Engagement Models & Pricing
How to work with us — end-to-end managed delivery, a dedicated crowd, on-demand annotation and QA, dataset licensing and sovereign delivery, each with clear scope, deliverables and process. Final pricing is scoped per project.
How clients work with us
Five ways to engage — pick the one that fits your stage
Whether you need Cognegica to run the whole pipeline, a ring-fenced native-speaker crowd stood up for your languages, flexible per-volume annotation and QA, a proprietary dataset to license, or a sovereign delivery posture — there's a clear model. Every one is rights-cleared, documented and available with an India-residency option.
-
End-to-end managed delivery
We run the full pipeline — recruitment, collection/transcription/annotation, multi-layer QA and delivery — against your spec and our SOPs.
See the process -
Dedicated crowd / dedicated team
A ring-fenced native-speaker crowd and linguistic reviewers stood up for your languages and domain over time.
How it works -
On-demand annotation / QA
Flexible per-volume transcription, annotation, diarization, validation and linguistic QA — including evaluation of human and AI outputs.
Scope a batch -
Proprietary dataset / data licensing
License a rights-cleared dataset from the catalog, or have one built and licensed under terms that match your rights needs.
Browse the catalog -
Sovereign delivery
Residency, consent and provenance built in — collection, processing and storage kept in-country for regulated and government programs.
Sovereign data
Core delivery models
Three ways we deliver — scope, deliverables and process
These are the primary shapes an engagement takes. The card lists the scope and the deliverables; the process for each is documented below. All three run on the same foundations: native-speaker sourcing, structured SOP workflows and multi-layer QA.
End-to-end managed delivery
We run the whole pipeline to your spec
Scoped per project
Scoped per project by languages, modalities & volume
- Scope: full pipeline — recruitment → collection / transcription / annotation → multi-layer QA → delivery
- Deliverables: versioned data drops, a data card, QA reports and documented provenance per batch
- Process: project spec and SOPs agreed up front; native-linguist review at every gate
- You define the spec; we own staffing, throughput and the quality bar
- India-residency / sovereign delivery option
Dedicated crowd / dedicated team
A ring-fenced crowd for your languages, over time
Dedicated capacity
Recurring capacity & roadmap — scoped per program
- Scope: a stood-up native-speaker crowd plus linguistic reviewers for your languages and domain
- Deliverables: sustained throughput, recurring drops, refreshes and roadmap alignment
- Process: calibrated and ramped against your guidelines; the same team carries domain context
- Best when needs are ongoing and domain knowledge compounds over months
- Sovereign / India-residency option across the program
On-demand annotation / QA
Flexible per-volume work, no long-term commitment
Per-volume / per-batch
Priced by modality, complexity & quality bar — request a quote
- Scope: transcription, annotation, diarization, validation and linguistic QA on data you already have
- Deliverables: labelled or validated data with per-batch QA metrics and adjudication notes
- Process: guideline calibration, two-pass labelling, native-linguist QA, adjudication
- Includes evaluation of both human and AI outputs (transcription QA, audio validation, eval)
- Scale up or pause as your volume changes
Every engagement is scoped per project against languages and their rarity, modalities, volume, your target quality bar, rights and residency.
Managed delivery — the process
How end-to-end managed delivery runs
The full pipeline for managed delivery, against your project spec and our SOPs. The dedicated-team model runs the same pipeline as a standing capability; on-demand work picks up from the annotation step.
-
1
Recruitment & calibration
Native speakers and linguistic reviewers sourced from the distributed network against a defined demographic and domain matrix, then calibrated on your guidelines.
Guideline calibration signed off
-
2
Collection / transcription / annotation
Field-grade collection, transcription against the seven transcription standards, or annotation/diarization — run inside structured SOP workflows.
SOP adherence checked per task
-
3
Multi-layer QA
Two-pass review with native-linguist QA and adjudication of disagreements, against the agreed quality bar.
Quality bar met per batch
-
4
Delivery & provenance
Versioned drops with a data card, QA report and documented consent and provenance per record.
Provenance recorded per record
At a glance
Delivery models compared
Scope, deliverables and process for the three core models, so you can match the shape of the engagement to the shape of your need.
| Model | Scope | Deliverables | Process | Best for |
|---|---|---|---|---|
| End-to-end managed delivery | Full pipeline: recruitment, collection/transcription/annotation, QA, delivery | Versioned drops, data card, QA reports, provenance | Spec + SOPs agreed up front; native-linguist multi-layer QA at every gate | A defined deliverable you want run for you |
| Dedicated crowd / team | Ring-fenced native-speaker crowd + reviewers for your languages/domain | Sustained throughput, recurring drops, refreshes, roadmap | Calibrated and ramped on your guidelines; team carries domain context | Ongoing needs where domain knowledge compounds |
| On-demand annotation / QA | Per-volume transcription, annotation, diarization, validation, linguistic QA | Labelled/validated data with per-batch QA metrics and adjudication notes | Calibrate, two-pass label, native-linguist QA, adjudicate | Flexible volume on data you already have |
Indicative scope. Final scope, deliverables and quality bar are documented in the SOW per project.
Data products & sovereign delivery
License a dataset, or deliver it sovereign
Beyond delivered work, you can license a proprietary dataset from the catalog under terms that match your rights needs, or wrap any engagement in a sovereign delivery posture.
Proprietary dataset / data licensing
License rights-cleared data, or build-and-license
Contact us
Commercial non-exclusive and enterprise/sovereign exclusive tiers — scoped per dataset
- Scope: license an existing proprietary dataset, or commission one and license it
- Deliverables: a documented data card — provenance, methodology, quality metrics
- Process: tier chosen to your rights — commercial non-exclusive or exclusive transfer
- Quarterly refresh and new-data additions where offered
- Full provenance transfer for procurement and audit on exclusive tiers
Sovereign delivery
Residency, consent and provenance, built in
Available on any model
Adds residency, consent and provenance controls — scoped per project
- Scope: collection, processing and storage kept in-country, no cross-border transfer of sensitive data
- Deliverables: documented consent per contributor and provenance tracked per record
- Process: residency and chain-of-custody controls layered onto any delivery model
- Aligned with the DPDP Act, 2023 for government, BFSI and regulated buyers
- Evidence you can put in front of a regulator
Licensing is scoped per dataset. Sovereign delivery is an option on any model, scoped per project against residency, consent and provenance requirements.
How we price
What moves the number
Pricing is transparent about its drivers even when the final figure is scoped per project.
-
Languages & rarity
Low-resource and field-collected languages cost more than high-resource ones.
-
Modality & complexity
Speech, multimodal and 3D/sensor annotation carry different effort than plain text.
-
Volume & quality bar
Target quality bar, QA passes and adjudication depth scale the work.
-
Rights & residency
Exclusivity, IP transfer and India-residency change the licence and cost.
Trust & sovereignty
Every engagement is rights-cleared and documented
Whatever the engagement model, the foundations are the same: written, commercial-reuse consent on every record; documented provenance and license terms; published quality metrics; and an India-residency option for government, BFSI and regulated buyers. You get data you can defend in a procurement, security and compliance review.
FAQ
Engagement & pricing — common questions
- Which delivery model should I choose?
If you have a defined deliverable you want run for you, choose end-to-end managed delivery — we own staffing, throughput and the quality bar. If your needs are ongoing and domain knowledge compounds, a dedicated crowd / dedicated team stands up native speakers and reviewers for your languages over time. If you already have data and want flexible, per-volume work, on-demand annotation / QA covers transcription, annotation, diarization, validation and linguistic QA — including evaluation of human and AI outputs. All three run on the same multi-layer QA and native-linguist review.
- How is pricing determined?
Every engagement is scoped per project against languages and their rarity, modality and complexity, volume, your target quality bar (IAA/QA), and rights/residency. Tell us your requirements and a senior PM returns a scoped quote — usually within one business day.
- Are there minimums?
Project engagements have practical minimums so we can staff a calibrated team and hit the quality bar. We'll tell you the smallest viable scope for your goal.
- What about rights and exclusivity?
Dataset licensing runs from non-commercial research, to commercial non-exclusive (others may license the same set), to enterprise/sovereign exclusive (you own the rights, with full provenance transfer). For custom collection you can take work-for-hire ownership, or retain IP and license it back to us. We make the rights explicit in the SOW.
- Can you guarantee data residency?
Yes. We offer sovereign delivery: collection, annotation, processing and storage performed entirely within India, aligned with the DPDP Act, 2023. See Sovereign Data for the full residency and compliance posture.
- How do we get started?
Tell us your languages, modalities, volume, quality bar and timeline via request a quote. A senior PM scopes the work and replies within one business day; we typically move from NDA to a signed SOW in about 14 days.
Tell us what you need — we'll scope it and quote.
Share your languages, modalities, volume and quality bar. You'll get a scoped plan and indicative pricing, or browse the catalog to license data today.