Frequently Asked Questions

HubSpot Breeze Data Agent is a genuinely excellent in-app AI tool. It reads live CRM data, updates records, enriches contacts, and answers pipeline questions — all without leaving HubSpot. It's the right tool for real-time CRM operations. If you need to update a deal stage, send a sequence, or pull a quick contact summary, Breeze is exactly the right place to do it.

The structural ceiling appears when the question spans more than one object. HubSpot's own documentation states: "Reports cannot be created using multiple objects." The Breeze AI report builder is a frontend to the same single-object engine — it even labels results "Generated from SQL," which confirms the right idea. The constraint is what SQL it can generate. Questions like pipeline weighted by each rep's real historical close rate, or email click-rate computed across joined engagement and contact tables, require a join before aggregation — something a single-object engine cannot express.

AI Context Bridge adds three things that sit outside Breeze's scope: joins before aggregation (per-deal margin, multi-hop deal → contact → company → source, correctly weighted ratios), a full SQL receipt with every answer so the number is auditable, and your choice of frontier model — full Claude or ChatGPT, not a cost-optimised default. This isn't a Breeze replacement; it's an analytics layer on top of HubSpot.

No — and this distinction matters. AI Context Bridge is a data bridge, not an AI application. There is no AI packaged into the product. No model is baked in, no prompts are hidden, no black-box reasoning is happening on our side.

You bring your own AI assistant. Think of AI Context Bridge as a specialised translator: it connects your chosen AI — Claude, ChatGPT, or any MCP-compatible model — to your HubSpot data through the Model Context Protocol standard. The AI does the reasoning; the bridge supplies the data in a form the AI can use efficiently.

This architecture has a practical consequence: as frontier models get dramatically better each year, your analytical capability improves automatically. You're not waiting for a vendor to upgrade an embedded model — you simply point to the new one.

AI Context Bridge operates through a five-step process. The key architectural choice — and the one that makes complex analytics possible — is that your AI queries a dedicated SQL database, not HubSpot's API.

1. Secure Connection: You authorise DataLabs.Store to access your HubSpot portal using OAuth — no passwords or API keys to manage.
2. Complete Data Replication: We create a dedicated SQL database exclusively for your organisation and replicate your entire HubSpot portal into it — contacts, companies, deals, tickets, custom objects, and all relationships between them.
3. MCP Bridge: You connect your AI assistant to our MCP server using a unique URL. The AI authenticates securely with your credentials.
4. Intelligent Querying: When you ask questions, your AI uses specialised tools to understand your database schema, build optimised SQL queries, and retrieve exactly the data needed — including cross-object joins.
5. Continuous Learning: The system saves frequently-used queries and business semantics for increasingly accurate future analysis.

For a detailed step-by-step walkthrough with visual explanations, see our comprehensive guide:

How AI Context Bridge Works

HubSpot's MCP is a genuinely useful tool — especially for developer workflows, record updates, helpdesk automation, and quick CRM lookups. It's free, reads live data, and requires no sync. If you're building a support agent that needs to update a ticket or pull a contact record in real time, HubSpot's MCP is the right choice.

The limitation is structural, not a product quality issue. HubSpot's MCP routes through the CRM Search API — every question becomes a series of individual API calls. For analytical work that spans thousands of records across multiple object types, this creates real constraints:

We offer multiple hosting tiers because the primary cost driver is engagement volume, not record count. A portal with 500 contacts and five years of email logging can outgrow SQL Express faster than a portal with 100,000 contacts and minimal engagement history. Here's how to choose:

SQL Express — Recommended for Most

Hosted on our dedicated virtual servers.

Performance: Superior and consistently fast. Each server hosts only a handful of clients, ensuring you never compete for hardware resources. Queries execute quickly, reports generate smoothly, and your AI assistant gets rapid responses.

Storage limit: 10 GB database size.

Best for: Businesses prioritising responsive performance. Works well for portals with moderate engagement history.

Understanding the 10 GB Limit

Many HubSpot portals will eventually reach this cap, but it's driven almost entirely by engagement volume, not contact or deal counts:

Primary consumers: Emails logged in HubSpot consume the vast majority of database space. Meetings and calls can also take significant space depending on usage patterns.

Negligible impact: Custom objects, associations, contacts, companies, and deals — even with hundreds of thousands of records.

Key factors: How intensively engagements are logged, and how mature your portal is. A 5-year-old portal with extensive email logging will hit the limit faster than a brand-new portal with 10× more contacts.

Azure SQL Database

Hosted on Microsoft Azure's cloud infrastructure.

Performance: Good, but variable. Azure optimises for hardware efficiency across many customers. Complex analytical queries tend to run slower due to shared infrastructure and resource contention.

Storage limit: 250 GB — accommodates portals with years of intensive engagement history.

Best for: Mature portals with extensive email logging or very high-touch sales processes where engagement volume is substantial.

Upgrade: Individual approach — contact us to discuss. Azure offers several price tiers with different performance characteristics.

Custom SQL Server Hosting

Hosted on your own SQL Server infrastructure.

Performance: The absolute best — your dedicated hardware, your specifications, zero resource sharing.

Storage limit: Whatever you provision — no DataLabs.Store limitations.

Best for: Organisations already licensing SQL Server, enterprises with strict data residency requirements, or businesses demanding maximum performance.

Consideration: SQL Server licensing adds significant cost (typically $14,000+ for Standard, more for Enterprise). If you're already paying for SQL Server licenses, this option leverages your existing investment.

Setup: Available on request — we handle database setup and synchronisation, you provide the SQL Server instance. Full support included.

No — and the difference is fundamental to why the analytics work. An API wrapper translates AI requests into HubSpot API calls, fetching data on demand. That sounds reasonable until you ask a question requiring a join across 10,000 deals, their stage history, engagement timeline, and owner records simultaneously.

Why API wrappers fall short for analytics

• Rate limits: Complex questions require dozens or hundreds of sequential API calls. Rate limits interrupt the reasoning loop before it completes.
• Latency: Each API call adds round-trip time. Analysing patterns across 10,000 deals means 10,000+ API calls — taking minutes or timing out entirely.
• No pre-aggregation joins: APIs return data in predefined per-object formats. Joins and aggregations require fetching everything and processing locally — defeating the purpose of a server-side query.
• No historical context: APIs show current state. How deals progressed through stages over time isn't available through standard CRM Search endpoints.
• Schema opacity: HubSpot's API format requires extensive parsing; AI models reason better over a structured SQL schema with explicit relationships.

The independent analytical infrastructure approach

1. Complete data replication: We maintain a full copy of your HubSpot portal in a dedicated SQL database, updated continuously. Your AI queries this database directly.

2. Optimised schema: Data is restructured for analytical queries and AI comprehension. Relationships are pre-mapped, indices are optimised, and the structure is designed for the questions you actually ask.

3. Historical preservation: We track changes over time, enabling analysis of trends, progression, and historical patterns unavailable through API access.

4. Unlimited query complexity: Your AI can join 15 tables, aggregate millions of records, and compute complex metrics in a single SQL statement.

5. MCP-native tools: Specialised tools help AI understand your specific business context, learn from previous queries, and generate increasingly sophisticated analyses.

Your data security and privacy are our highest priorities. The short answer on AI training: no, your data is not used to train AI models. Here's the full picture:

Complete Data Isolation

Your data exists in a dedicated SQL database that serves only your organisation. There's no shared infrastructure, no data pooling, no mixing with other customers — a private data vault, physically and logically isolated.

AI Training: Your Data Stays Yours

• You use your own AI assistant (Claude, ChatGPT, etc.). We don't provide the AI — we provide the bridge.
• When you use Claude or ChatGPT through their applications, you're protected by their privacy policies. Both Anthropic (Claude) and OpenAI (ChatGPT) have clear commitments: data accessed through their APIs and desktop applications is not used for training.
• We never see your queries or your AI's responses. The conversation happens directly between you and your AI assistant; we only provide the data access layer.
• Your database is read-only from our perspective. We replicate data from HubSpot into your database; we don't monitor, log, or access your AI interactions.

Authentication & Access Control

Access requires two factors:

1. MCP Server URL: Unique to your organisation, acts as the connection endpoint.

2. Secret Key: Your AI must authenticate with this key for every session. Without it, access is denied. Lost your key? We revoke it and issue a new one — previous keys become immediately invalid.

HubSpot Connection Security

We connect via OAuth 2.0 — the same secure protocol used by Google and Microsoft. We receive read-only access tokens, never your HubSpot password. You can revoke our access at any time from your HubSpot settings, and we'll automatically stop data synchronisation.

Encryption

All data in transit is encrypted using TLS 1.3. Database connections are encrypted. Authentication credentials are encrypted both in transit and at rest.

GDPR Compliance

We are fully GDPR compliant. You own your data. You can request complete data deletion at any time — we'll permanently remove your database within 24 hours. No backups, no archives, no remnants.