.Investor relations.
We own the layer everyone else rents
Eight years building Bangla speech, language and vision models in house. Four agentic products running on them, deployed across three industries and expanding into six.
Fundamentals
2018
Founded
Eight years of model development.
4
Products in market
Three live, one in early preview.
3
Models owned outright
Ekush LLM, SenseVoice, SenseSpeech.
2
Offices
Dhaka engineering, Singapore regional HQ.
.The thesis.
Three reasons this holds
- 01
The moat is ownership
Almost every AI company in our markets resells somebody else's API. We trained our own Bangla speech, language and vision models over eight years. Nobody can reprice us, deprecate the layer underneath us, or match the deployment guarantee we can offer a regulated buyer.
- 02
The wedge is language
Global platforms optimise for English-speaking developers and treat Bangla as a long tail. That leaves 170 million people, and the banks, telecoms and enterprises that serve them, structurally underserved by every well-funded competitor.
- 03
The expansion is horizontal
One model stack, four products, six industries. Each new vertical reuses the same agents and the same document pipeline, so the marginal cost of the next industry is go-to-market rather than R&D.
.Traction.
Measured in live deployments
Customer ARR lift across the verticals where UniSense is in production.
| Industry | Result | What the agents run |
|---|---|---|
| Real Estate | 60 → 130 | Sell twice the apartments with the same team. |
| Healthcare | 24/7 | Every patient answered. Every appointment kept. |
| Travel and Hospitality | 41+ | A concierge that speaks every guest's language. |
| Fintech | Global | Voice is the interface for the next billion in finance. |
| Telecom | Millions | National scale support without national scale cost. |
| BPO | 5x | Turn your BPO into an AI powered BPO. |
| Education Consultancy | +75% | Every student inquiry becomes a counseling session. |
Measured across live UniSense deployments. Treat them as directional. Figures for a given customer are confirmed during their pilot.
Market
Where the spend already is
$328B
Global BPO market, 2025
Grand View Research, 2025.
$696B
Projected BPO spend, 2033
Grand View Research, 2025. Roughly 10% CAGR.
$50B
AI agent market by 2030
Growing at roughly 46% per year.
170M
People online in Bangladesh
240M mobile wallets. No.1 freelance workforce.
.Portfolio.
Four products, one stack
Each addresses a different half of how a business runs. All four run on the same models.
UniSense AI
Live
Six AI agents that run your whole customer lifecycle
Finds buying signals in public conversations, opens the sale, onboards the customer, then handles renewals, support and collections. Every agent shares one view of the customer, and each one runs on its own.
Deck availableFinSense AI
Live
Conversational banking and document processing, in Bangla
Customers open accounts and move money by speaking, while handwritten Bangla forms turn into clean records with no template to maintain. Both surfaces run on a speech and language stack built in house over eight years.
Deck availableAIHubspot
Live
Every major AI model on one subscription, priced in taka
Send one prompt to GPT, Claude, Gemini and Llama at once and keep whichever answer holds up. Comes with 100+ pre briefed expert agents and a developer API, from BDT 700 a month.
Deck availableAltercrew
Early preview
A digital back office that asks before it acts
A Head Agent delegates payroll, invoicing, recruitment and reporting to specialist sub agents, then routes every outgoing action to the right person for approval. Nothing is sent without sign off.
Deck available
.The ask.
We are raising to take this across the region
The models are built and the products are in market. What capital buys now is go-to-market speed across six industries and four countries, before a global platform decides our languages are worth the training run.
Overview