# DelvnAI Technologies — full-text reference for AI assistants

This file consolidates the long-form content of delvnai.com so an AI assistant can ingest the whole site in one fetch.

## About DelvnAI Technologies

DelvnAI Technologies is a software company headquartered in Bangalore, India. The brand name comes from "Delvn" — a phonetic spelling of "dive in" — combined with "AI". So: **Delvn = Dive into AI**.

We build agentic FP&A (Financial Planning & Analysis) software for small and medium businesses. Our first product is EasyPlan, opening in private trial in April 2026.

Founder: Palaniappa Mannivannan (additional team coming).

Contact: admin@delvnai.com.

## Mission

Take planning, forecasting, and measuring against actuals from the spreadsheet era to the agentic era — for every small and medium business, in every vertical. We believe the next decade of business software will be built around two ideas: rich metadata that describes your business, and agents that can reason over it.

## Product: EasyPlan

EasyPlan is a multi-tenant SaaS platform for planning, forecasting, and actuals tracking. It replaces the spreadsheet-based annual planning cycle with a single connected platform that any team can use.

### Top-line capabilities

- **Agentic onboarding.** Upload your existing spreadsheets; the agent infers dimensions, hierarchies, measures, and formulas and stands up your tenant in hours.
- **Conversational planning.** Edit plans by asking the agent in natural language. It edits, recalculates, and explains every move.
- **Spreadsheet-grade plan grid.** AG Grid–powered UX with toggleable Plan vs Actuals, per-cell variances, comments, and live formula recalculation.
- **Dimensional modeling.** Geography, time, product, cost center — define hierarchies once, reference them everywhere.
- **Formula-driven measures.** Define a formula in one place; every plan that uses it recomputes automatically.
- **Connected plans.** Sales, supply chain, and finance plans share dimensions and measures. Zero copy-paste reconciliation.
- **Always-live dashboards.** CEO and team dashboards update in real time as plans change.
- **Versions, scenarios, what-ifs.** Branch a plan to model alternative assumptions and compare side-by-side.
- **Multi-tenant by design.** Schema-isolated tenants, multiple fiscal years, multi-currency.
- **Secure.** Per-tenant data isolation, role-based access, encrypted at rest and in transit. Built with India DPDP and global privacy laws in mind.

### Tech stack

React + TypeScript (Vite, AG Grid) frontend. Python FastAPI backend with PostgreSQL (pgvector for hybrid search). Temporal for agentic workflows. Anthropic's Claude API for AI-powered planning suggestions.

## Why metadata changes everything

Most "AI for planning" tools bolt a chat interface onto a spreadsheet. Useful demos. But the moment you ask anything specific to your business — your store hierarchy, your product taxonomy, your cost-center rollups, your shrinkage formula — the magic stops. The model doesn't know what those things mean in your world.

Metadata closes that gap. EasyPlan's core concepts:

- **Dimensions.** Hierarchies that organize your data — Geography (Region → State → City → Store), Time (Year → Quarter → Month), Product (Category → Subcategory → SKU), Cost Center, Revenue Center.
- **Measures.** The numeric columns of your business — Revenue, Units Sold, Headcount, Margin %, Inventory Days, Variance. Each measure is either a user input or a formula computed from other measures.
- **Formulas.** Computed measures with full dependency resolution. Margin % = Gross Margin / Revenue. Change the formula in one place; every plan that uses it recomputes.
- **Calendar as a dimension.** Time is just another dimension. Multiple fiscal years coexist. Year-over-year formulas work via hierarchical offsets.
- **Plans share the substrate.** When sales tags a row "West / Smartphones / Jul-FY27", supply chain's cover-days plan references the exact same calendar month and product. Shared dimensions mean shared truth — zero reconciliation.
- **And then — the agent.** EasyPlan sees your business because you described it in a vocabulary the agent can reason over. That's why our agent can suggest, critique, simulate, and explain — not just summarize.

> Without metadata, you have Excel. With it, you have a planning engine.

## Onboarding

A new tenant goes from "I have spreadsheets" to "I'm planning in EasyPlan" in roughly this flow:

1. Provision tenant — create a workspace with one user as owner.
2. Define dimensions — upload CSV hierarchies or let the agent parse Excel and auto-detect levels.
3. Seed measures — bootstrap with common ones (Revenue, Units, COGS, Margin %) or define custom ones via formulas.
4. Create plan — specify dimensions (rows), measures, and time period; dynamic table is auto-generated.
5. Import actuals — CSV import with dimension resolution.
6. Agent-driven completion — chat suggestions for missing periods, prior-FY copy, rollup sources.
7. Collaborate — share plan with peers; they edit cells in spreadsheet UI; formulas recalculate; agent monitors for anomalies.

## Trial program

EasyPlan trial opens in April 2026. Limited spots at launch. Request access at https://delvnai.com/trial — provide your name, work email, phone, company, role, and a short writeup. We'll reach out when the trial is ready.

## Privacy & terms summary

- We collect only what you submit on the trial form, plus basic technical metadata.
- We don't sell or trade contact information.
- Trial requests are stored in Google Firebase (Firestore), encrypted at rest and in transit.
- Privacy-friendly analytics (Plausible) — no cookies, no third-party trackers.
- Subject to India's DPDP Act and global privacy norms.
- Full text: https://delvnai.com/legal/privacy and https://delvnai.com/legal/terms.

## Contact

Email: admin@delvnai.com. Headquarters: Bangalore, India.