Agentic Data Access

Agentic Search & Data

We turn your internal data into safe, agent-ready tools.

We make your databases, warehouses, and documents answerable by AI — governed by semantic contracts and permissions, with every answer cited or refused. Our RecoSearch system is the reference implementation, working inside the AI clients your team already uses.

Architecture

The governed runtime, layer by layer

Eight layers turn a natural-language business question into a proof-carrying, cited answer — or a principled refusal. Click any module on the left to see why it exists, which companies set the precedent, what was rejected, and how it wires to the rest.

question in→ access → runtime brain → meaning → federation → execution → sources → evidence →proof-carrying answer (or refusal) out
control planesemantic modelfederationexecutionsourcesevidence & trustcross-cutting
1 · Access tier — who is asking, and may they (MCP + REST, one core)
2 · Governed runtime brain — decide the posture before any data is touched
3 · Semantic operating model — what the data MEANS · 3 stacked layers
4 · Federation & conflict — combine evidence only through declared relations
5 · Execution — compile the plan, run governed SQL + document retrieval
6 · Source estate — where the data lives (degrade source-by-source)
DuckDB · zero-infraPostgresSnowflakeOpenSearch / ElasticsearchQdrantFreshdeskERPNextS3 · origin-only
7 · Evidence & trust — prove it, certify the claim, or refuse
8 · Cross-cutting spine — wraps every layer
Access tier

MCP gateway

MCP gateway / agent-client interface

The northbound interface where RecoSearch presents itself to an AI assistant. It speaks the Model Context Protocol (MCP) — the open standard AI clients like Claude and ChatGPT already use — and exposes a small set of governed, named tools (e.g. ask-a-business-question, list-metrics, run-verified-query) rather than raw database access. Every tool call is authenticated to a caller identity, permission-checked, and returns a proof-carrying, cited answer or a structured refusal.

Why it is in the architecture

RecoSearch's whole value proposition — proof-carrying, cited, governed answers — only matters if an AI assistant can actually reach it without bypassing it. The Access tier is the chokepoint: if the assistant can also reach the database directly, governance is theater. MCP is the right transport because it is the standard the target clients already speak ("build once, integrate everywhere"), so RecoSearch needs zero bespoke client work to be usable from Claude, Cursor, VS Code, or ChatGPT. Exposing governed *tools* (verbs) instead of a SQL endpoint keeps interpretation, permissioning, and citation on RecoSearch's side of the boundary, which is exactly where the trust guarantees live.

Company precedents

Snowflake (Cortex / Horizon Context)Exposes governed Semantic Views over MCP so any external AI agent — explicitly Claude, Cursor, or any MCP-compatible framework — can consume them 'with governance enforced at the source.' MCP is positioned as one activation surface alongside Cortex Analyst, treating the protocol as the agent-facing front door to governed definitions.

Snowflake (Cortex Analyst)Ships an API-first, 'convenient REST API' that turns natural-language questions into governed answers without the caller writing SQL, and fully integrates Snowflake RBAC so generated/executed SQL adheres to established access controls — i.e. the agent-facing endpoint enforces governance at the source rather than handing out a DB connection. Cortex Agents wraps this as the tool-calling orchestration layer.

Microsoft (Fabric data agent)Provides a conversational agent over governed semantic models where data access runs under the caller's Microsoft Entra ID identity and workspace/data permissions — the agent reads schemas and runs SQL/DAX/KQL only if the user has access, honors Purview policies and sensitivity labels, and 'simply retrieves and processes structured data' (no exfiltration of raw DB credentials to the model).

Alternatives rejected

Bespoke REST/GraphQL API as the only agent interfaceA custom REST surface forces every AI client to be taught RecoSearch's endpoints, auth, and schemas individually — losing MCP's 'build once, integrate everywhere' leverage across Claude, ChatGPT, Cursor, and VS Code (scraped/docs/modelcontextprotocol.io-docs-getting-started-intro.md). Vendors that lead with REST (Cortex Analyst) still bolt MCP on top precisely to reach 'any MCP-compatible framework' (scraped/open-interoperable-agent-ready-122d8a2d67c2.md). RecoSearch keeps an internal REST/contract layer but leads with MCP as the agent-facing transport.

Direct database access (hand the agent a connection string / full DB credentials)This is the failure mode the sources explicitly indict: 'hardcoded credentials, full database access, zero audit trail, and prayer' (scraped/universal-agent-connector-mcp-ontology-production-ready-ai-infrastructure-0b4e35f22942.md). It moves interpretation, permissioning, and citation to the model's side of the boundary, which destroys RecoSearch's proof-carrying/governed guarantees. The Universal Agent Connector and Fabric data agent both replace this with identity-bound, permission-checked, audited tool access (scraped/github/cloudbadal007-universal-agent-connector.md; scraped/docs/learn.microsoft.com-en-us-fabric-data-science-how-to-create-data-agent.md).

Text-to-SQL exposed as the primary tool (let the agent generate arbitrary SQL the gateway runs)The 'semantic-layer-is-dead' thesis warns that raw SQL generation makes the agent a 'translator' guessing at meaning, when what is needed is interpretation: agents must call meanings expressed as functions/protocols, not tables (scraped/the-semantic-layer-is-dead-now-its-an-api-for-ai-agents-f91d48a0c74a.md). RecoSearch exposes governed question/metric tools so interpretation stays server-side; any internal SQL is constrained by the semantic/metric layer, not free-form from the model.

Depends on

Identity & authentication (caller identity binding)Authorization / policy engine (permission + access-control checks)Semantic / metric layer (the governed verbs and definitions the tools expose)Audit & observability log (records every tool call)

Feeds

AI assistant / MCP client (Claude, ChatGPT, Cursor, VS Code)Refusal & error contract (structured denials returned to the caller)Proof / citation envelope (cited, evidence-bearing answers returned over the protocol)

Encapsulates

MCP server runtime / transportTool registry & schema definitions (named governed verbs)Request authentication & API-key/identity validation at the edgePer-call permission gate / SHACL-style delegation guardRate limiting & per-agent quotasTool-call audit emission

Evidence sources (6)

open-interoperable-agent-ready-122d8a2d67c2docs.snowflake.com-en-user-guide-snowflake-cortex-cortex-analystlearn.microsoft.com-en-us-fabric-data-science-how-to-create-data-agentmodelcontextprotocol.io-docs-getting-started-introcloudbadal007-universal-agent-connectorcloudbadal007-agentic-mesh-security

Reference system: RecoSearch

RecoSearch is our agentic data gateway, built on this architecture — the moat is semantic governance, not wrapping a database with MCP.

Case studies

Representative scenarios we've delivered with RecoSearch.

Finance

Finance close acceleration

Challenge
Month- and quarter-end close spans ledgers, budgets, and approvals — controllers need fast variance answers without losing the audit trail.
What we did
Semantic contracts across ledger, budgets, approvals, and policy route close questions through approved metrics and return answers with lineage and citations.
Outcome
Immediate, auditable variance analysis; material exceptions surface for review and unsafe data exports are refused automatically.
Semantic contractsPolicy-aware refusalAudit evidence
SaaS · Customer Success

SaaS renewal-risk intelligence

Challenge
CSMs juggle renewal risk across tickets, usage trends, and invoices with no single governed view that respects data sensitivity.
What we did
A semantic contract bridges CRM accounts, subscription metrics, usage events, and support cases; permission boundaries redact PII and enforce aggregate-only access.
Outcome
Ranked renewal-risk tables and usage-decline diagnostics grounded across sources, with customer data protected through the cross-source query.
Multi-source routingHealth scoringPII redaction
Finance · RevOps

Revenue variance root-cause

Challenge
Revenue misses are hard to decompose, and dashboards or ad-hoc SQL often contradict each other — eroding trust in the numbers.
What we did
A variance-waterfall contract decomposes actual vs. forecast across volume, price, churn, discounting, and mix, each component traced to canonical sources.
Outcome
One reconciliation source of truth with clear driver attribution and embedded evidence, so reviews and approvals move quickly.
Variance decompositionReconciliationConfidence labels
Retail · Ecommerce

Ecommerce margin & returns

Challenge
Finance, merchandising, and marketing each define profit and returns differently, leading to inconsistent SKU-level decisions.
What we did
Semantic contracts for the gross-to-net bridge and return-driver ranking route margin questions through unified SKU, order, return, and inventory sources.
Outcome
Unified margin reporting with transparent definitions, return root causes by category, and early inventory-risk warnings.
Margin decompositionPolicy citationInventory signals

Make this part of your stack agentic.

Tell us the operation or dataset. We'll scope a governed, production-shaped build.

Start a conversation →