Business systems
- Analytics
- Advertising
- CRM
Controlled analytical access to the evidence.
Lytically Platform
The software your investigations run on.
The Platform connects purpose-built analytical capabilities across your business systems, keeps the context those results depend on, and exposes both to compatible AI tools through MCP. Customers who operate it themselves and customers on Managed Lytically use this same software. Routines, when you add them, use it for investigations that run on a schedule.
Connections
Controlled access to the systems you permit
Analytical tools
Operations framed as business questions
Business context
Definitions, rules, and history you can correct
MCP
How ChatGPT, Claude, and other agents reach it
Architecture
Systems and knowledge connect once. Several tools can use them.
The Platform keeps analytical access and maintained context together. ChatGPT, Claude, Routines, and other compatible AI reach that layer through MCP, the interface described below.
Business knowledge
- Goals
- Definitions
- Rules
- Business history
Maintained context, with source and recency visible.
Lytically Platform
Shared evidence and definitions
- Connected evidence
- Shared definitions
- Purpose-built analysis
MCP is the standard that lets AI tools access Lytically.
ChatGPT
Claude- Routines
- Other compatible AI
Separate connections to each app
Each application answers from its own data, with its own definitions.
Through the Platform
One investigation can call several analytical capabilities and the shared context that applies to them.
Integrations
What each connected system can answer.
A connection exposes a short list of analytical capabilities, such as traffic sources, search terms, or deal performance. These are the integrations available now. A system that is not listed here is not a live connection. Read what each connection can investigate.
Google Analytics
Website Analytics
Traffic, engagement, and conversion
- Traffic sources
- Page performance
- Events and key events
Google Search Console
Search & SEO
Visibility, clicks, and landing pages
- Search performance
- Opportunities
- Queries and pages
Google Ads
Advertising
Spend, delivery, and query quality
- Campaign performance
- Search terms
- Efficiency
HubSpot
CRM & Sales
Contacts, pipeline, and deals
- Conversion funnel
- Pipeline
- Deal performance
Acquisition evidence
Website behavior
Pipeline and deals
Analytical capabilities
Tools shaped around business questions.
A connection does not hand the assistant a raw API response and leave it to invent the analysis. Each integration exposes operations such as traffic sources, search terms, and deal performance. Flexible queries can cover evidence a named tool does not, but the named tools are the point.
| Capability | System | Question it helps investigate |
|---|---|---|
| Traffic sources | Google Analytics | Which channels changed, and did sessions move with them? |
| Page performance | Google Analytics | Which pages moved with the traffic? |
| Search performance | Google Search Console | Which queries gained or lost clicks? |
| Search opportunities | Google Search Console | Where is the site close to earning a click it does not have? |
| Campaign performance | Google Ads | Did spend and results move together? |
| Search terms | Google Ads | Which queries did the ads actually match? |
| Conversion funnel | HubSpot | Where did new opportunities slow down? |
| Deal performance | HubSpot | Did volume, value, or timing move reported revenue? |
Business context
Definitions and history the numbers do not contain.
An investigation can consult goals, terminology, operating rules, and earlier explanations alongside the measured evidence. Your team records and reviews that context. It keeps a source, a scope, and a status.
Supplied context can be incomplete, stale, or disputed. The Platform should help an agent show that, not treat a connected note as fact because it is available. Correcting and retiring context is part of keeping the layer useful.
Goals and priorities
What the company is trying to change this quarter.
Terminology and definitions
What counts as revenue, a qualified opportunity, or a conversion.
Rules and processes
How a lead is handed over, when an invoice is recognized, what a pause means.
Decisions and history
Why a metric moved the last time, and which explanation still applies.
Cross-system investigation
One question can use several systems and the shared context.
Connecting ChatGPT to HubSpot, and separately to Google Analytics, gives the assistant two independent views. Each view can be accurate and still leave the relationship, and the company’s own definition, outside the answer.
On the Platform, one investigation can call traffic sources, campaign performance, search performance, and deal performance, then apply the revenue definition your team maintains. The agent follows the evidence. Lytically supplies the capabilities and the relevant context. It does not host a separate Lytically chat analyst.
Illustrative output
Why did revenue fall last week?
01 Evidence
Acquisition and deal volume are broadly steady. Reported revenue is the material movement, $184k to $112k.
02 Context
The maintained definition says revenue is reported when the invoice is issued. A $68k renewal moved from Friday to Monday.
03 Reading
Timing accounts for most of the gap. Recount both weeks before changing spend. A smaller mix shift is still possible.
Figures are illustrative. The homepage walks through each source, the context record, and the limits of the reading. See the full investigation
MCP
Compatible AI tools use the Platform through MCP.
MCP is how ChatGPT, Claude, other compatible agents, and workflows you build call Lytically’s analytical tools and relevant business context. It is a core interface of the Platform. The primary experience is the assistant your team already uses, not a separate Lytically chat product.
MCP access and ordinary MCP use are included on every paid Platform plan. Your AI provider bills its own service separately. Abusive or runaway automated workloads can be limited. Connection limits apply to provider accounts, not to the number of AI clients using them.
- ChatGPT
- Claude
- Other MCP-compatible agents
- Customer-built workflows
Routines
The same capabilities, on a schedule.
Platform add-on · priced separately
A Routine starts from an instruction such as “every Monday, investigate last week and say what deserves attention.” Each run can choose which connected evidence and which current context matter. A scheduled report repeats a fixed export. Routines are part of the Platform architecture and are not included in Starter, Growth, or Business.
01
Instruction
A standing question, with a schedule and a boundary.
02
Same layer
The run uses the Platform’s analytical capabilities and relevant context.
03
Fresh finding
What changed, why it matters, and what deserves attention this time.
Access
Controlled analytical access.
Investigations run inside the permissions of the connected account and the capabilities Lytically exposes. This is the access model. It is not a certification claim.
Account isolation
A Lytically account reaches the provider accounts connected to it, not another customer’s.
Permitted resources
A connection discovers the properties, accounts, and objects you have allowed it to use.
Credentials stay with Lytically
The agent receives analytical tools and their results. Provider credentials are not passed through as tool output, and a connection is not unrestricted access to the account.
Implementation
Set the Platform up yourself, or have Lytically do it with you.
Self-managed customers connect systems, record context, and use the Platform from a compatible AI tool. That is a complete way to use the software.
Managed Lytically is the optional service for companies that want help implementing, configuring, and maintaining the same Platform. It is priced separately from Starter, Growth, and Business. Scope follows the engagement. The list below is not a package or a price.
Work the service can include
Scope follows the engagement. This is not a package, a price list or a promise that every item is included every time.
Implementation and setup
Stand up the Platform against the playbook.
Integration configuration
Connect the relevant live systems and the resources you permit.
Custom integration work
Where a standard connection is not enough, and where that work is in scope.
Business-context upkeep
Keep goals, definitions, rules and historical explanations current.
Semantic clarification
Record the meaning you intend, and the ambiguity that is still open.
Routine and investigation design
Shape recurring investigations and the questions they are allowed to pursue.
Ongoing analytical support
Help operate the layer as the business and its definitions change.
You still decide goals, definitions, and which sources are authoritative. Managed operation keeps the Platform in step with decisions you own. The two routes are also summarized in how companies use Lytically.
Book a free Blueprint Call.
Talk through the systems you use and the questions you want answered. A Blueprint engagement, if you choose one, is scoped separately.