You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Croohm is currently a mobile-first CRM and shared communication platform for SMBs, with foundations in leads, workspaces, teams, WhatsApp, Facebook Lead Ads, webhooks, campaigns, and controlled platform access.
The future product should not be positioned as "another CRM". The stronger position is:
Croohm is an AI business coworker for SMBs: it finds prospects, enriches them, follows up, handles customer conversations, runs playbooks, and keeps the team in control from a mobile-first inbox.
The immediate product path should be disciplined: complete the manual platform, add prospect enrichment as a source feeding the lead lifecycle, then evolve the backend into clearer service boundaries gradually.
Croohm is a mobile-first, multi-channel CRM and business communication platform for SMBs. The working product is centered on:
Lead/contact management
WhatsApp conversations
Facebook Lead Ads ingestion
Website/webhook lead ingestion
Campaigns and broadcast flows
Team/workspace permissions
Library assets and WhatsApp templates
Mobile-first lead follow-up
The platform is best approached in a practical sequence: complete the manual operating system first, then add intelligence and automation feature by feature.
Existing Product Surfaces
Mobile product:
Lead list and lead details
Chatbox for WhatsApp conversations
Team management and permissions
Campaigns and broadcasts
Library for files/templates
Connectors for WhatsApp, Facebook, website webhooks, and controlled platform access
Croohm is an AI business coworker that runs communication-heavy business workflows from a mobile-first lead inbox.
This is stronger than "CRM", "chatbot", or "automation tool". Those categories are too narrow.
Product Pillars
1. Lead Hub
The lead hub is the system of record for every prospect/customer.
It should combine:
Lead profile
Company/contact enrichment
Conversation history
Status/stage
Owner/assignee
Notes/interactions
Campaign/source
Next action
AI confidence and recommendations
2. Communication Hub
Croohm should remain conversation-first. Channels should converge into the same operational inbox:
WhatsApp
Facebook/Instagram
Website/webhook
Email
SMS
Future voice/call logs
The user should not care which connector created the lead. They should care: who is this, what do they want, what should happen next?
3. Dispatch
Dispatch is Croohm's prospecting and lead discovery engine.
It should find prospects from external sources, enrich them, score them, and load them into the lead hub with draft outreach and a recommended next action.
This is the "Vibe Prospecting-like" capability, but India/SMB-aware and deeply integrated into Croohm.
4. Conversation Automation
This layer handles inbound intent and repetitive follow-up:
Detect intent
Qualify leads
Ask missing questions
Route to the right teammate
Suggest or send replies
Start approved playbooks
5. Support Agent
The support agent uses the Library, past conversations, FAQs, and business docs to answer repeat customer questions. It should begin as a suggestion/handoff system, then graduate into supervised auto-replies.
6. Playbook Engine
The Playbook Engine turns plain-language process docs into executable workflows:
Send message
Wait
Check reply
Update status
Assign owner
Notify team
Escalate
Require approval
This is the bridge between "AI assistant" and reliable business operations.
7. AI Assistant
The assistant is the natural-language control plane:
"Find leads not contacted in 10 days."
"Draft follow-ups for hot leads."
"Import new MCA leads for Kochi and assign to sales."
"Run onboarding for all leads tagged paid."
The assistant should plan, preview, ask for approval where needed, execute, and leave an audit trail.
Differentiation
Croohm should not try to beat enterprise CRMs feature-by-feature.
The winning wedge is:
Mobile-first
WhatsApp-first for India/SMB markets
Lead/conversation unified
AI as an execution coworker, not a decoration
Prospecting/enrichment connected directly to follow-up
Simple enough for a small team owner to use daily
Strategic Product Narrative
Use this mental model:
Old CRM:
Record what happened.
Modern sales tools:
Help you find and message people.
Croohm:
Finds the right people, enriches the context, starts the right conversation, follows up, and tells the team what needs attention.
Build Principle
Every AI feature should attach to an existing manual workflow first.
Do not build isolated AI demos. Build:
AI on top of leads
AI on top of conversations
AI on top of campaigns
AI on top of library/playbooks
AI on top of assignments and reminders
That keeps the product reliable and gives users a fallback when AI is wrong.
Croohm can become much more valuable if it does not only manage incoming leads, but also helps clients find new prospects.
For clients selling GST, workspace, accounting, Google Workspace, website, or operational services, freshly registered companies in MCA are high-intent prospects. These companies often need setup help immediately.
This fits Croohm's future "Dispatch" capability:
The user describes who they want. Croohm finds prospects, enriches them, scores them, drafts outreach, and puts them into the inbox.
MVP Use Case: MCA New Company Leads
Input:
Vendor daily CSV/API of newly incorporated companies/LLPs
Optional filters: state, city, company type, capital range, industry/activity
Output into Croohm:
Lead record
Company profile
Contact candidates
Source and confidence
Compliance status
Suggested offer
Draft outreach
Assignment/campaign
Recommended Data Fields
Company
Company name
CIN/LLPIN
Incorporation date
Company type: Private Limited, LLP, OPC, etc.
Registered address
City/state/pincode
Industry/activity
Authorized capital/contribution
Website/domain if discovered
GST status if available
MCA/vendor source
Data freshness timestamp
People
Director/partner names
Designation/role
Email
Phone
LinkedIn/profile URL if discovered
Contact source
Contact confidence
DND/NDNC status where applicable
Enrichment
Company website
Social profiles
Google Business profile where relevant
Tech stack where relevant
Hiring/funding/news signals
Similar company category
Business maturity estimate
Residential-address likelihood
Croohm-Specific Fields
prospectScore
fitScore
urgencyScore
sourceConfidence
recommendedOffer
recommendedChannel
nextBestAction
enrichmentStatus
complianceStatus
Pipeline
Vendor MCA feed
-> raw import storage
-> schema validation
-> dedupe by CIN/LLPIN, phone, email, domain
-> normalize company/contact fields
-> enrich company
-> enrich people
-> validate email/phone
-> DND/NDNC and suppression checks
-> score against client ICP
-> create/update Croohm lead
-> assign campaign/team/owner
-> generate outreach draft
-> notify inbox
Product Surface
Do not build this as a separate "database export" screen first.
Croohm currently uses Firebase as a backend-as-a-service:
Firebase Auth for identity
Firestore for primary operational data
Firebase Functions for server-side actions and integrations
Firebase Storage for media/assets
FCM/Notifee for notifications
Mobile client reads key collections directly with Firestore listeners
This is appropriate for the current stage because it gives fast iteration, realtime UX, and simple deployment.
Current Service-Like Boundaries
Even though the backend is Firebase, several domain boundaries already exist inside Cloud Functions:
WhatsApp messaging
WhatsApp webhooks
WhatsApp templates
Meta OAuth
Facebook Lead Ads import/webhooks
Website lead webhook
Campaigns
Member management
Agent key management
croohmAgentTools
These should become the seeds of future services.
Migration Risks
Direct Firestore Coupling
The mobile app reads and writes Firestore directly for core domains. A hard switch to microservices would break:
Realtime listeners
Offline cache behavior
Optimistic message sending
Security rule assumptions
Query/index behavior
Authorization Split
Tenant/workspace authorization currently exists across:
Firestore rules
Cloud Function assertions
Client filters
Legacy role structures
Before serious service extraction, authorization must be centralized and made testable.
Legacy Data Shape
Some data/rules still accommodate older shapes, especially around workspaceId and messages. Backfills are needed before strict service boundaries.
Storage Rules
Storage access needs a stricter tenant/workspace membership check. Broad authenticated reads are risky for multi-tenant media.
Callable IAM Maintenance
The callable IAM postdeploy script is manually maintained. It should be kept in sync through CI checks against exported callables to prevent silent 401/403 failures.
Recommended Migration Strategy
Do not rewrite into microservices immediately.
Use a strangler pattern:
Current Firebase app
-> add API gateway for mutations and external clients
-> normalize events
-> backfill data
-> extract one domain at a time
-> keep Firestore as realtime read model until there is a strong reason to replace it
Phase 0: Stabilize Firebase
Before migration:
Complete current manual platform features.
Backfill workspaceId across messages/conversations/lead subrecords.
Remove legacy security-rule branches where possible.
Tighten Storage rules.
Add CI checks for callable exports vs IAM script.
Standardize collection names and casing.
Add event/audit records for critical mutations.
Phase 1: API Gateway
Create croohm-api as the first backend boundary.
Responsibilities:
Verify Firebase ID tokens.
Resolve tenant/workspace server-side.
Enforce role/permission checks.
Provide stable mutation APIs.
Rate-limit.
Own idempotency keys.
Write to Firestore initially.
Keep Firestore direct reads for realtime UX during this phase.
Phase 2: Event Normalization
Every external source should emit canonical events:
lead.created
lead.updated
conversation.message.received
conversation.message.sent
campaign.created
prospect.imported
prospect.enriched
playbook.started
agent.action.proposed
agent.action.executed
Use Firestore initially for event/audit records. Add Pub/Sub or Cloud Tasks where queueing is needed.
Phase 3: Extract Domains
Recommended extraction order:
Identity and tenant access
Connector token service
Messaging service
Lead service
Campaign service
Notification service
Prospecting/enrichment service
Agent/orchestrator service
Reasoning:
Identity/authorization must come first.
Connector tokens are security-sensitive and already server-only.
Messaging has external APIs, retries, rate limits, and webhooks.
Leads/campaigns depend on the earlier boundaries.
Agent/orchestration should come after stable domain APIs exist.
Phase 4: Agentic System
The agentic system should not write directly to Firestore.
It should call tools/APIs:
searchLeads
upsertLead
addNote
assignLead
sendMessageDraft
startPlaybook
importProspects
enrichProspect
Croohm already has croohmAgentTools; evolve it into a governed tool layer with permissions, rate limits, audit logs, and approval policies.
This document defines what Croohm is building, what it is not building, and how product and implementation decisions should stay aligned with the core direction.
One-Sentence Product Definition
Croohm is a mobile-first AI business coworker for SMB teams that finds leads, enriches context, manages conversations, follows up, and helps the team convert customers without dropping work.
The Core Job
Every Croohm feature should help answer at least one of these questions:
Who should I talk to?
Why should I talk to them now?
What happened last?
What should happen next?
Can Croohm safely do it for me?
If a feature does not answer one of these, it is probably not core.
Product Principles
1. Lead Inbox First
The lead list is not just a table. It is the daily operating surface.
The lead row should eventually show:
Person/company identity
Source
Last message or last activity
Owner/assignee
Urgency
Fit/quality score
Next best action
AI/playbook status
2. Manual Before Autonomous
Every autonomous workflow needs a manual version first.
This brief looks at where CRM, prospecting, enrichment, support automation, and agentic workflow platforms are moving in 2026, then translates that into Croohm's opportunity.
Market Direction
The market is moving from:
CRM as database
-> CRM as workflow system
-> CRM as AI-assisted workspace
-> CRM as agentic operating layer
The important shift is not "AI chat inside CRM". The shift is that CRM platforms are trying to become execution systems:
Research prospects
Detect buying signals
Draft outreach
Resolve support conversations
Execute playbooks
Hand off to humans
Price by outcomes
Govern AI actions
Evidence From Adjacent Platforms
HubSpot Breeze
HubSpot is moving toward embedded agents inside the Smart CRM. Its public material says Breeze Agents automate work across prospecting, customer service, content, and more. HubSpot describes prospecting capabilities such as researching target accounts, identifying buying signals, and crafting personalized outreach using CRM context and brand voice.
HubSpot also moved Breeze Customer Agent and Prospecting Agent toward outcome-based pricing in 2026: resolved conversations and leads recommended for outreach. This matters because it shows AI CRM monetization is shifting from seats/features to completed work.
Salesforce is pushing CRM toward agentic workflows and contact center automation. Salesforce describes Agentforce Contact Center as unifying voice, digital channels, CRM data, and AI agents in one system, with self-service, handoff, and visibility across interactions.
This validates Croohm's direction: conversations, CRM data, channels, and agents need to live together.
Intercom Fin shows the customer-support side of the same trend: AI agents are measured by resolution rate, not just message generation. Intercom reporting includes Fin AI Agent resolution rate and related operational metrics.
For Croohm, this means every agent should have outcome metrics:
Clay and Apollo represent the prospecting/enrichment side.
The market distinction is clear:
Clay-style: multi-source waterfall enrichment and flexible workflows.
Apollo-style: integrated database, enrichment, and sequencing in one sales platform.
Croohm should not copy either directly. Croohm's advantage can be an execution layer for SMBs: prospect data becomes a lead, with context, score, draft outreach, assignment, and follow-up inside the same mobile inbox.
Manychat validates conversational automation across Instagram, WhatsApp, Messenger, and Telegram. Its help docs highlight channel messaging windows, including the 24-hour window for Facebook Messenger, Instagram, and WhatsApp.
For Croohm, the lesson is: channel rules are product constraints. AI and automation must respect messaging windows, templates, and platform limits.
Attio is relevant because it shows the demand for flexible CRM data models, clean UI, enrichment, workflows, and AI features that are integrated into the CRM rather than bolted on.
Croohm should borrow the product lesson, not the exact product shape:
The enterprise players are building broad agentic platforms. The prospecting tools are strong but often sales-team centric. Conversation automation tools are channel-first. Support agents are support-first.
Croohm can win by being:
The SMB communication coworker that combines prospect discovery, lead management, WhatsApp-first conversation handling, follow-up automation, and support handoff in one mobile-first workspace.
Strategic Wedge
Start with markets where the pain is immediate:
Small businesses that live on WhatsApp
Agencies handling leads for clients
Workspace/coworking providers
GST/accounting/compliance firms
Real estate and local services
Education consultants
Clinics and appointment-heavy services
D2C/ecommerce sellers
These users do not need Salesforce-level configurability. They need:
Leads coming in
Leads being enriched
Follow-ups happening
Conversations not getting lost
Team members knowing what to do next
Competitive Moat
Croohm's moat should be built around:
India/SMB-specific workflows
WhatsApp-first UX
Prospecting sources like MCA, Google Maps, Meta Leads, Shopify, local directories
Cross-channel conversation memory
Playbooks learned from actual SMB operations
Outcome data from prospecting through support
Trust, auditability, and compliance controls
Research-Based Product Bets
Bet 1: Outcome-Based AI Will Matter
Users will not want to pay only for "AI tokens". They will pay for:
Qualified lead found
Conversation resolved
Follow-up completed
Meeting booked
Lead enriched
Croohm should track these outcomes from the start.
Bet 2: Data Quality Will Be A Differentiator
Every prospecting product claims data. The winner is the one that shows quality and source clearly.
Croohm should make data quality visible in the product.
Bet 3: SMBs Need Guided Autonomy
Enterprise customers can configure agents. SMBs need presets:
"GST setup lead follow-up"
"Coworking inquiry follow-up"
"WhatsApp pricing inquiry"
"No reply in 48 hours"
"New MCA company outreach"
Bet 4: Mobile Control Is Underserved
Most advanced GTM tools are desktop-first. Croohm can own the "approve and operate from phone" use case.
Bet 5: Compliance Is A Feature
For outreach-heavy markets, especially with phone and WhatsApp, compliance cannot be hidden in policy docs. It should be part of lead quality, campaign setup, and agent permissions.
What To Avoid
Avoid:
Becoming a generic CRM clone
Selling raw lead data without workflow value
Letting AI send uncontrolled messages
Building too many dashboards before daily workflows work
Building microservices before product-market learning
Over-customizing before vertical templates exist
Research Conclusion
The market is validating Croohm's long-term direction. The opportunity is not just CRM plus AI. It is an operational coworker for small teams where prospecting, enrichment, conversations, follow-up, and support all share the same context.
Croohm's current goals are already meaningful: lead management, WhatsApp/Facebook/webhook ingestion, campaigns, team inbox, and incremental AI.
But the platform can go much further if built around the right primitives.
North Star
Croohm can become:
The business communication operating system for SMBs, where prospects, customers, conversations, workflows, and AI agents all operate from one shared memory.
This does not mean building every feature immediately. It means every feature should compound toward that system.
Expansion Levels
Level 1: Lead Inbox
Current core.
Croohm helps users:
Capture leads
See conversations
Assign work
Follow up
Send WhatsApp messages
Track campaign source
Success metric:
Fewer missed leads
Faster first response
More organized follow-up
Level 2: Lead Intelligence
Next product layer.
Croohm helps users:
Enrich leads
Score fit/urgency
Detect duplicates
Recommend next action
Show source and confidence
Success metric:
Better qualified leads
Less manual research
Higher reply and conversion rates
Level 3: Dispatch
Prospecting layer.
Croohm helps users:
Find new companies/people
Import MCA/vendor/source data
Run waterfall enrichment
Draft outreach
Start campaigns/playbooks
Success metric:
New qualified prospects created per week
Cost per qualified lead
Reply rate
Meeting/demo/booked rate
Level 4: Conversation Coworker
Assistant inside the inbox.
Croohm helps users:
Summarize conversations
Suggest replies
Detect intent
Ask missing qualification questions
Route to a person/team
Handle safe FAQ responses
Success metric:
Time saved per conversation
Faster response time
Reduced unresolved conversations
Higher customer satisfaction
Level 5: Playbook Engine
Workflow layer.
Croohm helps users:
Convert SOPs into workflows
Run follow-up sequences
Pause for approvals
Track every lead's step
Escalate when stuck
Success metric:
Follow-up completion rate
Stalled lead reduction
Playbook conversion rate
Level 6: Agentic Business Brain
Coordinated autonomy.
Croohm helps users:
Monitor the workspace
Decide what matters
Propose actions
Execute approved work
Coordinate across prospecting, conversations, support, and campaigns
Success metric:
Outcomes completed by Croohm
Human approvals required per outcome
Error/escalation rate
Net time saved
Level 7: Ecosystem Platform
Platform layer.
Croohm exposes:
Agent tools
MCP-compatible access
Partner connectors
Custom data sources
Vertical playbooks
Marketplace of workflows and enrichment providers
Success metric:
Active integrations
Agent/tool usage
Partner-sourced leads/workflows
Revenue from usage/credits
Platform Capabilities To Add Over Time
1. Universal Lead Source Layer
Sources:
MCA new company data
Facebook/Instagram Lead Ads
WhatsApp inbound
Website forms
Google Maps/local businesses
Shopify stores
CSV uploads
CRM imports
Email parsing
Partner APIs
Manual entry
All sources should normalize into the same lead/prospect model.
2. Enrichment Waterfall
Croohm should support multiple enrichment providers and internal sources:
existing CRM data
-> vendor data
-> website/domain lookup
-> social/profile lookup
-> email verification
-> phone validation
-> AI research summary
The system should stop when enough quality is reached, not spend credits blindly.
This lets Croohm prove ROI and price by outcomes later.
5. Vertical Packs
Croohm should ship opinionated packs:
GST/accounting firm pack
Coworking/workspace pack
Real estate pack
Clinic pack
Education consultant pack
Agency pack
Ecommerce pack
Each pack can include:
Lead sources
Fields
Scoring model
Templates
Playbooks
Compliance defaults
Dashboard metrics
6. Agent Permissions
Agents need scoped permissions:
Can read leads
Can enrich leads
Can draft messages
Can send approved templates
Can assign leads
Can start playbooks
Can close conversations
Needs approval above risk threshold
Agents should not be all-powerful.
How Far Croohm Can Go
Near-Term Product
Mobile CRM + WhatsApp inbox + lead capture.
Mid-Term Product
AI lead intelligence and follow-up automation.
Long-Term Product
AI coworker for sales, support, and operations.
Platform Vision
An agentic operating layer for SMB communications where external data sources, business playbooks, customer conversations, and team workflows all connect.
Recommended 18-Month Roadmap
0-3 Months
Finish manual lead inbox
Assignment filters
Broadcast completion
Lead pagination/search
Enrichment schema
MCA CSV import MVP
3-6 Months
Dispatch review queue
Lead scoring
Data quality badges
Outreach draft generation
Basic suppression/compliance center
Conversation summaries
6-9 Months
Scheduled prospect imports
Provider-based enrichment
Follow-up playbook MVP
AI reply suggestions
Email connector
Outcome tracking
9-12 Months
Natural-language assistant
Playbook builder from plain-language SOP
Agent permissions
Cloud Run API gateway
Cloud Tasks for scheduled actions
Vertical packs
12-18 Months
Agent orchestrator
Multi-source prospect search
Usage/credit billing
CRM sync connectors
Advanced analytics/BigQuery
MCP-compatible agent tools
Partner/source marketplace groundwork
Biggest Strategic Risk
The biggest risk is trying to become too many products at once.
Croohm can go far, but the path must stay anchored:
lead
-> conversation
-> next action
-> outcome
If the product keeps improving that loop, every layer compounds. If it drifts into disconnected AI features, it will become hard to explain and hard to trust.
