Question 1

Where does our data go?

Accepted Answer

Mesh sends each request only to the model providers that you approve. No other party receives the data.

We have agreements with all our model providers. These agreements prevent the providers from using your data to train their models. Mesh also does not use your data for training.

Zero data retention (ZDR) is on by default. Mesh does not store your prompts or your responses.

Each enterprise customer receives a dedicated instance in the region that they select, for example Singapore. Mesh can also operate in your cloud account or in your data center. Then, the requests do not go out of your network.

Question 2

How do we audit usage if you keep no data?

Accepted Answer

Mesh processes each request in memory. For each request, Mesh records only metadata: the key, the model, the provider, the tokens, the duration, and the cost.

To make repeated requests fast and free, Mesh can keep an identical response in a temporary in-memory cache. This cache is private to your account. Each entry expires after a maximum of 24 hours. You can turn off the cache for a key or for a request.

For your audit:
Each response has a request ID. Use it to find the record of that request.
Pull the usage logs and the audit logs through the API into your data warehouse or SIEM.
If your audit team must have the full text of prompts, record it in your own systems. Then, the full text stays only in your environment.

Question 3

How does PII redaction operate?

Accepted Answer

The gateway removes personal data before the data goes to a model. It uses pattern rules and local models. The types of data that it finds include, but are not limited to:
Names, email addresses, and phone numbers
Card numbers, national ID numbers, and IBANs
IP addresses and credentials

Mesh replaces each item with a token. You select the categories, and you can add your own sensitive values. Name detection is not only for English.

Question 4

Which certifications do you have?

Accepted Answer

Mesh has SOC 2, ISO 27001, and GDPR certifications.

The contract gives the data locations, the retention rules, and the subprocessors. All models go through the same controls: ZDR, redaction, and access rules. Thus, the level of security does not change when the provider changes.

Question 5

Can Mesh control the AI tools that we use now?

Accepted Answer

Mesh controls all AI traffic that goes through it. For most tools and SaaS platforms, change two settings: the base URL and the API key. If a platform does not let you change these settings, we can make a connector for it.

To stop AI use without approval:
Make Mesh the only approved path to AI providers.
Block the direct provider endpoints at your firewall.

Question 6

Can we control which models each team uses?

Accepted Answer

Yes. Each API key has an allow-list of models. A key can also have a budget for each model: a maximum cost, number of tokens, or number of requests.

Keys belong to teams in your organization. Owners and admins control the members, the teams, the keys, and the limits. An audit trail records each change.

Question 7

Can we set spend and token limits?

Accepted Answer

Yes. Mesh applies the limits. It does not only monitor. You can set these limits for each key, each member, each team, and the organization:
Spend cap
Tokens per minute
Requests per minute and per day

When more than one limit applies, the smallest limit wins. Mesh sends alerts by webhook or email before the usage reaches a limit.

Question 8

Can you control agents?

Accepted Answer

Yes, for all agents that send traffic through Mesh. Give each agent its own key. Then you can set the model access and the limits for that agent.

The logs show each model call from the agent, with its tokens and its cost. The logs also show each request that Mesh refused, for example because of a limit or a blocked model.

Question 9

Can Mesh operate in our environment?

Accepted Answer

Yes. Select one of these options:
Our hosted service
A dedicated instance in your region
A hybrid configuration
A full installation in your environment, on Docker, EC2, Kubernetes, or your servers

Question 10

Does Mesh connect to our tools?

Accepted Answer

Yes. Mesh is compatible with the OpenAI API and the Anthropic API. To connect an application, change its base URL and its API key.
Mesh has a built-in MCP server. Claude Code, Claude Desktop, Cursor, and your own agents can use it.
Claude Code can operate through Mesh, with your limits and logs.
Mesh has built-in RAG for your files.
You can send Mesh traffic through your own Apigee proxy.
The Mesh CLI operates in your terminal.

Question 11

Which models can we use?

Accepted Answer

You can use more than 1,000 models for text, image, audio, video, and embeddings. Usually, a new frontier model is available a few hours after its release. If you need a model that is not in the catalog, ask us. We can add custom models on request.

You can also connect your own AWS Bedrock, Google Vertex AI, OpenAI, Azure OpenAI, or Azure AI Foundry accounts (BYOK). Your cloud credits continue to apply. Open-weight models can operate on your infrastructure.

Auto-routing first classifies the task. Then it gives each model a score for quality, cost, and latency, and it selects the model with the best score. You can change these weights for each request.

Question 12

How fast and reliable is Mesh?

Accepted Answer

Mesh adds approximately 50 ms of latency at P90. The measured uptime is 99.9%.

When retry and fallback are on, Mesh first tries the same provider again. Then, it sends the request to the same model on a different provider. If necessary, it uses a comparable model. You can set a maximum of 6 retries. If a provider fails frequently, Mesh removes it from routing for a short time.

Each enterprise contract gives its own SLA.

Question 13

How do you calculate the price?

Accepted Answer

You pay only for the tokens that you use. There is no platform fee and no seat license for individual users. Only BYOK traffic has a fee: 5% of the provider cost, after the first 1,000,000 free tokens each month.

Mesh Code and the Mesh CLI are also free to use for individual users.

For enterprises, contact us: https://cal.com/raushan-sharma/30min

Question 14

How much do we save?

Accepted Answer

Major models, such as Claude, usually cost 10% to 15% less than the direct list price. Some models cost up to 40% less.

Routing gives more savings, because Mesh sends simple tasks to less expensive models. The total savings can be up to 50%. The response cache and batch pricing decrease the cost more.

Example: you spend $10,000 each month on Claude directly. On Mesh, the same usage costs approximately $8,500 to $9,000, before routing savings.

Question 15

How do we pay? Is there a minimum commitment?

Accepted Answer

There is no minimum commitment and no subscription. You pay into a prepaid balance, and Mesh deducts the cost of each request. Credits do not expire. You can turn on auto-recharge. Enterprise contracts can also use monthly payment after usage (postpaid).

Optional: in the Philippines, you can buy through our partner Nexus Technologies:
Send a purchase order to Nexus.
Nexus sends you an invoice.
The credits go directly into your Mesh account.

Question 16

What does white-label cost?

Accepted Answer

The cost depends on your requirements. Book a call with us: https://cal.com/raushan-sharma/30min

Question 17

Why use Mesh, and not a direct provider or a free gateway?

Accepted Answer

Mesh gives one contract, one key, and one invoice for more than 1,000 models. It also decreases your costs.
Direct providers: Mesh has lower prices, higher rate limits, failover, and central governance.
Claude Code: Mesh has no seat limits, and you pay only for usage. You can continue to use Claude Code with a Mesh key.
LiteLLM: Mesh is fully managed, so you do not maintain servers. Mesh also includes discounted models, guardrails, and support.
OpenRouter: Mesh has lower prices, because OpenRouter adds a fee of approximately 5%. Mesh also has enterprise governance.
Kong: Mesh includes the models. With Kong, you must supply them.

Question 18

Who uses Mesh?

Accepted Answer

Mesh is the infrastructure of AI Fiesta. AI Fiesta has more than 1 million users in more than 100 countries.

The telecom operators Globe (Philippines) and StarHub (Singapore) use AI Fiesta. This month, Globe does a PoC of the Mesh API. Now, several banks and large companies in the Philippines evaluate Mesh.

Fiesta Labs Inc. makes Mesh. It is a US company, and Y Combinator alumni started it. We give references after you sign an NDA.

Question 19

Which use cases are usual for companies like ours?

Accepted Answer

Banks:
AI coding with spend control
A RAG copilot for internal documents
A compliance assistant for the CISO
Agents for customer support and loans
Logistics companies:
Document processing for shipping documents, invoices, and customs forms
Support agents (text and voice) that give shipment status to customers
Assistants for operations, warehouse, and training teams
AI in RPA workflows
Enterprises:
Document intelligence
Questions and answers on your data
An internal copilot with your brand
Telecom operators:
White-label consumer AI bundles
Model-as-a-service for B2B customers

These are only examples. Mesh powers thousands of use cases across all industries. If your use case needs an AI model, Mesh can support it.

Question 20

How do we start a pilot?

Accepted Answer

Book a call: https://cal.com/raushan-sharma/30min

Asia's largestsovereign AI gateway.

Request trace

Spend by team

Provider health

Live log

Allowed and excluded models

Spend caps

Role-based access

Rate limits

Allowed IPs

PII redaction

Zero data retention

Encrypted in transit

Works with your stack

Certified

Open-Weight

Frontier models

Open-Weight

Frontier models

Runs wherever you do. Works with everything you use.

Mesh Cloud

Dedicated region

Self-hosted

The mostcost-efficient AI gateway.

Auto routing

Direct model discounts

Most efficient caching

Every call passes one checkpoint.

Agent memory

RAG

Sandbox

Agent skills

Tool checks

Realtime speech

Asia's largestsovereign AI gateway.

The most
cost-efficient AI gateway.