Product Requirements Document · B2B SaaS · Telegram-First CRM

Rethink CRM: the conversation
is the record.

A conversation-first CRM for early-stage founders who personally lead sales — built to replace the manual logging loop that kills every traditional CRM. The scrappy precursor to Siteline: same instinct (systems that fit how people actually work), rawer execution, earlier stage of the same product mind.

B2B SaaS AI / NLP Telegram Bot WhatsApp-First Sales
Documentation
GitHub → Prototype →
CONVERSATION-FIRST CRM — USER FLOW Founder WhatsApp chat Forward to Telegram Bot Bot Parses Intent + Data Confirmation Card shown CRM Entry Created parse fails Clarification Prompt sent
WIREFRAME — TELEGRAM BOT + CRM DASHBOARD Rethink CRM Bot Forwarded from WhatsApp: Meeting with Rahul from... Deal Captured Company: RahulCo Stage: Intro call Next: Follow up Mon Add any notes? BOT INTERFACE CRM Dashboard Company Stage Last Touch Next Action RahulCo Intro Call Today Follow up Mon Meera Labs Proposal 2d ago Demo Thu BrightStart Closed Won 5d ago Onboarding FinTech Co Cold 12d ago Re-engage WEB DASHBOARD
SYSTEM ARCHITECTURE — RETHINK CRM INPUT PROCESSING STORAGE OUTPUT WhatsApp / Telegram User input Intent Parser Claude API Data Extractor Structured deal model Airtable CRM Deal records Web Dashboard React · read-only Telegram Bot Webhook Telegram Bot receives forwarded messages; Claude API parses intent into structured deal fields

This page is the full Product Requirements Document for Rethink CRM — covering problem framing, founder research, competitive analysis, solution design, the Telegram bot architecture, and the honest trade-offs of a scrappy, unfunded MVP.

Save as PDF →

Document overview
& research scope.

Overview

This document analyses the problem space surrounding CRM for early-stage startups where founders personally manage sales. It covers how sales workflows operate in founder-led startups, where lead tracking breaks down, why existing CRMs fail this segment, and how primary and secondary research validates a narrowed problem statement and solution direction.

Target Segment: Early stage startups, 0–10 employees, founder-led sales, pre-PMF to early PMF, with no dedicated sales operations.

Primary Research

  • Founder interviews and workflow observation
  • Identification of operational friction in sales tracking

Secondary Research

  • Market reports on CRM adoption and startup sales
  • Community forums, product reviews, accelerator insights

The target segment is defined by a structural condition, not company size: founders who are simultaneously the product owner, salesperson, and system administrator, with no bandwidth left for CRM maintenance.

Confidence tags used throughout this document

LevelDescription
High ConfidenceMultiple founder interviews, consistent cross-industry patterns, strong secondary research support.
Moderate ConfidenceLimited primary signals + secondary findings. Valid but requires further validation.
Emerging SignalsEarly observations from few interviews or edge cases. Not yet a confirmed pattern.

Every founder intended
to use the CRM.

01 · Problem

The founder-led sales reality

In early-stage startups, founders handle the full sales cycle with no dedicated sales team or CRM admin. Typical behaviour includes managing 50–150 active conversations simultaneously across WhatsApp, LinkedIn, email, calls, and in-person meetings, with deals progressing informally and context stored in chat threads, voice notes, and memory. A typical deal path: LinkedIn intro → WhatsApp discussion → call → email proposal → multiple follow-ups before close. Because conversations move fluidly across channels, there is no single system of record capturing deal context.

The observable breakdown in sales tracking

When sales volume increases, founders attempt CRM adoption, but a predictable failure cycle follows:

  1. CRM setup and initial contact import
  2. Manual logging required after every conversation
  3. Logging fatigue within days
  4. Partial, inconsistent data entry
  5. CRM abandonment by week 4

CRM abandonment is not a failure of intent. Every founder in the sample intended to use the tool. The failure is structural: the tool demands effort at the exact moment the founder has none to give.

Evidence of revenue leakage

73% Leads never contacted after first message (Salesforce SoS)
93% Deals close at 6th touch or later (HubSpot research)
44% Founders give up after just 1 follow-up (primary + secondary)
60–70% Founders abandon CRM within 3–4 weeks (G2, Capterra, primary)

Not abstract — every family's Sunday evening

“I can't read the doctor's…” — no, this isn't Vitae. Here it's the founder equivalent: “I scroll through 50 WhatsApp messages before every call just to remember what we discussed last time.”

— Composite of 9 founder interviews. The pattern was identical whether the founder sold HR tech, fintech infrastructure, or services.

Industry-level and user-level problem

Traditional CRMs (Salesforce, HubSpot, Zoho) were designed for structured sales orgs: dedicated teams, defined pipelines, consistent data entry, centralised channels. Early-stage startups operate oppositely: founder-led, relationship-driven, informal, conversational, and highly dynamic. This mismatch makes existing CRMs feel too complex, too enterprise-focused, and too maintenance-heavy.

The CRM industry built a hammer optimised for enterprise nails. Early-stage founders are a different fastener entirely — and no one has built the right tool yet.

All existing solutions depend on manual logging. Since conversations occur outside the CRM and logging requires additional effort, the system structurally fails: input data is missing, follow-up triggers never fire, and pipeline visibility becomes unreliable. In India specifically, this is amplified because WhatsApp is the dominant B2B sales channel, yet no CRM can access or interpret personal WhatsApp conversations.

Target user definition

Pre-seed to seed stage, 0–10 employees, early revenue or pre-PMF. Founders are still discovering their sales process. A CRM that imposes structure before that process is understood will always be abandoned. The founder acts as the entire sales function: lead gen, qualification, demos, negotiation, closing, and onboarding — on an extremely time-constrained schedule.

BehaviourDescription
High context switchingFounders switch rapidly between product, hiring, sales, and fundraising.
Mobile-first interactionMany sales conversations happen directly from smartphones.
Low tolerance for admin overheadTools requiring regular manual updates are quickly abandoned.
Relationship-driven sellingTrust and personal connection are primary drivers of early-stage deals.

These traits translate directly into product design constraints: tools must support fast mobile interactions, data capture must require minimal effort, and the system must support unstructured conversational workflows.

The real CRM is starred
WhatsApp messages.

02 · Research

Lead sources & deal lifecycle

SourceRole
ReferralsHighest trust and conversion
CommunitiesEarly traction channels
LinkedInOutbound prospecting
EventsEpisodic bursts of leads

All high-converting lead sources — referrals, communities, events — immediately move to WhatsApp. The CRM never sees the lead at the highest-trust moment of the relationship.

Business impact — illustrative revenue leakage

StageEstimated NumbersExplanation
Active leads in conversation~100 / monthConversations across WhatsApp, LinkedIn, email, calls
Leads not followed up~40~30–40% receive no structured follow-up
Potential deals from these leads2–4Based on 5–10% B2B conversion rate
Average deal size₹1.5LTypical early-stage B2B SaaS/services deal
Monthly revenue leakage₹3L – ₹6LRevenue lost due to missed follow-ups

Estimated combined revenue impact: ₹10L–₹16L per month in leaked revenue across missed follow-ups, incomplete cycles, invisible pipeline, and context reconstruction (~330+ founder-hours/year). Annualized: ₹1.2Cr–₹2Cr.

Primary research methodology

ParameterDetail
Sample10 founder interviews + 1 aggregated insight document (11 total). 2 B2B SaaS founders interviewed separately.
FormatSemi-structured, remote, 45–60 min. Open-ended across lead gen, tracking, follow-up, tools, pain points.
SelectionSmall/early-stage Indian businesses, sub-20 person teams, founder-involved in sales.
Key BiasesSample skews toward non-adopters and unhappy CRM adopters. Zero satisfied CRM users. No pricing sensitivity explored.

The say-do gap

What Founders SayReality
“I track all my leads in Excel.”Business cards pile up post-event. CRM fields left blank. WhatsApp messages to self serve as the real database.
“I remember all the context.”When handing off to a team member, 60–70% of context is lost.
“We follow up 2–3 times.”Leads found untouched for 4 months. No reminder system in place.
“We use a CRM.”Salespeople skip data entry. Teams abandon tools like HubSpot within weeks.

A week in the life (composite)

Deals don't die in a dramatic moment. They die in the 48-hour gap between a good conversation and the follow-up that never happened.

Friction points, ranked

Friction PointEvidence
Follow-ups slip through cracks (7/10 — HIGH)“Some people had not been contacted since like 4 months.” — P06
Manual data entry kills adoption (8/10 — HIGH)“These guys don't always put in all the data.” — P06
Context lost across channels (6/10 — HIGH)“The real problem is remembering the context.” — Unnati
Exhibition lead digitisation bottleneck (4/10 — HIGH)“Digitalization of my leads. Number one.” — P03
CRM too complex for small teams (6/10 — MEDIUM)“Too complicated, too power-user centric.” — P08
Sales-to-CS handover breaks (1/10 — HIGH, B2B SaaS specific)“60–70% of cases, we never got the full context.” — P12

Recurring behavioural patterns

PatternDescription
The Pocket Database (HIGH)Deal context lives in the founder's head. Works until first hire or sick day. Implication: passive capture must replace active note-taking.
Post-Event Decay (HIGH)Events generate 50–400 leads in 2–3 days. By the time leads are entered, the warm follow-up window has closed.
Automation Paradox (HIGH)Founders want automation but resist the structured input it requires. Tools demanding the most data get the least adoption.
Spreadsheet Gravity (HIGH)Even after buying a CRM, founders revert to Excel because spreadsheets impose no structure. Product must feel like a spreadsheet but act like a CRM underneath.
The Build Reflex (MEDIUM)Technical founders build custom CRMs rather than buy. Product must be API-first or builders will build around it.

The Automation Paradox is the core design constraint — founders simultaneously want automation and refuse the data entry that powers it. The only resolution is a system that generates its own data through passive capture.

Key verbatim pain signals

“I was looking at a report where some people had not been contacted since like 4 months.”P06
“The sales team had to manually do a lot of things. They hated using HubSpot.”P12
“The real problem is remembering the context behind each conversation.”Unnati
“I don't think there is value in CRMs for small companies any more.”P07

Every quote is about missing context or lost momentum — not missing features. The product gap is not functional, it is temporal: the right information isn't available at the right moment.

Secondary research — market sizing

Synthesised from 85+ sources and 362 findings across market reports, community forums, and review platforms.

MetricFinding
India CRM market (2024)$2.48B (Fortune Business Insights)
India CAGR19.1% — 72% faster than global 11.1%
Projected 2034$14.24B
Intent vs adoption94% of SMBs (<10 emp) intend to adopt; only ~50% have structured usage
SMB CRM sub-segment16.2% CAGR with no dominant India-first player

Founder time allocation

ActivityTime Allocation
Product & Engineering35%
Customer Conversations25%
Fundraising & Investor Comms15%
Hiring & Team Management12%
Operations & Admin8%
CRM Maintenance5%

CRM gets 5% of founder time by default — and that ceiling doesn't move with better UX. The only lever is reducing how much time the tool requires, not improving how the time is spent.

Three industry insights

1. The Architectural Inversion

Every CRM treats conversation as an input requiring data entry. The winning product treats conversation as the log itself. Kommo comes closest (WABA conversations as deal records) but requires Business API and forces pipeline structure at onboarding.

2. Frequency Drives Retention

Only daily-frequency features (follow-up reminder, pre-call context recall, WhatsApp context capture) can sustain retention. Pipeline analytics (weekly/monthly) cannot. Design must prioritise daily-frequency features above all else.

3. The P0 Chain

Three failures compound in sequence: (1) Input failure — WhatsApp invisibility. (2) Habit failure — manual logging breaks. (3) Output failure — no system knows when to remind. Solving WhatsApp capture and auto-follow-up together addresses all three with a single architectural decision.

Sixteen tools. Zero support
for personal WhatsApp.

03 · Competitive & Prioritisation

The real existing solutions

Before evaluating CRMs, the baseline tools founders actually use:

The most popular CRM among early founders is WhatsApp starred messages. That is the benchmark the product must beat — not HubSpot.

CRM tool landscape — 16 tools, 4 tiers

TierDetails
Tier 1: Traditional CRMSalesforce, HubSpot, Pipedrive, Close. Built for multi-rep orgs. Data entry burden: 30–90 min/day. Founder fit: 1.5–4/10. Zero WhatsApp support.
Tier 2: Indian Mid-MarketZoho, Freshsales, LeadSquared, TeleCRM, Kylas, Interakt. WhatsApp support is WABA-only (business number + BSP middleware at INR 3K–15K/mo).
Tier 3: Founder-Adjacent LightweightFolk, Streak, Bigin, Kommo. Closest to solving founder pain. Each solves one dimension well; all share the same anchor failure: manual logging breaks at week 3–4.
Tier 4: Workaround StackNotion CRM, Google Sheets. Adopted because they impose no structure. Abandoned equally quickly because they provide no guidance.

Three architectural ceilings

Ceiling 1: The WhatsApp Wall

Zero of 16 tools support personal WhatsApp. 8 support WABA only — requiring a separate business number, BSP middleware (INR 3K–15K/mo), and per-conversation charges. Economically and structurally unviable for a founder at <INR 20L ARR.

Ceiling 2: The Manual Logging Loop

The only accurate logging moment is immediately after a conversation — also when founders are least available. Not a discipline problem. A timing problem.

Ceiling 3: The Pricing Cliff

Free tiers are stripped of useful features. Meaningful plans run $25–$75/user/mo. The upgrade ask arrives exactly when founders are ready to pay — but before the tool has proven value.

Feature coverage across 16 tools

FeatureCoverage
Email Integration13/16 have it. Solved. Not a differentiator.
WhatsApp Integration0/16 support personal WA. 8 support WABA only. Architecturally unsolved.
Mobile App13/16 have one. All are scaled-down desktop. None are mobile-first.
Auto Follow-up Reminder7/16 have partial capability. All reactive, not proactive.
Free Tier5/16 have a genuine free tier.
Stalled-Deal AlertingPartial in Pipedrive, Streak, Freshsales (paid). No tool proactively surfaces “this deal needs attention now.”

Email integration is table stakes. WhatsApp integration is the moat. The feature every Indian founder needs most is the one no tool has built — and the one no global competitor is incentivised to build.

Prioritisation: the P0 chain

ProblemDescription
P1 — WhatsApp conversation invisibility [HIGH]Personal WhatsApp cannot integrate with any CRM due to Meta API restrictions. 100% of target users affected.
P2 — Manual data entry failure [HIGH]Founders stop logging within 2–3 weeks. 8/10 cited as primary failure cause.
P3 — No proactive follow-up nudging [HIGH]Tools are passive. 7/10 cited follow-up failure as direct cause of lost deals.
P4–P6 — symptomatic [MODERATE]Pre-defined pipeline assumption, no behavioural feedback loop, 30-second mobile failure — downstream of P1/P2.
P7–P10 — adoption barriers [HIGH/MODERATE]Pricing cliff, slow time-to-value, customisation paralysis, team-designed UX — tractable once P1–P3 are solved.

WhatsApp invisibility (P1) → manual logging failure (P2) → follow-up failure (P3) → DEAL LOSS. Solving P1 + P2 + P3 together as a single architectural decision, not three separate features, is the product strategy. Treat conversation as the primary record, not the input to a separate record.

Why this problem, why not others

ProblemReason Excluded
D2C/B2C founder salesStructurally different (platform analytics). Would dilute core architectural solution. [Excluded]
Team CRM and collaborationRequires baseline individual data quality that doesn't exist yet. [Phase 2]
Analytics and reportingRequires data foundation. Reporting on incomplete pipeline produces misleading results. [Phase 3]
Enterprise CRM capabilitiesTarget user has no sales team, no multi-stage approvals. [Out of scope]

Trade-offs resolved in prioritisation

Trade-offAnalysis
Privacy vs convenienceAny conversation capture requires access to content. Founders must explicitly consent. Trust-critical design constraint.
Simplicity vs pipeline completenessSolving only P0 chain produces ~25–40% explicit coverage + AI inference. A 70–80% accurate pipeline with no manual effort beats a 100% accurate one the founder has abandoned.
Speed vs accuracyAI-inferred deal stages won't be 100% accurate. A pipeline with minor inaccuracies the founder uses is categorically more valuable than an accurate pipeline they've abandoned.

All three trade-offs resolve in the same direction: a slightly imperfect tool the founder actually uses is categorically better than a perfect tool they abandon. Adoption is the primary success metric, not accuracy.

Narrowed problem statement

Indian early-stage B2B founders who personally lead sales (0–5 person teams) lose deals not because they lack a CRM, but because every CRM requires the one behaviour they will not sustain: manually logging their conversations. The result is an invisible pipeline, missed follow-ups, and lost deals. The problem is architectural, not motivational. They lack a tool whose input model matches their actual workflow: short, asynchronous, multi-channel, WhatsApp-dominant, phone-first.

The conversation
is the record.

04 · Solution & Persona

The P0 chain — what we are solving

Three root-cause failures compound in sequence. Solving the first breaks the entire chain downstream:

What we built

Founder CRM is not a simpler version of HubSpot. It is a fundamentally different product category — one that inverts the relationship between conversation and record. In every existing CRM, a conversation is an input that requires a record. The founder must stop, open the CRM, and log what happened. In Founder CRM, the conversation IS the record. The founder forwards a WhatsApp thread, speaks a voice note, or sends a screenshot — and the system builds the deal record automatically.

Every CRM TodayFounder CRM
Conversation requires manual loggingConversation IS the log
Founder opens CRM after every callBot captures passively via forward or voice
Value arrives weeks after consistent usageValue arrives in the first 30 seconds — Day 1
Built for VP Sales with 10 repsBuilt for the founder who is the rep
Fails when the founder goes dark for 2 weeksCatches up automatically when they return
Desktop-first, mobile as afterthoughtMobile-first bot. Desktop as visibility layer only.

Architecture in one sentence: A Telegram bot captures sales context passively from the founder's existing behaviour. A desktop platform makes that context visible. Neither requires the founder to do anything they are not already doing.

The four non-negotiables

Zero Manual Logging

If the founder must type anything consistently to keep the system alive, the system dies. Full stop. (8/10 founders cited data entry as primary CRM failure cause)

Instant First Value

Value must arrive before end of Day 1. Not after 2–3 weeks of data input. (Abandonment peaks at week 2–6 across all 16 tools)

Mobile-First Capture

70%+ of sales activity happens on mobile in 30-second interaction windows. Any product requiring desktop for capture fails architecturally.

Conversation-Driven

The conversation IS the log. Not an input to a separate log. Any product that inverts this hits the same ceiling as all 16 tools evaluated.

Why Telegram as the capture layer

Primary user persona — the founding seller

Arjun Mehta, 29

B2B SaaS — HR tech tool for SMEs
CompanyPre-seed. 3-person team. Bengaluru (applicable to Mumbai, Delhi NCR, Pune, Hyderabad)
Sales RealityManages 10–20 active conversations simultaneously. Deals span LinkedIn intro → WhatsApp discussion → call → email proposal → multiple follow-ups. No dedicated sales hire.
Current ToolkitWhatsApp (primary), Gmail, LinkedIn, Google Sheets or Notion as a makeshift CRM, personal memory for context.
Primary GoalClose deals without becoming a CRM administrator. Keep pipeline moving without stopping to update systems.
Core Frustration“I scroll through 50 WhatsApp messages before every call just to remember what we discussed last time.”
CRM HistoryTried HubSpot Free, Zoho, Notion CRM. Abandoned all within 6 weeks. Reason: manual entry fatigue, not product quality.
Technical ComfortComfortable with apps and smartphones. Uses Telegram already. Not a developer. Zero tolerance for setup overhead.

Key assumptions behind the bet

AssumptionEvidence / Basis
WhatsApp will remain the primary B2B sales channel [HIGH]8/10 primary, secondary consensus, Meta India stats
Founders will not sustain manual data entry [HIGH]8/10 primary, 18-month CRM abandonment rates, identical failure across all 16 tools.
Founders will engage with <30 sec capture from phone [MODERATE]3/10 cited 30-sec window; secondary research on mobile CRM adoption
Founders will trust AI summaries if they can review/edit quickly [MODERATE]Trust earned incrementally, not assumed from Day 1
Price ceiling: INR 1,500–3,000/month [HIGH]Products above require explicit ROI in deal terms

Five flows.
Zero forms.

05 · Product Flow & Platform

The product has five distinct flows, each designed to require zero friction at the point of use. The founder never opens a CRM interface to log something. They interact with the Telegram bot in the same way they already interact with colleagues — via forward, voice note, or text command.

Onboarding

StepActionWhat the Founder SeesTime
1Visits landing pageOne-line value prop. Single CTA: 'Start Free'30 sec
2Clicks CTA — opens TelegramDeep link opens Telegram app. Taps 'Start'.5 sec
3Bot sends welcome message'Forward me any WhatsApp conversation or send a voice note after a call. No setup needed.'Instant
4Founder forwards first conversationBot extracts lead card. Founder confirms. First deal logged.< 30 sec

Input layer — three interchangeable modes

Input TypeFounder ActionWhat AI Extracts
WhatsApp Conversation ForwardLong-presses a thread → Forward to Founder CRM botContact name, company, deal stage signals, intent phrases, budget mentions, objections, next step indicators
Voice Note (Post-Call)Opens bot → Presses mic → Speaks a 30-second summaryStructured deal note: contact, discussion points, outcome, next action — transcribed and tagged automatically
Screenshot / ImageScreenshots a business card, LinkedIn message, email → sends to botContact identity, company, stated intent, message summary

Deal capture — the founder confirms, doesn't fill in

StepSystem ActionFounder Sees
1AI processes inputTyping indicator. 2–3 second processing.
2Bot generates Lead CardContact, Company, Stage, Budget signal, Key blocker, Suggested next action
3Bot checks missing fieldsIf contact name or stage missing: bot asks one clarifying question only. Never more than one.
4Founder confirms or editsQuick-edit buttons for common changes. No form. Inline reply to edit any field.
5Deal stored'Saved. Ankit Sharma from TechCorp is now in Evaluating. I'll remind you in 3 days if no update.'

Context recall — the highest-frequency use case

The founder needs deal context in the 5 minutes before a call. Currently this requires scrolling 50+ WhatsApp messages. The product reduces this to a single command:

InputFounder ActionBot Response
Slash commandTypes: /context TechCorpFull deal brief: last conversation, current stage, open questions, suggested talking points
Natural language (text)Types: 'What's the status with Rahul from Juspay?'Conversational summary of interactions, last touchpoint, next action pending
Natural language (voice)Speaks: 'Prep me for my call with TechCorp in 10 minutes'Voice-friendly deal brief, optimised for a walking/commuting founder.

Nudge and follow-up — triggered by inactivity, not calendar

TriggerBot Message
No update for 3 days (Evaluating)'You haven't updated Ankit Sharma in 3 days. Last status: evaluating pricing. Follow up?'
No update for 7 days (any active stage)'Rahul at Juspay has been quiet for 7 days. Follow up or mark cold?'
Daily digest (8 AM)'You have 3 deals needing follow-up today. 2 are overdue. Your hottest deal: [Company X].'
Deal marked Closed'Congratulations on closing TechCorp! Want to log what made this one work?'

Follow-up rules by stage: New (nudge at 24h), Contacted (3d), Evaluating (3d), Proposal Sent (2d). Closed deals exit the nudge queue.

A day, before and after

Before — Reactive Mode

  • 8 AM: Scrolls WA to reconstruct yesterday. 12–15 min.
  • 10 AM: Good discovery call. Means to log it later. No record created.
  • 3 PM: Warm lead from 2 weeks ago goes cold, unnoticed.
  • 8 PM: Investor asks for pipeline update. Reconstructs from memory.

After — Structured Mode

  • 8 AM: Reviews daily digest. 3 deals need follow-up. 2 min, day planned.
  • 10 AM: Sends a 30-second voice note after the call. Deal created automatically.
  • 3 PM: Bot nudges on the 12-day-quiet lead before it goes cold. Deal saved.
  • 8 PM: Types /deals. Live pipeline snapshot in 10 seconds. Accurate.

Desktop platform — a visibility layer, not a capture layer

The desktop platform is a visibility layer, not a capture layer. The bot captures. The desktop displays. The founder never needs to open the desktop to keep the system alive. This is a deliberate inversion of the standard CRM architecture: in HubSpot, Pipedrive, or Zoho, the desktop is where work happens. In Founder CRM, the Telegram bot is where work happens and the desktop is where patterns become visible.

FeatureWhat It ShowsPhase
Pipeline BoardKanban view across 5 stages. Read-only in Phase 1.Phase 1 (P0)
Deal Timeline ViewEvery touchpoint for a contact in a single thread — chronological.Phase 1 (P0)
Contact Heat ScoresHot / Warm / Cold, calculated from recency, sentiment, engagement frequency.Phase 1 (P0)
Follow-Up QueuePriority-ranked list of deals needing attention.Phase 1 (P0)
Win/Loss PatternsClosed-won vs lost ratio, average deal length, common blockers. Requires 20+ deals.Phase 2
Investor Pipeline ExportOne-click clean pipeline snapshot for investor conversations.Phase 2
Team Pipeline (Shared)Multi-user deal visibility when the founder makes their first sales hire.Phase 3

Read-only in Phase 1 by design: editable pipeline boards invite manual entry, letting the same failure re-enter through the desktop. Heat score over deal stage as primary sort, because 4/10 founders do not think in pipeline stages.

MVP now.
Sales coach later.

06 · MVP & Moonshot

MVP in one sentence: A Telegram bot that captures sales conversations from WhatsApp forwards and voice notes, creates structured deal records automatically, and reminds the founder when deals need attention — with no setup, no data entry, and no learning curve.

The five hypotheses driving MVP design

H#HypothesisKill Signal
H1Founders will forward WhatsApp conversations to a Telegram bot when it requires < 5 seconds.< 30% forward ≥1 msg/week in 30 days
H2Founders will use voice notes to log post-call summaries.< 25% send ≥1 voice note in 14 days
H3Founders will use /context to recall deal information before calls.< 40% use /context within 30 days
H4Nudge-triggered follow-ups increase second-touchpoint rate within 5 days.No measurable improvement
H5Founders will prefer a bot-first interface over a traditional CRM interface.Majority of updates happen via desktop form entry

In scope for MVP

Capture

  • WhatsApp conversation forward → AI extracts lead card
  • Voice note → Whisper transcription → structured deal note
  • Screenshot / image → OCR + AI extraction

Recall & Visibility

  • /context — full deal brief in < 5 seconds
  • /deals — pipeline snapshot, heat-ranked
  • Generic 5-stage pipeline (New → Contacted → Evaluating → Proposal → Closed), not editable in MVP

Follow-Up

  • Inactivity-triggered nudges, stage-based timers
  • 8 AM daily digest: overdue, needing follow-up, hottest deal

Moonshot — AI Sales Coach (Phase 3+, separate surface)

The AI Sales Coach moves Founder CRM from a context management tool to an active competitive intelligence layer. The system continuously enriches every deal record with scraped intelligence about the prospect, their company, and their buying context — and uses this to tailor the founder's pitch, timing, and talking points.

“I know my product can solve their problem. But I go into every call not knowing enough about their business. I wing it. Sometimes it works.” Navdeep, B2B Services Founder
TrackWhat HappensOutput to Founder
Continuous EnrichmentAs soon as a company is identified: LinkedIn page, funding news, headcount signals, job descriptions, leadership changes, competitor mentions.Deal record enriched silently. Richer context card on recall.
Pre-Meeting Coaching'Coach me for my call with TechCorp tomorrow.' Pulls enrichment + deal history + objections + competitor context.What matters now, the angle most likely to land, tailored talking points, known objections, one unlock question.
Touchpoint SequencingAnalyses closed-won deals over time to find sequences correlating with faster close rates.Personalised outreach sequence recommendation.

Not in the MVP because it requires a deal corpus to be valuable (20–30 logged deals minimum), a separate web surface, non-trivial scraping infrastructure, and H1–H3 validated first.

Strategic implication: The AI Sales Coach is the product's long-term moat. The data asset the bot builds through passive capture becomes the training input for the coach. The flywheel: more captures → richer intelligence → better coaching → higher win rates → stronger retention.

Behaviour, not sentiment,
is the signal.

07 · Metrics & Trade-offs

Metrics are behaviour-first. We measure what founders do, not what they say they will do. User-reported satisfaction is not a success signal. Behaviour change is the only signal that matters.

Metric30-Day TargetKill SignalHypothesis
Weekly Forward Rate> 50%< 30%H1
Voice Note Adoption> 40%< 25%H2
/context Usage Rate> 60%< 40%H3
Nudge Response Rate> 45%< 20%H4
Bot vs. Desktop Update Ratio> 80% via bot< 50% via botH5
D7 Retention> 65%< 40%Core
D30 Retention> 40%< 20%Core

What this product does not solve

AreaWhy It Is Out of Scope
Deal quality or product-market fitFounder CRM cannot make a weak product win deals. It surfaces context and timing.
Outbound lead generationThe product manages conversations after they begin. It does not generate new leads.
Sales coaching / process definitionThe MVP does not tell founders how to sell. It captures what they do. Process intelligence is Phase 3.
Prospect responsivenessThe product cannot make a prospect reply. It ensures the founder follows up.
Stop-start founder rhythmWhen founders go dark for 2+ weeks, pipeline data goes stale. The system catches up when they return.

Known trade-offs in MVP design

Trade-offWhat We Gave UpWhy We Made It
Telegram over WhatsApp nativeThe forward action adds 2–3 seconds of friction.No viable alternative. Personal WhatsApp integration is permanently prohibited.
Generic 5-stage pipelineFounders with non-standard processes will find the stages don't map to their reality.Setup friction kills adoption faster than stage mismatch.
No desktop at MVPFounders who expect a dashboard will feel the product is incomplete.A desktop with no validated data asset is a failure surface, not a value surface.
Bot-only interfaceNot optimal for bulk deal review or complex pipeline analysis.The target persona's primary device is mobile.
Founder-triggered capture onlyTrue ambient capture is not possible.DPDPA 2023 compliance requires explicit consent.

Dependency risks

Rethink CRM is the precursor to Siteline — same instinct for reading how people actually work before designing a system around them, applied a stage earlier and rawer. Where Siteline shipped a matured, validated product, this was the discovery-through-MVP exercise that sharpened the instinct: solve the architectural constraint, not the feature gap, and measure adoption before accuracy.