.
That is the company.
Built for the middle
ToggleLogic is built for mid-market companies: large enough that AI decisions carry real cost and real exposure, small enough that there is no platform team to build a governance layer in-house.
That market is not served. The tools built for the largest enterprises assume staff those companies do not have. The tools built for developers assume expertise those companies do not want to hire. ToggleLogic is built to be owned and operated by the business itself.
An independent company, starting clean
ToggleLogic is being formed as a Delaware C corporation in September 2026, separate from the operating business where it was built. The intellectual property will be assigned to the new company by Al Harlow personally, under a written assignment agreement.
The company starts greenfield. No legacy platform to maintain, no acquired codebase to reconcile, no prior obligations. What it does carry is the part that is hard to acquire: a product that has already run in production, and the operational judgment of the business that had to live with it.
Leadership
Al Harlow has built five companies since 1984 and is ToggleLogic's first customer. The product ran his own business before it ran anyone else's.
That is deliberate. A company being built for where it is going should be staffed for where it is going.
Evaluate ToggleLogic
Map your models, execution environments, business policies, and the evidence your organization needs before routing can change.
Start a conversation →