December Unwrapped: We're Building Workers, Not Tools | Niti AI Manifesto, Dec 2025
Niti AI | Dec 2025
Our Manifesto
Hey there! 👋 December was our most transformative month yet. We didn't just ship features: we fundamentally reimagined how AI agents work for B2C growth teams. We rebuilt our entire architecture around real marketing roles, codified our product philosophy into the 3Cs framework, and converted our first customer from the evaluation pipeline. Here's what went down.
Announcements 📊
Role-Based AI Agents: Your Marketing Team in a Box
We moved from generic "AI assistant" to specialized agents that mirror real marketing roles:
Growth Analyst: Answers "Show me revenue by cohort" or "Why did churn increase?" Generates insights, computes metrics, performs root cause analysis.
Campaign Manager: Tracks campaign performance across channels. Unified view of retention campaigns + ad performance with customer-level attribution.
Marketing Strategist: Orchestrates segmentation, offer generation, and playbook creation based on your business goals.
Offer Manager: Dynamic offer modeling with margin awareness. Optimizes discounts, bundles, and non-monetary offers in real-time.
The unlock: These aren't features bolted onto a platform. They're autonomous agents that replace expensive strategic work at 93% cost reduction.
The 3Cs Framework: Our Product Philosophy
We spent December obsessing over three principles that separate useful AI from AI theater:
CLEAR - Insights in plain language, not technical jargon. "High Value Customer Rate declining with AOV increasing" instead of acronym soup. Metrics organized by priority. Task views that surface what needs action.
COHESIVE - 360° data coverage across your entire stack. Bidirectional flows between paid ads and retention behavior. Unified view across platforms: no more data silos between acquisition and retention teams.
CREDIBLE - Zero hallucinations. Every insight backed by actual data with Root Cause Analysis showing why metrics moved. Transparent reasoning so you see the data behind every recommendation.
This is our competitive moat against platforms slapping GPT wrappers on dashboards.
Agent Primitives: Programmable AI That Learns
We introduced standing instructions and feedback mechanisms that make agents programmable to your business:
- Set guardrails: "Always prioritize margin over revenue" or "Never discount hero products below 15%"
- Provide feedback: Agents learn from approved strategies, successful segments, effective offers
- Contextual memory: Agents remember your business model, competitive position, and margin constraints
The result: AI that gets smarter about your business specifically, not generic best practices.
Human-in-the-Loop Workflows: Automation with Control
We built approval gates for repeated tasks: what we call "workblocks":
- Strategy approval before campaign launch
- Segment review before targeting
- Offer validation before discounting
- Performance review before budget reallocation
This isn't full autonomy (yet). It's automation of repetitive analytical work while keeping strategic control where it belongs: with you.
Product releases 🖥️
Metric Tree Visualization
We built a visual tree showing how 100+ KPIs influence each other:
- Upstream drivers vs. downstream outcomes
- Tier classification (Critical → Info) for prioritization
- Time series for influenced/influencing metrics side-by-side
- Proper units and formatting (currency, percentages, counts)
Example: See that "Repeat Purchase Rate" drives "Customer LTV" which drives "Revenue": and which moved first when performance changed.
The Ads ↔ Retention Bridge
Not all customers from the same ad campaign behave the same. We built bidirectional data flows between Meta/Google Ads and retention behavior to show:
- Which ads drive profitable vs. margin-killing customers
- Customer-level attribution: Which ad → Which customer → What LTV
- Campaign performance metadata layer unifying paid and owned channels
Real impact: One customer identified that 40% of their ad-driven cohort never repurchased. They reallocated spend to creative driving profitable customers: same budget, 4x better ROAS.
Root Cause Analysis with 5 Whys
Traditional dashboards tell you "Revenue dropped 15%." We tell you why with recursive logic:
- Revenue dropped → repeat purchase rate declined
- Repeat rate declined → high-value customers less active
- High-value churn → product discovery stalled
- Discovery stalled → email engagement dropped weeks 3-5
- Email engagement dropped → subject lines weren't margin-aware
All explained in plain language a marketer can understand and act on.
Agent Reasoning Architecture
Under the hood, we rebuilt everything:
- ReAct framework for recursive reasoning through complex problems
- MCP integration for tool calling
- Streaming responses so you see agents think in real-time
- Validation layers preventing hallucinations
This is infrastructure that would take competitors 12-18 months to replicate.
Chat Interface
Ask questions to specialized agents in natural language:
- "Show me my most profitable customer segments"
- "Why did my email campaign underperform?"
- "Create a winback strategy for lapsed customers"
Generative UI adapts to your question. Follow-ups maintain context. Export insights to Slack or email. The experience feels like messaging your best strategist: except this one works 24/7.
Team and social updates 🛠️
Achieving Ramen Profitability
We hit operational efficiency giving us infinite runway at current ARR. The team is lean and delivering 10x output per person through AI tooling, clear ownership, and ruthless prioritization. This puts us in control of our timeline: we grow when product-market fit is proven, not because we're burning runway.
GTM Team Additions
Bharani joined to automate competitor analysis: web scraping + LLM aggregation for personalized outreach at scale.
Rishu owns outbound automation and pipeline management: CRM workflows, lead scoring, metrics tracking.
Result: Enterprise-grade GTM with startup budget.
Forward Deployed Model
Our engineers own both product development and customer success for their features. When a customer asks "Why is this metric showing X?" the engineer who built it answers directly. No handoffs. No translation loss. Product improvements ship within days of customer conversations.
Customer stories ✨
First Pipeline Conversion: Pokonut
Pokonut became our first conversion from evaluation to signed, paying customer in December. Their path: Initial skepticism → Saw insights on their data → Requested deeper analysis → Converted to annual contract.
The unlock: They saw value before we asked for payment. Product-led conversion, not sales-led convincing.
Dwij: When 40% of SKUs Kill Margins
Dwij (upcycled fashion) came thinking they had a retention problem. Product-level margin analysis across 6 categories revealed something worse:
40% of SKUs were margin-negative after accounting for production costs, shipping, returns, and discount dependency. They were acquiring customers profitably on hero products, then losing money on cross-sells.
Our recommendation: Strategic SKU rationalization. Kill the margin-killers, double down on profitable categories.
Impact: Strategic product decisions based on true unit economics, not just revenue contribution. One decision paid for our annual fee.
Weekly Automated Intelligence
Customers now receive automated weekly reports with AI-generated insights and ready-to-execute retention strategies.
Example: "Your repeat purchase rate among November first-time buyers is tracking 12% below October cohort. Recommended action: Launch 'Product Discovery' campaign targeting single-SKU buyers with educational content + complementary product offers at 20% off. Expected lift: 2-4% conversion, 2.5x ROI."
What used to take 6 hours of analysis now takes 20 minutes of review and execution.
Customer campaigns launched 🚀
Youthnic Agency Partnership: B2B2C Model
We validated a new GTM motion: Agencies use Niti AI to generate retention strategies for their brand clients, white-label our insights, add creative execution, and charge normal retainer.
Why it works:
- Agencies get differentiation and faster delivery
- Brands access sophisticated intelligence without hiring strategists
- We get B2B2C distribution (one agency = 10 customer logos)
Youthnic is testing with 3 brands. Success means 5-10 more agency partnerships in Q1 2026.
Qaadu: Rescuing 99% Inactive Customers
Qaadu (kids fashion) faced brutal economics: 25% repeat rate, 99% inactive last 90 days, 40% GMV discounted away. We set up Meta + WhatsApp campaigns with margin-aware segmentation:
- High-Margin Loyalists: VIP treatment, early access, premium messaging (no discounts)
- Discount-Dependent: Strategic deals on clearance inventory
- Lapsed High-Value: Founder outreach with satisfaction guarantee + 25% off
Campaigns launch early January with automated WhatsApp flows live.
Wakefit: Mattress Upsell Opportunities
Multiple insight cycles identified customers who bought pillows/bedding rarely upgraded to mattresses despite high satisfaction. Our strategy: Target pillow buyers 60-90 days post-purchase with "Sleep System" bundles leveraging margin structure (preserve mattress margins, use accessories as loss leaders).
Strategy handover in progress. This demonstrates strategic depth: not "send 20% off" but "leverage your product portfolio and margin structure for profitable growth."
GTM Updates 📈
Lead Scoring: Routing for Efficiency
We implemented automated lead scoring separating high-value signals (human follow-up: $5M+ revenue, clear budget authority) from low-value signals (auto-route to self-service: <$2M revenue, no budget).
Impact: Sales team focuses on qualified opportunities. Conversion rates up 3x.
ICP Refinement: MM+ at $1K/mo Minimum
Target: Mid-market+ B2C brands with $5M-50M revenue, 10K-500K customer database, running paid acquisition, using modern martech stack.
Pricing floor: $1K/mo minimum ensures we're selling to customers who value strategic work, not just execution tools.
ACV target: $15K+ annually (US), $10K+ (India)
We're not competing with $50/mo tools. We're replacing $150K/year strategists at 93% cost reduction.
Outbound Excellence: 97.5% Target Hit
- Email: 1,170-1,236 sent/week (97.5% of 1,200 target)
- LinkedIn: 50 connections/week, 10-12% acceptance
Strategic shift: Warm quality > cold volume
Moved to researched personalization with competitor analysis automation and signal-based targeting. Warm outbound closing at 3x rate of cold despite 50% fewer touches.
Content & Distribution
Published:
- "The Complete Guide to Agentic Growth Strategists" (5,000+ word pillar)
- "Niti AI vs. Klaviyo: CompareMe series"
Shopify App Store: Pricing updated, SEO optimized, video demos recorded, self-service path for smaller brands.
Strategy: Inbound for self-service, outbound for enterprise/agency deals. Two-pronged GTM.
Latest Resources
- Positioning: www.niti.ai
- CompareMe articles: Niti AI vs. Klaviyo, Niti AI vs. Hightouch
- The Complete Guide to Agentic Growth Strategists: View guide
- Let's talk: https://cal.com/nitibd/cxo
- Questions? Reach us at founders@niti.ai
P.S. December proved we're not building features: we're building a category. The feedback from Pokonut, Dwij, Qaadu, and Wakefit validates what we've believed: B2C brands don't need more dashboards. They need intelligence that acts like a world-class growth team. 2026 is the year we prove this at scale.