Insight Opens an Agent Store for Enterprise AI Work

On 12 November 2026 Insight launched an Agent Store with 15 offerings, direct deployment, and a first named client. The release lists the agents. It does not publish an independent scorecard.

PUBLISHED //
DURATION // 10 MIN READ
A quiet operations desk with a laptop, printed workflow cards, and a city window in late afternoon light
FIG.01 // VISUAL DOCUMENTATION A quiet operations desk with a laptop, printed workflow cards, and a city window in late afternoon light

Insight Enterprises told investors on 12 November 2026 that it had opened an online Agent Store. The release, also carried on Business Wire, is a catalog announcement, not a benchmark. It names six new offerings, a library that now totals 15, a deployment model that keeps Insight in the install, and one external client. Anyone reading it as proof that agent projects now finish in days should slow down. The “days to weeks” line is the company’s claim about its own service. It is not a measured result from a customer study published in the same document.

That still leaves a real shift in how a Fortune 500 integrator wants to sell agentic software. The store is not described as a self-serve download. Insight says it deploys each offering itself, either alone or inside a wider job that includes consulting, implementation, and managed services. A catalog with a services contract attached is a different product from a marketplace tile. The release is clearest when it stays on that distinction, and thinnest when it borrows outcome language it does not tabulate.

What the store actually lists

The deck line says a catalog of AI agents, enterprise-ready platforms, and integrated services helps companies achieve value in as little as days to weeks. Treat that as marketing until a client publishes cycle times. The operational sentences are more useful.

Insight is adding six agentic offerings and says the library now holds 15, after nine introduced earlier this year. The release does not name the earlier nine. It does name what is new at this launch, plus two products it wants readers to associate with the store even if they are not in the “new at launch” trio.

New at launch:

Quote Advisor looks at pricing, inventory, freight, and contract requirements, and turns a customer order into a sourced quote. That is a sales-operations job with four inputs the release bothers to list. It is also a job where a wrong freight assumption or a missed contract clause costs money. The release does not say which systems the agent reads, how a human approves the quote, or what happens when inventory data is stale.

NDA Review Agent checks non-disclosure agreements against a company’s legal policies. Again the boundary is stated and the control is not. An NDA review that “determines whether they align” can mean a checklist, a redline, or a yes-no score. Legal teams will care which. The release does not say.

Parts Order Assist is for warehouse, maintenance, and field staff. It uses an AI-powered catalog search to help with parts ordering. Field service is a plausible agent job because the user is often away from a full desktop and the catalog is large. The release does not give catalog size, error rate, or whether the agent can place the order or only suggest a part number.

Two other names sit in the store description. Insight AI Flight Academy is a gamified platform meant to speed adoption through training and change management. TokenScope is a cost tool for finance teams that measures token usage and consumption. Those are not agents that do a business transaction. One is training. One is a meter. Putting them in the same store as Quote Advisor is a clue about what Insight thinks “adoption” requires: a thing that does work, a thing that teaches people to tolerate the thing, and a thing that tells finance what the tokens cost.

The release says the offerings were first built for clients or used inside Insight, including in security, finance, and legal. That is the “Client Zero” story chief executive Jack Azagury tells in the same document. It is a reasonable way to build a catalog. It is not the same as publishing the internal before-and-after. He says Insight has driven over 90 percent internal AI adoption. That figure is his. The release does not define adoption, name the denominator, or attach an auditor. A reader can believe the company uses its own tools and still refuse to treat 90 percent as a transferable result.

Deployment is the product

The sentence that should organize any evaluation is this one: Insight deploys each offering directly for clients, on its own or as part of a broader AI transformation supported by consulting, implementation, and managed services. The store page exists so a buyer can see the catalog. The work, as described, still runs through Insight’s people.

A commentator posting on 14 November, after reading the 12 November announcement, put the same point in plainer language. Quote Advisor touches price, inventory, freight, and contract terms. The NDA agent has to follow the company’s legal policy. Insight deploys the software and can wrap it in consulting, implementation, and hosted service. That is not an install file and a goodbye. He also said he had not tested the agents and would not treat the vendor’s effect numbers as independent evidence. That caution matches what the release itself fails to supply. There is no third-party timing study in the document. There is a promise of speed.

Speed, in the release, is “days or weeks rather than months.” Months is the implied baseline of a pilot that never leaves a lab. Days or weeks is the implied baseline of a packaged offering plus a services team that has done the integration before. Both baselines can be true for different buyers. A firm that already runs Microsoft’s stack and already has a clean parts catalog is not the same buyer as a firm that still reconciles inventory in spreadsheets. The release does not split those cases. It offers one clock.

The country list is more concrete than the clock. Agent Store is generally available in 14 countries: the United States and Canada; Australia, New Zealand, and Singapore; and Austria, Belgium, France, Germany, Italy, the Netherlands, Spain, Sweden, and the United Kingdom. If your operating company is outside that list, the 12 November launch is not your launch. If you are inside it, availability still means Insight will sell and deploy there, not that a local team can turn the agents on without that deployment.

Insight describes itself as a solutions integrator, a Fortune 500 firm, with a network of over 6,000 partners, selling hardware, software, and services for more than 35 years. The Agent Store sits on that reseller-and-services history. A software lab that ships a model weights file is selling a component. An integrator that ships Quote Advisor plus a managed-service contract is selling a finished workflow and the labor to keep it pointed at the client’s data. Buyers who wanted a component will find this announcement frustrating. Buyers who have already failed an internal pilot because nobody owned the data plumbing may find the frustration familiar and the offer legible.

Culligan is the one named outsider

The release names Culligan International as the first external client to implement Flight Academy, to build AI fluency across the workforce. Read that narrowly. Flight Academy is the training and change-management platform, not Quote Advisor, not the NDA agent, and not Parts Order Assist. Culligan is not presented as a case study of those three.

What Culligan is using, according to the release, is a Microsoft set: Copilot, Cowork, Copilot Studio, Azure, and Microsoft Fabric, plus governance for AI cost and for responsibility. Laura Walsh, Culligan’s global chief information officer, is quoted saying technology does not transform a company, people do, and that generative AI adoption is far ahead of the learning curve. She says Culligan prioritized education with Insight and set up a functional champions network so adoption would spread by function and geography rather than by a single mandate.

That quote is the most specific client evidence in the document, and it is about training and a champion network, not about a measured drop in quote cycle time. A champions network is a real organizational object. It is also the sort of object companies announce when the technology roll-out is ahead of the skill roll-out, which is the gap Walsh names. Flight Academy is Insight’s product for that gap. TokenScope is the product for the finance version of the same gap, token bills that arrive before anyone has a budget line for them.

If you are scoring the announcement, separate three claims:

The catalog exists, with named jobs, in 14 countries, deployed by Insight. That claim is checkable against the release and, later, against the agent solutions page the release points to.

Insight uses the tools internally and says internal adoption is over 90 percent. That claim has one speaker, the chief executive, and no published method in this release.

Customers get value in days or weeks. That claim has no customer table. Culligan’s quoted experience is about education and a champions network, which can be true and still say nothing about Quote Advisor’s error rate.

What a buyer should ask before treating this as a transformation plan

The pressure Azagury describes is real in the sense that he is not the only executive saying pilots are stuck. This release does not prove the store unsticks them. It does tell you the questions that decide whether a packaged agent is a product or a slide.

Who approves the output? Quote Advisor produces a sourced quote from price, inventory, freight, and contract terms. Someone still has to send that quote. If the release does not name the approval step, the contract should.

Whose policy file does the NDA agent read, and who updates it when legal changes a position? An agent aligned to last year’s NDA playbook is a liability with a login.

What catalog does Parts Order Assist search, and can it order or only suggest? Field staff do not need a second search box. They need a part that arrives.

What does TokenScope count, and who is allowed to see another team’s token bill? Cost tools fail when the meter is right and the org chart is wrong.

What does Flight Academy require of managers, separate from what it asks of learners? Walsh’s champions network is the part a gamified course cannot substitute. If the local champion has no time and no authority, the course completion rate will not move the workflow.

None of those questions are in the 12 November text. They are the reading the text forces if you refuse to promote “days to weeks” into a fact. Insight’s media contact on the release is Tracey Workman, (781) 336-1789. A buyer who wants the missing control points should ask them in the statement of work, not in the press deck.

This is not investment advice, and the ticker in the dateline is not a reason to infer a revenue ramp. The release does not attach Agent Store bookings, price list, or attach rate. NASDAQ: NSIT is how the company identifies itself. It is not a forecast.

For a different cut of the same month’s agent news, the note on Meta’s acquisition of Conductor is about a coding workspace being absorbed and then shut down as a cloud product. Insight’s move is the inverse shape: a services firm putting a catalog in front of clients and keeping the deployment. One story is a lab buying a control plane. The other is an integrator refusing to pretend the control plane installs itself. Both can be true in the same fortnight. Neither one, on the documents published so far, tells you the error rate of an agent left alone with a contract template.