How to standardize on one AI vendor without your team going around you

Harmonic Security analysed 22.4 million AI prompts and uploads across its customers and found 665 distinct tools in use. ChatGPT accounted for 71.2% of detected sensitive exposure instances. Standardizing on one vendor is not about picking favorites. It is about making the approved option so good that nobody bothers looking elsewhere.

On this page
  1. The 665-tool problem nobody sees coming
  2. Why vendor standardization beats vendor governance
  3. The technical enforcement stack
  4. Making SSO the only door in
  5. What to do this week

Key takeaways

  • 665 AI tools in the wild - Harmonic Security found hundreds of distinct GenAI tools running across enterprise environments, in pooled customer telemetry
  • Blocking is one control - Zscaler data shows AI/ML transactions grew about 36x, with 59.9% blocked. Those figures do not measure bypass. In my view, supply beats restriction
  • Identity needs separate controls - SSO, provisioning, session revocation and audit logging each need configuration and testing
  • Start with visibility, not enforcement - You can't standardize what you can't see. CrowdStrike reports a catalog of more than 1,800 distinct AI apps across observed endpoints; prompt inspection has a separate supported-app scope

Every mid-size company I advise has the same conversation at some point. The CTO wants to standardize on Claude. The VP of Sales already bought ChatGPT Team seats. Marketing is using Jasper. Someone in finance found a PDF tool powered by an AI model nobody has heard of. Legal is panicking about all of it.

The question is always the same: “How do we get everyone on one platform?”

The answer isn’t just picking a vendor. It’s building a system where the approved option is so good, so accessible, and so deeply embedded in the workflow that going around it feels like more effort than using it. That requires an actual AI governance framework, not just a vendor decision. And then test whether the technical controls cover the routes your company manages and find where gaps remain.

The 665-tool problem nobody sees coming

Harmonic Security analysed 22.4 million prompts and uploads in its US and UK customer telemetry from January through December 2025 and what they found is deeply alarming. They found 665 distinct GenAI tools in active use across their client base. Not 6. Not 60. Six hundred and sixty-five.

Most IT teams I talk to guess their exposure at maybe 5 to 10 tools. Harmonic counted tools across its customer base, so that total does not tell you how many one company uses.

The data breakdown is painful. ChatGPT accounted for 71.2% of the sensitive exposure instances Harmonic detected. That is an association in its telemetry, not proof that ChatGPT caused those exposures. Across five AI services, 16.9% of detected sensitive instances, about 98,000, occurred on personal accounts. Corporate controls need to account for that route too.

Among detected sensitive exposure instances: Code at 26.5%. Legal discourse at 22.3%. M&A data at 12.6%. The kind of information that makes your general counsel lose sleep.

IBM’s 2025 Cost of a Data Breach report studied 600 breached organizations. Those with high shadow-AI usage had average breach costs $670,000 above those with low or no shadow AI. That association is a reason to investigate the exposure, not a price tag for every prompt.

The thing is, an exposure need not be malicious. An employee may be trying to get their work done faster. They may be pasting a code snippet to get a bug fix. They may be uploading a contract to get a summary. Productive intent can still lead to data exposure. And that distinction matters for how you solve it.

Why vendor standardization beats vendor governance

Some companies try the governance route. Approve five tools. Write policies for each. Train everyone on which tool to use for what. Monitor compliance across all five.

I’ve watched this approach fail at every company that tried it. The governance overhead alone is a nightmare. Five vendor relationships. Five data processing agreements. Five security reviews. Five sets of usage policies. Five training programs. And still, people use tool number six because their friend recommended it.

Standardizing on one vendor is a different philosophy. It’s not about restricting choice. It’s about making one choice so obviously superior that alternatives feel unnecessary.

The practical difference: governance says “you can use these five tools within these rules.” Standardization says “here’s one excellent tool that does everything you need, and it’s the only door in.” One approach creates a shadow AI prevention challenge with more vendor policies and integrations to manage. A default provider reduces that work, but its connectors, accounts and devices still need controls.

The pattern I’ve seen work is straightforward. Pick one vendor. Make it available to everyone on day one. Connect it to every tool your team already uses. Invest in training. And then enforce compliance through technical controls, not just policy. The sharpest of those controls sits at identity: stop a personal account from wearing a corporate face, so standardizing on one vendor actually means one vendor.

Does standardization on one vendor mean you’ll never use another model? No. It means your default platform is one vendor, with SSO and audit logging configured separately. Power users who need specific capabilities from other models can access them through approved API channels with proper logging. But the default, the place employees start routine work, is one platform. Standardize the door. Just keep the company brain behind it portable, an AI context layer you own rather than rent from any one vendor.

(June 2026 note: the line between “one vendor” and “one model” has blurred, and that helps this argument rather than hurting it. GitHub Copilot now spans OpenAI, Anthropic, and Google models behind a single governed surface, and Microsoft Copilot added Claude alongside its OpenAI models through the Frontier program. So standardizing on one platform no longer locks you into one model. You can offer people a choice of models inside the same SSO-gated, audit-logged door. The thesis holds. The door is still one door.)

November 17, 2026 correction: The June note’s “audit-logged door” needs a qualification. Check the chosen plan, model provider terms and logging coverage; one interface does not make those identical. SSO alone does not record every interaction.

For help setting up a shared AI workspace and reusable skills, see Blue Sheen.

The technical enforcement stack

Four categories of AI security controls configured under a security policy

Here’s where it gets specific. After the access strategy, consider four types of technical control. The diagram is conceptual; coverage depends on the applications, routes, licenses and policies you actually configure.

Layer 1: DNS and firewall rules. Inventory the web, desktop and API routes your team uses, then block unauthorized destinations on traffic you manage. Keep approved API exceptions explicit. Palo Alto Networks provides granular AI URL categories, but categories and endpoints change. Use current vendor lists and test web, desktop and API access; the application type alone does not tell you whether a network rule covers it.

Layer 2: Cloud access security. Zscaler reported AI/ML transactions growing about 36x in its February to December 2024 analysis, with 59.9% blocked. This measures growth and blocking, not how often employees bypassed a rule. Zscaler offers block, caution and isolate options. Coaching messages and browser restrictions depend on your policy and configuration. Caution mode is underrated. I would consider coaching before a blanket block where the risk allows it.

Layer 3: Endpoint detection. CrowdStrike’s March 2026 announcement describes Shadow AI Discovery for applications, agents, runtimes and MCP servers. It reports more than 1,800 distinct AI applications across observed endpoints. Falcon AIDR separately provides prompt visibility for supported applications. Check that support list and the required licenses; discovering an application does not mean every prompt in it can be inspected.

Layer 4: DLP tuned for AI patterns. Check how your DLP handles unstructured prompts as well as files and structured data. Harmonic Security uses purpose-built small language models for real-time prompt inspection. A Harmonic-commissioned Enterprise Strategy Group review described 96% fewer alerts at one investment adviser and about 75% lower deployment resources and costs at a different customer. Those are separate examples, not expected savings for every deployment. The key point: you need DLP that understands prompts, not just payloads.

No single layer catches everything. Network and cloud controls depend on traffic routing. Endpoint and content controls depend on application support and policy. Choose layers for the gaps you find, then test those routes. Buying all four categories does not by itself establish coverage.

Making SSO the only door in

This section matters more than the technical enforcement above. I know that sounds backwards, but hear me out.

SSO gives you a central authentication point. Provisioning determines who has an account, application session controls determine when existing access ends, and audit features determine what gets recorded. Configure and test each of those. Connecting an identity provider does not do all four automatically.

Without SSO, here’s what happens when your VP of Product joins a competitor: their personal Claude account goes with them. Conversations retained in that personal account may remain accessible after they leave, outside your corporate account controls. LayerX’s data-security research reported GenAI tools accounting for 32% of corporate-to-personal browser data transfers in its telemetry. That is not a measured rate for every enterprise or proof that each transfer was malicious.

The SSO gap in Claude’s pricing tiers is worth knowing about. The Team plan includes SSO through SAML 2.0 or OIDC. Claude’s provisioning guide distinguishes sign-in from account management. The Team plan doesn’t include SCIM for automated provisioning and deprovisioning. Team supports just-in-time provisioning for adding users through SSO. It does not provide the same automated lifecycle management as SCIM. Enforced SSO can block new logins when IdP access is removed, but existing sessions and organization membership still need an offboarding procedure. The Enterprise plan adds SCIM, and the bar is lower than it used to be: a 20-seat minimum if you buy self-serve, 50 through sales, billed annually.

Even with SCIM, test the timing. Microsoft says provisioning cycles typically run every 20 to 40 minutes. Applications also control their own sessions, so 40 minutes is not a guaranteed upper bound on access. Test an already logged-in account as well as a new login, and agree the acceptable revocation time for your data.

The point isn’t that SSO is perfect. For me, SSO is the starting point for managing employee sign-in. Data privacy design starts with identity. Block the network all you want, but if your employees can sign up for AI tools with their personal email on their personal phone, your network controls are incomplete for those unmanaged routes. And once SSO is the only door in, there’s a second thing to hang off it beyond access: your AI has no whoami covers loading company and team instructions from the directory group your IdP already resolves.

What to do this week

Not next quarter. Start this week. These five priorities may take longer than a day each, depending on your current setup.

Monday: Discover what’s actually running. Before you can standardize, you need visibility. Run an endpoint audit. Check DNS logs for AI-related domains. If your endpoint product and license support AI discovery, enable it and check its coverage. Don’t panic. Just know the scope.

Tuesday: Pick your vendor. If you haven’t already, make the decision. Claude, ChatGPT Enterprise, or Gemini. The choice matters less than making it and committing. The best platform is the one your team will actually use through approved channels. Whichever you pick, ensure it supports SSO, has admin controls for data retention, and offers usage reporting.

Wednesday: Enable SSO. Connect your AI platform to your identity provider. SAML 2.0 or OIDC, whichever your IdP supports. In my view, this is the single highest-impact action. Pilot the connection, enforce it after testing, and test offboarding against active sessions. Configure audit logging separately if the chosen plan supports the records you need.

Thursday: Apply the first restrictions. Use the discovery audit to prioritize unauthorized services and routes. Test DNS or cloud access rules in an observation or coaching mode where available, then enforce them. Keep exceptions for approved API use explicit and confirm the approved platform still works. Maintain the endpoint or URL-category list as vendors change it.

Friday: Communicate. This is the step everyone skips and the one that determines whether your standardization effort succeeds or gets routed around. Tell your team what you chose, why, and where to get started. Make the onboarding path dead simple. Record a 5-minute walkthrough. Pin it in Slack. The cost of your AI platform is nothing compared to the cost of people ignoring it and using personal accounts instead.

If you skip Friday, people have less reason to use the platform you chose. Turns out the technical controls are the easy part. Getting people to actually use the thing you bought? That’s the real work.

About the Author

Amit Kothari is an experienced consultant, advisor, coach, and educator specializing in AI and operations for executives and their companies. With 20+ years of experience, he is the Co-founder and CEO of Tallyfy® (raised $3.6m, the Workflow Made Easy® platform) and founding partner at Blue Sheen, which helps small and medium-sized companies get their whole team using AI. Originally British and now based in St. Louis, MO, Amit combines deep technical expertise with real-world business understanding. Read Amit's full bio →

Disclaimer: The content in this article represents personal opinions based on extensive research and practical experience. While every effort has been made to ensure accuracy through data analysis and source verification, this should not be considered professional advice. Always consult with qualified professionals for decisions specific to your situation.

Contact meMore about me

Blue Sheen helps small and medium-sized companies get their whole team using AI.

Contact me Follow 10k+