# Mindcat

> Salesforce implementation and AI advisory consultancy serving mid-market and enterprise clients in the UAE, India, and USA.

Mindcat implements Salesforce and assembles enterprise AI stacks: models, agent runtimes, memory, identity, and governance. Work spans new implementations, failed-project rescue, partner-selection guidance, and AI governance aligned to the EU AI Act, NIST AI RMF, ISO 42001, and DIFC Regulation 10.

Founder: [Shivanath Devinarayanan](https://mindcat.ai/about) — Salesforce MVP Hall of Fame. Writing: [shivanathd.com](https://shivanathd.com/).

Catalog: [/ai-tools](https://mindcat.ai/ai-tools). Machine-readable details: [/auth.md](https://mindcat.ai/auth.md) for API access, [/openapi.json](https://mindcat.ai/openapi.json) for the contact API spec, [/.well-known/mcp/server-card.json](https://mindcat.ai/.well-known/mcp/server-card.json) for MCP discovery. Request a path as Markdown via `/api/markdown?path=...` (returns a path-aware summary). This index is also at [/.well-known/llms.txt](https://mindcat.ai/.well-known/llms.txt).

Content-Signal preference: search and citation allowed; training on this site is declined (`ai-train=no`, `search=yes`, `ai-input=yes`).

## Brand and logo

The **Mindcat cat logo** is a periwinkle cat silhouette with no wordmark. Use the SVG. Do not use the PNG lockup as the cat — that file is a historic tulip with the name.

- Cat logo (canonical): [mindcat-mark.svg](https://mindcat.ai/media-kit/mindcat-mark.svg) — SVG, 3200 × 2400. Same graphic: [/favicon.svg](https://mindcat.ai/favicon.svg), [/logo.svg](https://mindcat.ai/logo.svg), [/icon.svg](https://mindcat.ai/icon.svg), [/.well-known/logo.svg](https://mindcat.ai/.well-known/logo.svg).
- Color lockup (not the cat): [mindcat-logo.png](https://mindcat.ai/media-kit/mindcat-logo.png) — historic tulip + Mindful Catalyst. Public name to write is Mindcat Consulting.
- [Media kit](https://mindcat.ai/media-kit): Logos, bios, partner badge, founder portrait. Markdown: [media-kit.md](https://mindcat.ai/media-kit.md). Zip: https://mindcat.ai/media-kit/mindcat-media-kit.zip

## Services

- [Salesforce Implementation](https://mindcat.ai/services/salesforce-implementation): New orgs, configured around your process.
- [Salesforce Integration](https://mindcat.ai/services/salesforce-integration): Connecting Salesforce to existing systems.
- [Salesforce Optimization](https://mindcat.ai/services/salesforce-optimization): Improving underperforming Salesforce orgs.
- [Enterprise AI Deployment](https://mindcat.ai/services/enterprise-ai-deployment): Agentforce, a custom stack, or both — pilot to production.
- [We set up the agent](https://mindcat.ai/services/ai-agents): An AI agent on a named process. UAE desk or US desk.
- [AI agents in the UAE](https://mindcat.ai/ai-agents/uae): Dubai HQ. WhatsApp, free-zone ops, book transfer. PDPL and DIFC.
- [AI agent implementation in the USA](https://mindcat.ai/ai-agents/usa): US teams. Slack-first. Texas mailing, not a second HQ.
- [Data Readiness for Agents](https://mindcat.ai/services/ai-readiness): Knowledge, permissions, and grounding before you build.
- [Data Foundations](https://mindcat.ai/services/data-foundations): Data Cloud as context. Agentforce as agency. Graph and vectors as memory, fronted by MuleSoft.
- [AI Governance](https://mindcat.ai/services/ai-governance): EU AI Act, NIST AI RMF, ISO 42001, DIFC Regulation 10. Gateway policy families, evals, and every connector named.
- [AI Observability](https://mindcat.ai/services/ai-observability): Traces from gateway to broker to agent to tool. Identity and cost on the same trail.
- [Failed Project Rescue](https://mindcat.ai/services/failed-project-rescue): Stuck Salesforce orgs and stalled AI pilots.

Decision: [Agentforce vs custom agents](https://mindcat.ai/compare/agentforce-vs-custom-agents).

## Architecture

Your agents fail in a named plane. Skip one and you get a demo with write access, not a shortcut.

- [Digital labor architecture](https://mindcat.ai/architecture): Your agents fail in a named plane, not in general. We name the missing plane and start with one slice.
- [Experience plane](https://mindcat.ai/architecture/experience): If the agent is not in Slack, the portal, or the voice channel your people already open, they will not use it.
- [Control plane](https://mindcat.ai/architecture/control): Every tool call and model call needs one front door. Otherwise you cannot say who acted, or stop it.
- [Runtime plane](https://mindcat.ai/architecture/runtime): Classify the step first. Agentforce, a code agent, or an SDK is a consequence, not a shopping list.
- [Memory plane](https://mindcat.ai/architecture/memory): A vector store is not memory. Your agent needs a profile, working context, and what was decided.
- [Durable plane](https://mindcat.ai/architecture/durable): The write is the risk. Queues, retries, and document intake so a timeout does not post twice.
- [Salesforce and MuleSoft by plane](https://mindcat.ai/architecture/salesforce-mulesoft): You already pay for several of these. We write down the job each cloud does once agents are in the picture.
- [Spec-driven engineering](https://mindcat.ai/guides/spec-driven-engineering): Your fleet of agents cannot be hand-written. Specs, shadow mode, then a human gate on the write.

## MuleSoft

MuleSoft is a first-class practice, not a footnote under Agentforce.

- [MuleSoft](https://mindcat.ai/products/mulesoft): Anypoint, Flex Gateway, Agent Fabric, and IDP. A simpler pipe when reuse is not real.
- [Anypoint](https://mindcat.ai/products/mulesoft/anypoint): API-led design, Exchange, DataWeave, MQ. A new source is a mapping, not a new deploy.
- [IDP](https://mindcat.ai/products/mulesoft/idp): Classify, extract, validate, write downstream. A person reviews the exceptions.
- [Governed agent architecture](https://mindcat.ai/products/mulesoft/agent-architecture): One path from the desk through the gateway to systems of record.
- [Flex Gateway](https://mindcat.ai/products/mulesoft/flex-gateway): On-behalf-of identity, MCP hosting, LLM proxy, rate limits, audit.
- [MuleSoft delivery](https://mindcat.ai/guides/mulesoft-delivery): Foundation, scale, then data. Named owners. Receipts, not slides.

## Salesforce platform products

- [Data Cloud](https://mindcat.ai/products/data-cloud): The profile your agent grounds on. Data spaces and vector search as context, not agency. Also sold as Data 360.
- [Slack](https://mindcat.ai/products/slack): The desk your people already live in. Fast commands, visible approvals, write-back to the org.
- [Salesforce Platform](https://mindcat.ai/products/salesforce-platform): The event fabric in your org. Flow, Apex, platform events, and change data capture.

## Agentforce practice

- [Agentforce](https://mindcat.ai/products/agentforce): Salesforce-native agents when the work and the permissions already live in the org.
- [Agentforce Network](https://mindcat.ai/products/agentforce/network): Brokers, A2A, and a registry. Shared space when it is safe.
- [Agent Broker](https://mindcat.ai/products/agentforce/broker): Guided determinism. Not every hop is a model call.
- [Trusted identity](https://mindcat.ai/products/agentforce/identity): On-behalf-of. The agent acts as the user.
- [Agent program operating model](https://mindcat.ai/guides/agent-program-operating-model): A claim, a receipt, and an owner.

## Dreamforce 2026

- [Dreamforce 2026](https://mindcat.ai/dreamforce-2026): Founder Shivanath Devinarayanan teaches two verified breakouts at Moscone, September 15-17.
- [Agent Fabric: See, Control, and Govern Every Agent](https://mindcat.ai/dreamforce-2026): Wednesday 16 September 2026, 10:00-10:40 AM PT, Moscone South, Content Pavilion, Stage 5.
- [Recruiting During the Talent Paradox](https://mindcat.ai/dreamforce-2026): Thursday 17 September 2026, 8:30-9:10 AM PT, Metreon AMC, Level 3, Theatre 11. Co-speaker Chris Davis.
- Markdown: [/api/markdown?path=/dreamforce-2026](https://mindcat.ai/api/markdown?path=/dreamforce-2026).

## AI Tools We Implement

### Models
- [Claude (Anthropic)](https://mindcat.ai/products/claude)
- [Google Gemini](https://mindcat.ai/products/gemini)
- [OpenAI](https://mindcat.ai/products/openai) (API and Azure OpenAI)
- [AWS Bedrock](https://mindcat.ai/products/aws-bedrock)

### Agentic systems
- [Agentforce](https://mindcat.ai/products/agentforce)
- [Claude Agent SDK](https://mindcat.ai/products/claude-agent-sdk): Multi-turn tool use. Schema and business-rule checks before your agent writes.
- [OpenAI Agents SDK](https://mindcat.ai/products/openai-agents-sdk): Structured outputs and typed routing. Rejected output becomes an exception with an owner.
- [LangGraph](https://mindcat.ai/products/langgraph): Loops with hop limits. A human interrupt before a high-impact write in your process.
- [Google ADK](https://mindcat.ai/products/google-adk)
- [A2A Protocol](https://mindcat.ai/products/a2a)
- [CrewAI](https://mindcat.ai/products/crewai)
- [MuleSoft Agent Fabric](https://mindcat.ai/products/agent-fabric)
- [Model Context Protocol (MCP)](https://mindcat.ai/products/mcp)

### Agentic memory
- [Neo4j](https://mindcat.ai/products/neo4j)
- [Cognee](https://mindcat.ai/products/cognee)
- [IndyKite](https://mindcat.ai/products/indykite)

### Automation and voice
- [n8n](https://mindcat.ai/products/n8n)
- [Retell AI](https://mindcat.ai/products/retell-ai)

## Featured use cases

- [First process](https://mindcat.ai/ai-use-cases/first-process-agent): One documented job. Then we stay.
- [Book of business](https://mindcat.ai/ai-use-cases/book-of-business-transfer): Pull the policy book from the last system.
- [WhatsApp or Slack seat](https://mindcat.ai/ai-use-cases/whatsapp-or-slack-agent): WhatsApp in the Gulf. Slack for the company.
- [Customer service](https://mindcat.ai/ai-use-cases/customer-service-automation-coo): Agentforce + Retell + OpenAI + Cognee
- [Sales pipeline](https://mindcat.ai/ai-use-cases/sales-pipeline-automation-ceo): Agentforce + OpenAI/Claude + n8n
- [Legal / contracts](https://mindcat.ai/ai-use-cases/legal-document-automation-clo): Claude + Neo4j + Cognee
- [IT helpdesk](https://mindcat.ai/ai-use-cases/it-helpdesk-automation-cio): Agentforce + Gemini/OpenAI + ADK + MCP
- [Compliance](https://mindcat.ai/ai-use-cases/compliance-monitoring-cro): Claude + IndyKite + AI Governance
- [Golden Visa screening](https://mindcat.ai/ai-use-cases/golden-visa-eligibility-checker): Claude + Cognee + n8n
- Full list: [AI use cases](https://mindcat.ai/ai-use-cases)

## Resources

- [Resource hub](https://mindcat.ai/resources): Salesforce and AI partner-selection, evaluation, deployment, and recovery guides.
- [Decision notes](https://mindcat.ai/resources/decision-notes): Six short posts — Agentforce vs custom, MCP, red flags, three agents in v1, context first, stalled-pilot rescue.
- [40 questions to ask an AI implementation partner](https://mindcat.ai/resources/questions-ask-ai-implementation-partner)
- [How to Choose an Enterprise AI Implementation Partner](https://mindcat.ai/resources/choosing-ai-implementation-partner)
- [From AI Pilot to Production](https://mindcat.ai/resources/ai-pilot-to-production-guide)
- [Claude Enterprise Deployment Guide](https://mindcat.ai/resources/claude-enterprise-deployment-guide)
- [MCP Governance Guide](https://mindcat.ai/resources/mcp-model-context-protocol-governance)
- [Enterprise AI Governance Playbook](https://mindcat.ai/resources/enterprise-ai-governance-playbook)
- [How to Choose the Right Salesforce Partner](https://mindcat.ai/resources/choosing-salesforce-partner)

## Guides

- [Salesforce technical guides](https://mindcat.ai/guides)
- [Implementation best practices](https://mindcat.ai/guides/salesforce-implementation-best-practices)
- [Data migration strategy](https://mindcat.ai/guides/salesforce-data-migration-strategy)
- [Integration patterns](https://mindcat.ai/guides/salesforce-integration-patterns)
- [MuleSoft delivery](https://mindcat.ai/guides/mulesoft-delivery)
- [Agent program operating model](https://mindcat.ai/guides/agent-program-operating-model)
- [Spec-driven engineering](https://mindcat.ai/guides/spec-driven-engineering)
- [Security and compliance](https://mindcat.ai/guides/salesforce-security-compliance)
- [Performance optimization](https://mindcat.ai/guides/salesforce-performance-optimization)

## Offices and field notes

- [Locations](https://mindcat.ai/locations): Dubai HQ, Kochi studio, US mailing desk.
- [Field notes](https://mindcat.ai/case-studies): Anonymized Salesforce stories. Clients asked not to be named.

## For executives

- [Salesforce for CEOs](https://mindcat.ai/for-ceos/salesforce-transformation)
- [Salesforce for CFOs](https://mindcat.ai/for-cfos/salesforce-roi-optimization)
- [Salesforce for CTOs](https://mindcat.ai/for-ctos/system-integration)
- [Marketing automation for CMOs](https://mindcat.ai/for-cmos/marketing-automation)

## Optional

- [Industries served](https://mindcat.ai/industries)
- [About](https://mindcat.ai/about): The firm and the founder.
- [Contact](https://mindcat.ai/contact): Form submission, no authentication required — see /auth.md.