Ask in plain English.
The answers exist. Your team just can't find them.
We've all searched emails by keywords and found nothing, or looked for an old PDF in a shared drive with no matching file names.
Search is a chatbot-style query tool that indexes your files by context. Ask in plain English. Get the answer with the source cited. Cloud or on-premise with open-source models.
Fixed fee + monthly · 4-6 weeks setup
The pain you know
The search box doesn't understand what your team is looking for.
Your team knows the document exists. They still can't find it. Email search returns seventy-three irrelevant results, SharePoint finds three with mismatched metadata, and the drive search returns nothing useful. Veterans lose an hour a day; new employees lose more. The problem isn't the search box. It's that the search box doesn't understand what your team is actually looking for.
73 results · the words match, the meaning doesn't
RE: RE: cancel my Thursday call.msg
Email · 'cancel' matched a calendar note
Cancellation_Policy_v2.pdf
Drive · the policy, not what you did
Proposal_template_v4.docx
Drive · 'project' matched, no context
Q2-project-tracker.xlsx
Drive · keyword in a cell
old_contracts/
Folder · 80 files inside
+ 68 more · ~12 minutes of scanning
Keyword search matches words. Search understands what you mean, and finds the precedent.
What it does
A chatbot interface on top of your documents.
Search puts a chatbot interface on top of your documents. Your team asks a question in plain English. The system reads the relevant passages, drafts the answer, and shows the exact source document with the relevant section highlighted. Click the citation to verify. The interface looks like ChatGPT. It works on your actual files.
What it looks like in use
Legal associate
Switch roles to see how each team asks. Every answer cites its source.
What gets indexed
- Email (Outlook, Gmail)
- File shares (SharePoint, Google Drive, OneDrive, Dropbox, Box)
- Document management (iManage, NetDocuments, Procore)
- CRMs (Salesforce, HubSpot, Microsoft Dynamics)
- Support tickets and customer communications
- PDFs you upload directly, plus PDFs in any of the above
- Your existing intranet or wiki, if you have one
Search reads from where the documents actually live. Nothing migrates into a new platform. Your access controls stay in place; if a user can't see a document today, Search won't surface it to them tomorrow.
How it answers
Every answer cites its sources. The interface shows two things: the drafted answer in plain language, and the underlying passages from the source documents that produced it. Click any citation to open the original document, with the relevant section highlighted.
When the system isn't confident, it says so. A low-confidence answer surfaces with a flag and the closest matches it could find. Your team verifies and gives feedback, which the system uses to improve.
What Search won't do: pretend to know things that aren't in your documents. It doesn't fill gaps with general knowledge. If your team asks a question and the answer isn't in your files, Search says so and tells you where it looked.
“…employees at the Director level and above are permitted to carry forward a maximum of 80 hours of accrued paid time off…”
Grounded answers cite the exact passage. Click through to the source.
Deployment options
This is where Search is most differentiated. Three deployment shapes, picked by your data sensitivity, not by us.
Cloud
Standard deployment
Documents indexed in our managed cloud. Queries processed using Claude or GPT. Fast, low-friction, suitable for documents that aren't subject to privacy or confidentiality concerns beyond standard business protection.
Your cloud (VPC)
Inside your AWS, Azure, or GCP
Search runs inside your cloud account. Documents indexed in your environment. Model calls go through providers you've contracted with directly. Nothing about the workflow leaves your boundary unless you've explicitly approved it.
On-premise or air-gapped
Open-source models, your hardware
Search runs on your servers. Indexing happens locally. Queries use open-source models (Llama, Mistral, Qwen, DeepSeek) running on your hardware. Nothing leaves your network. The deployment shape for legal client data, healthcare PHI, defense-adjacent IP, or anything where data sovereignty isn't optional.
Most regulated buyers (legal, healthcare, PE, manufacturing with sensitive IP) end up at option two or three. We don't push you toward the cloud for our convenience. The on-premises shape costs us more to deliver, but reduces your compliance risk. The economics work for both sides.
How it works
Four phases. Same model every Databender engagement runs on.
Audit
2 weeks
Senior diagnostic. We map your document sources, validate access patterns, confirm the deployment shape, and rank the workflows where Search will land hardest first. The Audit report tells you whether Search fits, what it'll cost, and what it'll save.
Implementation
4-6 weeks
Source integrations, indexing, deployment (cloud, your VPC, or on-prem), access controls aligned with your existing permissions, retrieval, and answer quality tuned for your specific document types. Your power users start using it during week three, in beta. Full team rollout at week six.
Adopt
1-2 weeks·Included
The chatbot interface is familiar. The prompting habits aren't. We train your team on how to ask Search well, what kinds of questions return reliable answers, and when to verify versus when to trust. Role-specific playbooks (associates use Search differently than partners; coordinators use it differently than billers).
Run
Ongoing·Monthly run fee
Continuous indexing of new documents. Ongoing tuning as your team's question patterns evolve. Monitoring for hallucinations, low-confidence patterns, and quality drift. Same-day fixes for production issues. The system you start with is not the system you have a year later; we tune it as your business changes.
What you own
When we're done, you have everything.
- The index and all retrieval code
- The prompts and evaluation suite tuned for your data
- Integration code for every source we connected
- Deployment runbooks for your environment
- Access control mappings
- Monitoring dashboards
- Adoption playbooks per role
- The training materials your team uses
You have full ownership. If you bring this in-house in three years, you have the runbooks to do it.
Pricing
Three tiers. Cloud, VPC, or on-prem.
Tightly scoped. Add-ons priced transparently for the cases where customers genuinely want more. The Audit determines which tier fits.
Search Starter
$15,000 cloud · 3-4 weeks
Best for: A single practice group, clinical site, or project team that wants chatbot-style search on a small document corpus without a full enterprise rollout.
Includes. 2 source systems, under 500K pages indexed, 1 reranking model, page-level citation grounding, basic ACL passthrough, a 50-question eval suite tuned to your document types, 1 week of Adopt with 2 training sessions, and cloud deployment.
Search Standard
$25,000 cloud · $30,000 VPC · $40,000 on-prem · 4-6 weeks
Best for: The typical Search engagement. Most mid-market deployments land here.
Includes. Everything in Starter, plus: 5 source systems total, 500K to 2M pages indexed, 1 cloud reranker plus 1 open-source fallback, passage-level citation grounding, role-based multi-tenant access control, a 150-question eval suite, and 2 weeks of Adopt with 5 training sessions. Available cloud, in your VPC, or on-prem.
Search Enterprise
$48,000 cloud · $60,000 VPC · $80,000 on-prem · 6-10 weeks
Best for: Regulated firms and multi-site groups that need privilege-aware retrieval, redaction handling, and a regulatory eval overlay. Legal, healthcare, PE-backed platforms, manufacturers with sensitive IP.
Includes. Everything in Standard, plus: 8 source systems total, 2M to 10M pages indexed, 2 reranking models, passage plus redaction-aware citation grounding, privilege-aware access control with a regulatory overlay, a 300-question eval suite, and 3 weeks of Adopt with 8 training sessions. Available cloud, in your VPC, or on-prem.
Run · Monthly
After delivery, Run covers continuous indexing of new documents, retrieval and citation quality tuning, hallucination and low-confidence monitoring, exception handling, and same-day fixes for production issues. Cloud infrastructure costs are passed through at cost.
- Starter: $1,500/mo cloud
- Standard: $2,000/mo cloud · $3,000/mo on-prem
- Enterprise: $4,000/mo cloud · $5,000/mo on-prem
Running Search alongside other Databender products? The Platform Bundle lowers setup and Run across your whole stack, and a longer Run term drops the monthly fee further.
Common add-ons
Tight tiers are the headline. Add-on prices are transparent so you know the cost upfront, no negotiation.
| Add-on | Price |
|---|---|
| Additional source system (cloud) | $5,000 |
| Additional source system (on-prem) | $6,500 |
| Additional 500K pages indexed | $3,000 |
| Additional re-ranking model | $5,000 |
| Privilege-aware retrieval + ethical walls (adds to Standard) | $6,000 |
| Regulatory overlay (HIPAA, SOC 2) | $10,000 |
| Air-gapped deployment package | $15,000 |
There's a usage layer on top of the base price for large query volumes (over 10,000 queries per user per month). Most mid-market deployments don't hit it. We tell you in the Audit if your usage pattern will.
The Audit confirms which tier fits, the final scope, and the deployment shape.
Common Questions
FAQ
Your team will look for a document this afternoon.
Imagine asking in plain English and getting the answer in seconds, with the source cited.