Enterprise
Private research infrastructure.Custom-built per client.
A standing synthetic panel of your market, built from your customers and calibrated to your own past studies, so the numbers line up with the tracker you already trust. Your team gets its own workspace, API and MCP access, and a deployment that runs on Lewis, our own survey model, with no external model API calls. A dedicated, physically separate instance can be included.
Pilot: 8 weeks on your questions, calibrated to your past studies. It rolls into an annual engagement. Pricing is scoped to your audiences and study volume on a short call.
Every Enterprise deployment includes a self-serve workspace built for your team.
Research agencies: simulated extra rows on a tracker you already ran, matched to your cells, holdout you score. Sample boosting.
Why teams deploy privately
What a private deployment includes.
Uncapped Studies
Screen concepts and messages on a calibrated panel before you field a live focus group.
The contract sets the study volume. Panel studies, message tests, and focus groups run on the calibrated panel.
Dedicated Inference
A dedicated instance can be included.
A physically separate GPU instance can be part of the written plan. Hardware is scoped per client. Lewsearch does not guarantee a named GPU SKU to every customer. On-premises deployment is a separate contractual question, not something inferred from hosted dedicated hardware.
Data Isolation
No external model API calls.
Enterprise study inference runs on Lewis. There are no external model API calls, including no third-party image-review models. Hosting, logging, and data-handling terms are written per client. Zero external model calls is not the same as a promise of zero data egress.
Economics
Traditional panels vs. private infrastructure.
Traditional focus groups
Recruit and field
Weeks to recruit a room
- Per-project vendor procurement
- Limited geography and sample size
- No iteration within the same budget cycle
Lewsearch Enterprise
Annual engagement
Standing calibrated panel · no external model APIs · scoped to your market
- ✓Study volume set by the contract
- ✓Study volume is set by the contract
- ✓One annual engagement covers the study volume in the contract
Deployment scope
What's included.
Each contract is scoped. A dedicated instance can be included when the plan calls for it. Your team gets the Lewsearch research stack written into that order: studies, focus groups, message testing, and API access, with no external model API calls on those workloads.
- ✓Dedicated physically separate instance, when the written plan includes it
- ✓All inference on Lewis: no OpenAI, Anthropic, or external model API calls
- ✓Uncapped studies, focus groups, and questions under contract
- ✓Sample boosting: simulated extra rows on an instrument you already fielded. You score the holdout.
- ✓Custom audiences from your customer data, with persistent memory and side-by-side scenarios
- ✓A self-serve workspace built for your team: your logo, your audiences, your question library
- ✓Multi-seat access with admin roles and permissions
- ✓Audit log of every study, export, and API call
- ✓Individual API keys per seat: issue, rotate, and revoke
- ✓White-label PDFs and workspace branding for research firms and agencies
- ✓MCP access so your AI assistants and internal agents can field questions to the panel. The Lewsearch team configures it for your workspace.
- ✓Onboarding and deployment with the team
- ✓Dedicated account manager
Architecture
Built for compliance review.
Zero external inference
All respondent simulation runs on Lewis, our proprietary fine-tuned model. No prompts sent to OpenAI, Anthropic, Google, or any third-party model API, including image-review helpers used on self-serve. That is a model-call guarantee, not a claim that no data ever leaves the account systems named in the contract.
Dedicated instance
Enterprise is a custom build per client. There are no external model API calls on Enterprise workloads. A dedicated physically separate instance can be included. When the plan includes that instance, the hardware is separate from the shared queue. This is hosted separation unless the written agreement says otherwise. It is not automatically on-premises.
Custom MSAs, security questionnaires, and vendor onboarding are part of the engagement.
Programmatic access
A research API for your team.
Every deployment exposes the same calibrated panel over a REST API POST /api/v1/cohort/query with per-seat keys. Enterprise workloads make no external model API calls. MCP access is part of every Enterprise deployment, so your AI assistants and internal agents can field questions to the panel directly. The Lewsearch team configures it for your workspace. Your data, pricing, and product teams can query and re-query at the rate your contract sets.
- ✓One endpoint, multiple-choice or free-form, JSON in and out
- ✓Rate limit and monthly respondent pool sized to your contract
- ✓Price-sensitivity and message testing at scale
- ✓Per-seat keys and an audit trail
Team workspace
Your brand. Your users. Full visibility.
Enterprise deployments include a private workspace built for your team: your audiences, your question library, and your brand.
Branded login
Your logo on a dedicated sign-in page at your company URL. Researchers enter your workspace, not a shared Lewsearch account.
Team seats
Invite analysts, strategists, and client stakeholders. Role-based access controls who can run studies, view results, or manage the workspace.
Admin & audit log
Workspace admins see who ran what, when, and what was exported. Full activity history, exportable for compliance and internal review.
API keys per seat
Each team member gets their own key for programmatic access. Issue, rotate, or revoke individually, with no shared credentials to leak.
Who deploys
Built for organizations with data handling requirements.
Market research firms
Large research firms
Pharma & life sciences
Behavioral studies, concept testing, patient insight
Financial services
Research departments, product insight, compliance-sensitive work
Government contractors
Custom build per client. No external model API calls on Enterprise workloads. A dedicated physically separate instance can be included.
Regulated enterprises
Any org where data isolation is a procurement requirement
How we score the panel
On the calibrated 50-state benchmark, the average miss is 7.07 points on the best 80% of 1,521 scored questions, and 10.02 points across all of them. 50-state check →
Next step
Start with a discovery call.
We scope your audiences, study volume, and deployment on a short call. Accounts are set up by the Lewsearch team.
Or email jeffrey@lewsearch.com · How teams engage