Occlum
A control layer so companies can use AI safely
AI · CybersecurityB2BAndijan
Every day, employees paste sensitive data into ChatGPT and other AI tools: client records, contracts, source code. Occlum lives on company devices and runs each AI-bound request through its detection intelligence. Safe requests pass untouched, sensitive ones are occluded so the request still works, and dangerous ones are blocked by the company's own policy. Employees keep their AI tools; the company gets control it can prove to auditors and regulators.
Traction
$40,000
Pre-seed raised, Sep 2026
$100,000
Soft commitments, 6 firms
1st
Batch ranking
- Pre-seed from United Ventures after due diligence; one of 30 in its first Startup Booster
- One of 5 AI startups representing Uzbekistan at Tashkent International Investment Forum 2026
- Reached the final stage of Draper University Hero training, Sep 2026
The problem
Employees paste client records, contracts, credentials and source code into AI chatbots daily, often from personal accounts. Banning AI drives it underground, and no security team can say what left yesterday.
- Employees paste client data, contracts and source code into AI tools daily
- Often from personal accounts the company cannot see
- Banning AI fails: usage goes underground
- Nobody can say what left the company through AI yesterday
The solution
Sits on the device, so it sees traffic that gateways never route. Context-aware detection masks data while keeping the request's meaning. Cloud or fully on-premise, with no customer infrastructure needed.
- On the device - Sees what employees send to any AI tool
- Occlusion - Masks sensitive data while the request keeps its meaning
- Policy - Blocks dangerous requests under the company's own rules
- Audit - Control the company can prove to regulators
The company
What makes it different
On the device, where unrouted AI traffic happens
Gateways and SaaS integrations only govern traffic routed through them. Occlum sees what employees actually send, rewrites requests by context rather than pattern, and runs in the cloud or fully on-premise.
Who it's for
- Buyers accountable for company data: CISO, head of IT, compliance officer or CEO
- Regulated companies: banks, fintech, telecom, insurance, healthcare
- Germany first, where the EU AI Act requires mid-sized firms to govern AI; design partners in Central Asia
How it makes money
- Occlum Sensor: a free 7-day scan that shows a company the real scale of its AI use
- Occlum Enclave (cloud): from $10 per seat a month; small teams from $3,600 a year
- Occlum On-Prem: annual licence from $15k plus implementation, for data that must stay in the network
Market
~$12B
Global TAM, 1.27M mid-sized firms
~$864M
Germany SAM, 84,000+ companies
$3M ARR
Early-years target, Germany only
Regulation is creating the market. The EU AI Act's core obligations are coming into force, and Gartner forecasts that by 2027 AI governance will be a requirement of sovereign AI laws worldwide. Companies that adopted AI will have to prove control over it.
Competitors
- Enterprise platforms - Cyberhaven, Zscaler, Netskope: serious technology, priced for large corporations with long sales cycles.
- AI-native DLP - Nightfall, Harmonic, Polygraf: SaaS or API integrations that only see the approved stack.
Founders
Started Occlum alone in late January 2026 and designed and built the product personally. Abdulloh has worked with AI for 4 years and has built and sold web products for local businesses for 3 years.
Sold a first software product, a Telegram bot, to around 30 customers while still at school. In 2025, when paper-based IELTS exams ended in Uzbekistan, built a computer-delivered prep platform with 2,000+ active users.
Admitted to five foreign universities, three of them in the world top 200, and offered a $170,000 scholarship in the US. Turned all of it down to stay in Uzbekistan and build Occlum.
Background
- Presidential specialised school: perfect-score graduate, national honours diploma
- Admitted to 5 foreign universities (3 in the world top 200); $170,000 US scholarship offer
- One of 5 AI startups representing Uzbekistan at Tashkent International Investment Forum 2026
- 5 district and 1 regional wins in Russian-language national olympiads
- Muay Thai champion of Uzbekistan, lightweight, 2019
- Founded the school's first coding club and taught 65+ students
Ask any security team “what left your company through AI yesterday?” and nobody can answer. That's the gap Occlum closes.
Milestones
- 3 Sep15-day sprint: $40,000 pre-seed raised from United Ventures; $100,000 in soft commitments from 6 businesses
- 9 SepFinal stage of Draper University Hero training; market research with 30+ AI agents; pilots targeted with 8 banks
- 10 SepPre-seed contract with United Ventures signed
The round
$100,000
raising
Use of funds
German market entry and first customers
40%
Product and detection intelligence engine
35%
Team building
25%
