Week 5 · Group 4 · Rethink Systems

Signal
/ CRM

Intelligence layer for founder-led B2B sales

The $73.4B CRM market is built for professional salespeople. This is not a CRM. It reads your inbox and tells you who you're about to lose.

Status: v1.0 Architecture Date: 11 March 2026 Research: 6 founder interviews · 80+ sources

The Problem

Early-stage B2B founders who personally close every deal have no system that tells them which conversations need attention today. At approximately 40 simultaneous active relationships, manual prioritization becomes unreliable, and qualified deals decay silently — not from rejection, but from the absence of a timely follow-up signal.

Core finding: Deals die not from rejection but from silence. The data showing decay exists in email and messaging logs. No tool reads it and acts on it. The failure is an information architecture failure, not a motivation failure.
₹1L+
Monthly recurring revenue deal lost to follow-up failure — went silent in WhatsApp for 2 months
40+
Simultaneous conversations = ~240 live data points that exceed human working memory
4 min
Prevention cost. Recovery takes 2–4 weeks and usually fails.

The 5-Stage Failure Cascade

1

Single-Channel Initiation

A lead arrives via LinkedIn, WhatsApp, or conference. The founder responds. No system entry is created. Risk is invisible.

2

Channel Fragmentation

Within days, the exchange migrates: LinkedIn → email → Zoom → WhatsApp. Context fragments across 4–5 apps. Context reconstruction required at every re-engagement.

3

Open-Ended Commitments Disappear

"Let me check with my team" — no deadline, no calendar trigger, no system entry. These vanish into unstructured transcript text that no one reviews. ← Signal intercepts here.

4

Cognitive Overload Threshold Crossed

At ~40 conversations, the founder shifts to recency-based prioritization. High-value quiet prospects get structurally deprioritized by default, not by decision. ← Signal intercepts here.

5

Silent Decay → Reactive Panic → Loss

Founder discovers the decaying deal by accident. Recovery attempt fails. Deal logged as "cold lead." Cause of death invisible. No learning is possible. Pattern repeats.

Research Findings

Six structured founder conversations spanning different stages, GTM motions, and industries. Each session ran ~60 minutes. Every pre-research assumption was stress-tested.

The Disciplined Founder at the Ceiling

4-tab Excel workbook, 2–3 hours of daily updates. His system accurately logs everything that happened. It generates no signal about what should happen next. Some deals fall through — not by decision, but because he runs out of hours.

The Scaling Founder Watching Revenue Leak

Runs HubSpot, a custom internal tracking portal, and Firefly on every call. A ₹1L/month prospect had expressed genuine interest 2 months earlier — unfollowed-up. Neither HubSpot nor the portal surfaced it. He was reminded during a research conversation.

The Repeat Founder Who Learned the Hard Way

Used HubSpot for 4 years with a Chief of Staff enforcing logging. Result: 50% of features used, retrospective logging, and a tool that never once told anyone who needed a follow-up today.

Key Finding

The original hypothesis — "founders don't log, so there's no data" — is wrong. Every founder already captures data. The gap is intelligence on top of the data, not the data itself. The intervention point is downstream of capture.

Assumptions Overturned by Research

AssumptionStatus After Research
The core issue is adoptionPartially wrong. Adoption was achieved by one founder (4 years enforced HubSpot). Deals were still lost. Adoption is necessary but insufficient.
Deals are lost because data is not recordedWrong. Every founder already captures data. The gap is intelligence on top of the data, not the data itself.
The intervention point is at captureWrong. The intervention point is downstream of capture — at the moment when a prioritized follow-up signal is needed and none exists.
Pipeline visibility is the product's core valueOverturned. The repeat founder achieved complete visibility (4 years enforced HubSpot). She still lost deals. The real need is direction: which conversation, why now.

Primary Persona

This persona is a deliberate bet, not a confirmed archetype. It is built from three adjacent evidence sources. The MVP is the validation instrument.

The Solo Closer
B2B FOUNDER · 1.5–3 YRS IN · 10–30 EMPLOYEES · NO DEDICATED SALES HIRE

B2B founder, no dedicated sales hire, or one junior SDR who generates leads but does not close. The founder closes every deal personally.

A spreadsheet or basic CRM, maintained with decreasing discipline over time. Secondary system: the email inbox, which doubles as an unintentional CRM.

The system reflects the past but generates no signal about the future. At ~40 simultaneous conversations, manual prioritization becomes unreliable.

Three conversations that need attention today, ranked by urgency, with enough context to act immediately. No configuration required. No maintenance required.

A specific, memorable loss. A deal they thought was progressing that closed with a competitor — and when they checked, their last contact was three weeks ago.

If the product looks like a CRM, they will assume it requires the same discipline. The first week must produce value before they have done anything to earn it.

Product Direction

"It reads my inbox and tells me who I'm about to lose."

Signal is an active prioritization layer that converts a founder's existing email and calendar data into a directed daily action list — telling them which conversations need attention today, why they are urgent, and what context they need to re-engage effectively.

Deal State Machine

Warm

Active — reply received within decay threshold window

Cooling

Decay threshold reached — no reply, no commitment on file

Critical

Commitment overdue — prospect made a promise and went silent

Stalled

No activity 14+ days — window may be closing

Daily Brief — Per Deal Card Output

FieldWhat It Shows
ContactName + company (e.g. Meera Iyer · Acme Corp)
Status badgeCritical / Cooling / Warm / Stalled with color indicator
Last contact"6 days ago · via WhatsApp"
Verbatim commitmentExact words: "I'll check with my team" — with source channel and date. Never paraphrased. Never inferred.
Channel coverage barShows which of the 5 channels Signal is currently monitoring for this contact
Action button[Open Thread] — 1-click deep link to originating Gmail thread, WhatsApp conversation, or LinkedIn DM
Anti-hallucination rule (non-negotiable): The system must never paraphrase prospect language. Every alert displays the verbatim text of the detected commitment — date-stamped, channel identified, with a deep link to the originating thread. Trust is built on receipts, not summaries.

Technical Architecture

Architecture flow: Input Layer → Intelligence Layer → Output Layer

Input Layer — V1 Channels

ChannelSetup RequiredTime to ActivatePriority
GmailGoogle OAuth consent screen<5 minutesP0 — must connect at signup
Google CalendarSame Google OAuth (bundled)<1 minuteP0
WhatsApp BusinessBSP account (Wati/Interakt/AiSensy) + Business number2–7 business daysP0 — onboarding begins at signup
LinkedIn DMLinkedIn app OAuth (official API only — Section 8.2 compliance)4–6 weeks review + 5 min consentP0 — review initiated before launch
Zoom TranscriptZoom app install + host recording enabled<10 minutesP1 — optional, high value

Intelligence Layer — Commitment Detection Logic (V1)

IF message contains:
  • future-tense first-person phrase
    ("I'll", "let me", "I will", "main baat karta hoon")
  AND temporal marker
    ("next week", "by Friday", "a few days", "kal")
  AND no calendar invite exists from this domain

THEN Activate Commitment Timer for this contact

CROSS-CHANNEL MERGE RULE:
IF prospect silent on WhatsApp for 48h
AND replied to Gmail today
THEN reset all decay timers — deal is alive in a different channel

Decay Thresholds

ChannelBase DecayWith Active CommitmentRationale
Email4 days2.8 days (−30%)B2B email open rates ~30%; 4-day silence = seen-but-not-replied
WhatsApp48 hours33 hours (−30%)98% open rate; 48-hr silence = deliberate non-reply
LinkedIn DM3 days2.1 days (−30%)Mid-frequency channel; 3-day silence is meaningful
Zoom (post-call)24 hoursMeeting happened; no follow-up within 24 hrs = decay begins

Solutions Evaluated — What Was Eliminated and Why

SolutionScore (/25)VerdictKey Reason
Gmail + Calendar API (Signal V1)25/25Build nowCorrect architecture, zero config, Week-1 value, legal safe
HubSpot + Zapier15/25Hard eliminatedDirect evidence of failure in our interviews. 4 years enforced logging, still produced zero proactive output.
Browser Extension DOM Scraping16/25Hard eliminatedLinkedIn ToS Section 8.2 violation. hiQ v. LinkedIn precedent — ToS breach is the operative risk. MV3 kills continuous monitoring.
Unofficial WhatsApp API (Baileys)20/25*Existential riskSingle Meta enforcement = account ban = all deal visibility gone instantly. No appeal path. India B2B deals run on WhatsApp.
AI Draft Reply GeneratorPermanently eliminatedDestroys the product's core competitive advantage. Founder-to-founder authenticity is the moat. This line never moves.
Daily GTD Discipline SystemStructurally impossible40 conversations × 6 variables = 240 data points. Exceeds human working memory regardless of discipline or intention.

Non-Negotiable Product Rules

These are not features to be balanced against each other. They are architectural constraints. Any feature that violates a rule does not ship.

R1

Passive Always

If a feature requires the founder to trigger it, it is not a Signal feature. Manual input breaks at 40+ conversations. Evidence: every interview.

R2

Evidence Before Conclusion

Every alert shows the verbatim quote that triggered it. No AI summaries. No paraphrasing. Receipts build trust. Week-1 trust is the product.

R3

First Win in 48 Hours

Signal must surface a real forgotten deal before the founder has done anything to earn it. Adoption trigger for Solo Closer is a specific memorable deal-save. Must happen Week 1.

R4

Honest About Limits

Channel coverage bar visible at all times. Alert language hedges on channels not connected. A trusted 70% beats a mistrusted 65%.

R5

Never Automate the Human Part

Signal never writes, drafts, or sends messages on any channel. It surfaces context only. Founder outreach trust premium is the competitive advantage. This boundary is permanent — not in V1, V2, or any future version.

R6

Legal-First Architecture

Every data source uses only official APIs with user consent. No browser extensions, no unofficial APIs. Disclosable in VC due diligence. Sustainable at Series A scale.

R7

Channel Data Is the Moat

Every entity and commitment record carries a channel field and raw message reference from day 1. Multi-channel coverage is what competitors cannot replicate. The data model must not be retrofitted.

V1 to V2 Roadmap

FeatureV1V2Reason for Deferral
Gmail ingestionMUST HAVECore architecture
Google Calendar ingestionMUST HAVECore architecture
Daily Brief (ranked cards)MUST HAVEPrimary output
Verbatim quote retrievalMUST HAVETrust foundation (R2)
WhatsApp Business APIMUST HAVEHighest-volume India B2B channel. Requires business number + BSP approval.
Configurable decay thresholdsShould HaveDefaults sufficient for V1; personalize after 60-day calibration
AI deal qualification scoringV2Requires 60–90 days of history per founder before meaningful signal
Automated outreach / AI email draftsNEVERNEVERDestroys founder outreach trust premium. Permanently eliminated.
Product Vision — Moonshot: The long-term vision is an AI Sales Co-Founder — a system that doesn't assist the founder with sales but thinks alongside them in real time. It carries persistent memory of every relationship across every channel, understands deal psychology, and encodes Indian B2B sales culture. The founder doesn't manage it. It runs alongside them.