Brevoir
← All coverage/Research report
BREVOIR ANALYSISApr 17, 2026

Shofo

shofo.ai/
YC W26AI data infrastructure, video dataset curationSeed, YC W26San Francisco, CA, United States
INVESTABILITY
48
PASS
CONFIDENCE 74%
VERDICT

We would watch this, but not underwrite it as a clean moat yet. The product is timely and the team has shipped before, but the business depends on content access, labeling economics, and whether labs truly want pre-curated video over in-house builds.

// Contrarian angle
What everyone sees: Shofo looks like a YC-backed Common Crawl for video, built for multimodal AI teams that need custom datasets fast.
What we flagged: The real moat is not the index, it is whether platforms keep Shofo legally and commercially connected to the underlying video sources.
SCORE BREAKDOWN
Team
16/25

The founders have relevant UCSB technical depth, and Misra has already shipped and scaled Correkt to 40k plus users.

Market
15/25

Multimodal training data is a real bottleneck, but Shofo addresses only a narrow slice of the broader annotation market.

Traction
6/25

Public traction is thin, with no revenue, customers, or usage disclosed beyond the prior Correkt project.

Timing + Moat
11/25

The timing is good for video data demand, but the moat is constrained by platform access and labeling economics.

COMPANY
Founded
2025
Total raised
Not publicly disclosed
Key investors
Y Combinator

Shofo indexes billions of videos and helps AI labs extract custom training datasets on demand.

PRODUCT + TECHNOLOGY
Shofo combines a continuously updated index of billions of videos with an agentic search and labeling pipeline. The company says it can find exact subsets, apply reasoning annotations, and deliver custom datasets in days rather than months. It also exposes an API for querying 100M plus videos across social platforms, which makes this more than a pure annotation shop.
MARKET + TIMING
The company is riding a real pull from multimodal model training, where teams need video, image, audio, and text data with stronger provenance and faster turnaround. The broader data annotation market is large and growing, but Shofo’s practical TAM is narrower, focused on AI labs and research teams that need pre-curated video datasets. That wedge is credible, but it is still a subsegment of a crowded and mature market with strong incumbents.
TEAM
The team is young, UCSB-rooted, and technically oriented, with the strongest signal coming from Alexzendor Misra, who previously built Correkt to 40k plus users before the pivot. Bryan Hong and Andre Braga bring data science and NLP background, while Braiden Dishman adds an AWS line on the résumé, though public detail is thin. We think the team has proven it can ship and iterate, but we do not yet see evidence of experienced enterprise GTM or scaling operations.
Bryan HongCEO

He is described as Co-Founder and Head of AI at Shofo, with a UCSB BS in Statistics and Data Science and a prior MIT affiliation that is not clearly detailed. Public track record outside Shofo and Correkt is limited.

Andre BragaHead of AI

He is a UCSB statistics student and researcher in the UCSB NLP group under Professor Xifeng Yan. The research background fits Shofo’s multimodal search and annotation thesis.

Alexzendor MisraCTO

He is a UCSB computer science dropout and previously served as CEO of Correkt, which reached 43k users. This is the strongest execution signal on the team, since he has already shipped a consumer AI product and pivoted it.

Braiden DishmanCOO

He is a UCSB economics graduate and previously worked at AWS, though the title and tenure are unclear. Public evidence for his operating contribution is limited.

TRACTION SIGNALS
Shofo is very early, with no disclosed revenue, customers, MRR, or ARR. The main traction signal is indirect, the founders’ prior project Correkt reportedly reached 40k plus users, and Shofo has already surfaced in YC W26 coverage and launched publicly in early 2026. That said, we found no public proof of meaningful paid usage, marquee logos, or hiring momentum yet.
BUSINESS MODEL
Public pricing is not disclosed. The visible model appears to blend custom dataset delivery, API access to large video and social indexes, and likely project-based enterprise work for AI labs. Margins could be attractive on the API side, but the human-in-the-loop labeling layer may compress economics if scale rises before automation does.
COMPETITIVE LANDSCAPE
Shofo sits between managed annotation vendors and tooling platforms. Direct competition for its exact index-first, agentic curation layer appears limited, but it faces strong substitutes from Scale AI, Appen, iMerit, TELUS, and in-house dataset teams. The core competitive question is whether buyers value pre-aggregated video discovery enough to switch away from established labeling workflows.
Scale AI
Scale AI is a mature training data and workflow platform, but it usually expects customers to bring raw data. That makes it an adjacent substitute rather than a direct index-first competitor.
medium threat
Appen
Appen is a large-scale managed annotation provider with deep labor and enterprise relationships. It is slower and more labor-heavy, but it can still win on breadth and trust.
high threat
iMerit
iMerit focuses on complex, domain-specific annotation with strong compliance posture. It is less index-centric than Shofo, but still competes for the same training-data budgets.
medium threat
TELUS International
TELUS offers broad data annotation services at large scale, but it is a generalist player rather than a specialized video discovery engine.
low threat
Labelbox
Labelbox is a tooling platform for teams that want to manage labeling in-house. It is more complementary than directly competitive.
low threat
MOAT + DEFENSIBILITY
Shofo’s best moat candidate is data, specifically a large and continuously refreshed video index combined with semantic search and labeling workflows. We do not see strong network effects yet, and switching costs look low unless the company deeply embeds into customer pipelines. The moat is therefore real but fragile, because it depends on content access, index quality, and product depth more than on brand alone.
Datamoderate

The company claims to maintain a searchable index of billions of videos and to continuously update it across sources.

Scale economicsemerging

Once the index exists, marginal access to additional queries should be cheaper than repeatedly sourcing and labeling fresh footage from scratch.

Switching costsweak

No public evidence yet shows deep workflow lock-in, long contracts, or embedded customer integrations.

RISK ASSESSMENT
Content access and licensing
critical0-6mo

The business relies on access to public and private platform video, but the research notes no disclosed partnerships or licensing rights.

Labeling bottleneck
high0-6mo

Human-in-the-loop labeling could become the throughput constraint if customer demand rises faster than automation.

Slow buyer adoption
medium6-18mo

AI labs may still prefer to build datasets internally or use incumbents they already trust.

Weak public traction
medium0-6mo

We found no public revenue, customer logos, or meaningful usage metrics for Shofo itself.

STRENGTHS
  • +The founders already proved they can build and ship a prior AI product.
  • +The product addresses a real bottleneck in multimodal training data.
  • +Shofo’s index-first positioning is different from conventional annotation tooling.
  • +YC backing gives the company early distribution and credibility.
WEAKNESSES
  • −Public traction is extremely limited, with no disclosed customers or revenue.
  • −The business depends on video source access that is not publicly secured.
  • −Human-in-the-loop labeling may cap margins and throughput.
  • −The team lacks visible enterprise sales depth or operating scale experience.
SOURCES
Sources cited above. Not investment advice.
// For founders

Want Brevoir to cover your startup next?

Submit to Brevoir Discover. We publish a page, investors tracking your sector find you. Five minutes.

Submit your startup
Brevoir Coverage: Shofo | Brevoir Terminal