V7, Kognic, Scale AI — operator notes from a former buyer
V7 Labs — operator notes
What V7 does well
- UI is the best in class for computer vision annotation. Bounding box, segmentation, keypoint workflows are fluid.
- Auto-annotation is genuinely useful. SAM-style integration for segmentation cuts annotation time materially.[^1]
- Document AI workflows are strong. V7 Go handles complex PDF + form workflows.[^2]
- Customer success is responsive for mid-market accounts.
Where V7 breaks for MENA workloads
- Arabic NLP is not a design center. RTL handling works at the locale level, not the design level. Subtle annotation drift on Arabic text spans.
- No KSA / UAE / Egypt cloud regions. Sovereign deployment is custom-engagement, not standard.
- No PDPL operational alignment. Article 24 72-hour breach notification is a custom incident-response negotiation.
- Workforce is global gig pool by default. No curated Cairo / Riyadh PhD-linguist option.
- Pricing is mid-market enterprise. Annual minimums + USD-only invoicing make pre-Series-A MENA founders pay a premium for capability they don’t fully use.
When V7 is right for you
- US / EU computer vision workload with no MENA component
- Strong existing AWS / Azure US footprint
- No Arabic NLP component
- Budget for mid-market enterprise pricing
See the V7 alternative page for the full comparison.
Kognic — operator notes
What Kognic does well
- Best-in-class for sensor-fusion AV annotation. Multi-camera + LiDAR + radar workflows are deeply engineered.[^3]
- AV-specific tooling is unmatched. Scenario library, edge-case mining, corner-case workflows are AV-native.
- Quality management for AV is rigorous. Inter-annotator agreement + adjudication workflows tuned for AV.[^4]
Where Kognic breaks for MENA workloads
- Not designed for Arabic anything. Kognic is AV-specialised by design — Arabic NLP isn’t its target.
- Heavy procurement weight. Enterprise AV contracts run 6-12 months from first contact to first label. KSA / UAE startup AV teams need faster ramps.
- Workforce is European-centred. No Cairo / Riyadh option.
- Pricing is enterprise AV. Mid-six-figure annual commitments are standard.
When Kognic is right for you
- Large AV program with $1M+/year annotation budget
- European OEM with European compliance needs (TISAX)[^5]
- Sensor-fusion-heavy workload
- No MENA market component
See the Kognic alternative page for the full comparison.
Scale AI — operator notes
What Scale AI does well
- Largest workforce + capacity in the industry. Throughput at billion-token scale for foundation-model labs.
- Strong defense + government track record. US government clearance options.[^6]
- Strong foundation-model lab track record. OpenAI, Anthropic, Meta worked with Scale.[^7]
- API + tooling is mature. Programmatic workflows are first-class.
Where Scale AI breaks for MENA workloads
- Arabic is not a strategic priority. Coverage exists but is not differentiated.
- No KSA / UAE / Egypt cloud regions. Sovereign deployment is custom-tier engagement.
- No MISA / GOSI / ZATCA procurement story. KSA buyers requiring in-Kingdom procurement face friction.
- Workforce is global gig pool. No curated dialect-stratified Arabic option.
- Pricing is enterprise / hyperscaler. Annual minimums in the high six figures are common for serious workloads.
- Customer success is enterprise-paced. Pre-Series-A MENA founder gets routed to self-serve.
When Scale AI is right for you
- US foundation-model lab with $10M+/year annotation budget
- US defense / government with classified workloads
- Multi-billion-token English pretraining corpus curation
- No MENA market component
See the Scale AI alternative page for the full comparison.
The structural pattern
Looking across all three:
| Strength shared | Weakness shared (for MENA) |
|---|---|
| US / EU customer track record | No KSA / UAE / Egypt cloud regions |
| Mid-to-large enterprise procurement maturity | Annual minimums punish pre-Series-A MENA founders |
| English-default product surface | Arabic as locale, not design center |
| Global gig pool workforce | No curated PhD-linguist Cairo / Riyadh tier |
| Mature compliance for US / EU regimes | No PDPL operational, sometimes no GDPR baseline |
| English customer success | No native Arabic customer success + MENA market context |
None of this means V7, Kognic, or Scale are bad vendors. They are the right vendors for non-MENA, non-Arabic, mid-enterprise-plus workloads. They are the wrong vendors for MENA-first, Arabic-first, founder-stage workloads.
Why we are building Annota8
Noureldin and I were paying customers across multiple roles. We saw the gap from the inside. The decision to build Annota8 is not about doing what V7 / Kognic / Scale do, only cheaper. It is about closing the five structural mismatches every MENA AI team hits. The design targets:
- Pricing structure — line-itemed transparent, no annual minimums for pre-Series-A
- Workforce — curated Cairo + Riyadh + Dubai, with a PhD-linguist QA layer
- Sovereignty patterns — KSA / UAE / Egypt deployment options + customer-cloud sovereign tenancy
- Procurement readiness — MISA-registered KSA entity, SAR / EGP / AED invoicing, ZATCA-aware VAT handling
- Cultural — Vision 2030 + HUMAIN context, bilingual operations, MENA-resident leadership
That’s the design centre. The product is being built around it; the exact delivery shape — sovereign tenancy, dedicated workforce, BAA paths, certifications — is scoped per customer engagement, not baked in as a blanket claim.
What this means for you
If you are a MENA AI team — foundation-model lab, enterprise, or AI-native startup — and you are currently using or evaluating V7 / Kognic / Scale, the question to ask is structural, not feature-by-feature:
- Does the vendor’s design center match your design center?
- Does the vendor’s compliance posture match your regulatory exposure?
- Does the vendor’s procurement model fit your buyer side?
- Does the vendor’s workforce produce the dialect coverage you need?
- Does the vendor’s customer success understand your market?
If the answers are no, an Arabic-first alternative is structurally a better fit. That’s true whether you choose Annota8 or another regional vendor.