Your company’s AI, under control
The AI Control Platform that lets teams keep using their preferred AI tools, with company login, data protection and per-user budgets built in.
Overview
This month · all users
Claude Sonnet 5.546%- Claude Opus 5.527%
Kimi K317%
OpenAI Astra10%
- Amir C.Claude Code · Claude SonnetAnswered
- Dana L.Claude Desktop · Claude OpusMasked
- Noa P.Claude Desktop · GeminiAnswered
Users
Synced from Entra ID · sign-in with your work account
- Dana LeviLegal Admin Active
- Amir CohenEngineering User Active
- Noa PeretzFinance Viewer Active
- Yossi MizrahiEngineering · contractor User Active
Remove someone in your directory and their access ends within about a minute. In this demo it is sped up.
Budgets
A monthly cap per user, with alerts on the way
- 70% reachedAlert email sent
- 90% reachedAlert email sent
- 100% reachedRequests paused
Your monthly AI budget is used up. Your admin can raise it.
Agents
Each agent has an owner, a model and a budget of its own
- Invoice triageFinance · Claude Sonnet
n8n 42% Active - Release notesEngineering · Sol 5
LangChain 67% Active - Contract reviewLegal · Claude Opus
Strands Agents 88% Active - Ticket summariesSupport · Gemini
Make 23% Active
Agent requests count against the agent’s own budget, and every one is logged.
Guardrails
Every prompt is checked before it reaches the model
Summarise my call with [NAME] (ID [ID NUMBER], [EMAIL]) about the overdue invoice and draft a polite reply.
Blocked. The person is asked to rephrase. Nothing reaches the model.
Secrets are always blocked.
Tools and models
Choose what your organisation can use
Model choices are enforced on the server, in every tool.
Requests
Every request, tied to a named person
All your AI traffic goes through one layer
Every request is checked, sent to a model you allowed, and recorded.
- One invoice All your models and tools, on one bill.
- A budget that stops A ceiling per person, with alerts on the way.
- Names and IDs masked IDs, names and card numbers masked or blocked.
- Who uses AI, and how much By person, team and model, any day of the month.
One AI gateway for every tool
Your people keep the tools they use and sign in with their work account, without a new app or personal API keys.
ClaudeIllustrationSummarise the Q3 pipeline for the board
Here is a one-page summary, grouped by stage and owner.
- Won: 9 deals, most of them enterprise renewals
- Negotiating: 6 deals, two waiting on legal
- Next: owners confirm their forecasts by Friday
J. Doe · signed in with MicrosoftClaude Opus
Claude Code ~/invoice-serviceIllustration› add tests for invoice_export.py
Reading invoice_export.py…
Writing tests/test_invoice_export.py
6 passed in 1.4s
J. Doe · signed in with MicrosoftClaude Sonnet
Codex ~/reportsIllustration› refactor the report builder
Planning changes in 3 files…
Updated builder.py, export.py, tests
Ready for review
A. Cohen · signed in with OktaSol 5
Cursor payments.tsIllustration12export function charge(order) {
13 validate(order);
14 const total = sum(order.items);
15 return gateway.pay(total);
16}
Add input validation
Checked charge() and its callers
Edited validate() · 2 lines
M. Rosen · signed in with MicrosoftKimi600+ models
AnthropicClaude Opus, Sonnet, Haiku- OpenAISol 5, Luna, Astra
GoogleGemini- MoonshotKimi
ZhipuGLM
DeepSeek- + 600 more
Any AI-gateway client
- Claude Desktop
- Claude Code
- Codex
- Cursor
Windsurf
Cline
Roo Code
Kilo Code
OpenCode
Goose
OpenClaw- + more
Security and control
What your security review will ask, including what neo does not do.
-
A dedicated environment for every customer. No stack, model account or database is shared with anyone else.
-
SSO and SCIM with Microsoft Entra ID or Okta. Remove someone in your directory and their access ends within about a minute.
-
Every prompt is screened before it reaches a model. Personal data is masked or blocked, secrets are always blocked. Responses are not screened.
-
Each request is recorded with the person, team, tool, model and tokens, so you can always answer who used what.
-
Encrypted in transit and at rest, operated under ISO/IEC 27001 and 27017, and tested externally in August 2026.
- Hosting
- Flexible to your needs, including where your data lives.
- Screening
- Prompts only, not responses.
- Managed
- We run the platform; your admins set access, budgets and data rules.
Transparent AI pricing
Provider prices.
Zero markup.
Pay for the AI you use. Pay neo for the platform. Simple as that.
AI usage itemised on your invoice.
See how billing worksGive your team AI.
Take back control.
See neo in action, talk through security, and leave with a clear rollout plan.
Book a sessionWe reply within one business day.
- “What are we using now?”We map your teams’ tools, models and spend.
- “Can I see it?”The console live, with example data.
- “Is it safe?”Hosting, logs, prompt checks and offboarding.
- “Where do we start?”A written rollout outline for your organisation.
FAQs
Will our teams have to switch AI tools?
Teams can continue using supported AI tools they prefer, with company login, data protection and budgets managed through neo. Compatibility and setup requirements depend on the tools and workflows your teams use.
We already have enterprise AI subscriptions or an AI gateway. Why would we need neo?
neo brings model access, business-system permissions, security policies and budgets into one platform. Its value depends on what your current setup already covers and where you need shared controls across teams and providers. We can help you assess that fit.
How does neo help protect company data?
neo provides configurable security guardrails to help detect and restrict sensitive data in supported AI interactions. These controls apply to usage routed through neo. They do not automatically cover separate accounts or workflows that bypass the platform.
Can we control AI spending, rather than just track it?
Yes. neo supports user and team budgets, spending limits and threshold alerts for consumption managed through the platform. Your admins can also review usage and costs by user, team and model to understand where spending comes from.
Can we connect AI to company systems without giving it unrestricted access?
neo lets admins define which supported connectors, tools and operations each team can use. For example, a team can be permitted to read records through a connector without being granted access to update them.
How do we manage access when people join, change teams or leave?
neo supports enterprise SSO and SCIM provisioning, alongside centralized user and team policies. This lets admins manage AI access through company identities and update permissions as responsibilities change.
Get in touch
Thanks. We’ll be in touch.
Expect an email from our team within one business day.
Meeting scheduler goes here (HubSpot Meetings, pending setup).