Where interpretation
becomes evidence.
Build a shared analytical framework, apply it to any body of evidence, and produce results that are verifiable — not just claimed. Human-led. AI-assisted.
No credit card required · Free during Open Beta
Source data
Title
Social media use and adolescent mental health outcomes
Author
Smith, J. et al.
Year
2024
Coding form
Sentiment *
Topic *
AI Suggestion
Negative
Grounded in
▸ Strauss & Corbin (1998)
▸ 6 similar coded rows
Codebook · v4
Used across research disciplines
From raw evidence to defensible findings.
No setup wizard. No implementation consultant.
01
Upload your evidence
Spreadsheets, PDFs, audio, video — bring the full body of material your analysis will rest on. Lenselot handles all three source types in a single project.
02
Build your interpretive framework
Define what you are looking for in terms your whole team can apply. Build a codebook from scratch, import an existing one, or let the AI suggest a starting structure from your data. You decide what stays.
03
Apply, verify, and publish
Code every unit. Measure inter-rater agreement automatically. Produce findings backed by a methodology that is versioned, transparent, and citable — not just described in a footnote.
Your framework, under version control.
Every change to your codebook is tracked, attributed, and reversible. Propose revisions, review diffs, merge with team sign-off. Your analytical framework has a history — and that history is part of your methodology.
- Full version history with per-change attribution
- Proposal and review workflow before merging changes
- Diff view shows exactly what changed and why
- Restore any previous version at any time
Version history
Added 'Resilience' code
Sarah K.
Merged proposal #12 — 3 codes revised
James R.
Refined 'Anxiety' definition
Priya M.
Initial framework
Sarah K.
Change · Anxiety — definition
- “Expressed worry or fear”
+ “Reported symptoms including worry, fear, or avoidance behaviours”
Built for teams. Designed for disagreement.
Multiple coders. Configurable overlap. Automatic IRR. When disagreement surfaces — which it should — adjudication resolves it with a complete audit trail, not a silent override.
- Concurrent multi-coder projects with row-level assignment
- Cohen's κ and percent agreement computed per dimension
- Calibration rounds to align the team before production coding
- Adjudication queue shows all coder responses side by side
IRR Overview · Sentiment
κ = 0.84
Percent agreement
91%
Coders
Sarah K.
142 rows
94%
James R.
142 rows
91%
Priya M.
142 rows
88%
3 rows need adjudication
Review →Agentic AI that works from your data — not thin air.
Our grounded AI agents read your uploaded literature, cross-references previously coded rows, and proposes codes with explicit reasoning. Every suggestion cites its source. You accept, modify, or reject — always.
- Suggestions grounded in your uploaded literature
- Cross-references similar rows already coded by the team
- Accept, modify, or reject — no silent auto-application
- AI attribution tracked separately in the audit log
AI Assistant · Row 12 of 47
Evidence
“The intervention helped participants recognise triggers before escalation, reducing crisis episodes over 12 weeks.”
Suggested code
Psychoeducation
Teaching individuals to understand and manage their own mental health symptoms.
Grounded in
The analytical rigour your work deserves.
Built for teams who need their methodology to hold up — under peer review, under editorial scrutiny, under replication.
Dataset, document & multimedia
CSV, XLSX, PDF, audio, and video all work in the same project workflow.
Reusable codebooks
Build a codebook once and apply it across multiple projects. Four code types, each with a description shown at coding time.
Focused coding interface
Each row as a vertical card. Source fields stacked above the coding form. Navigate by keyboard or jump directly to any row.
Multi-coder workspaces
Invite collaborators, configure row overlap for reliability testing, and manage roles — all within one workspace.
Inter-rater reliability
Cohen's κ and percent agreement per dimension, computed automatically. Export IRR statistics for reporting or publication.
Export at any time
Download as .xlsx or .csv at any stage. All original columns, one column per code, plus row status. No lock-in.
Also included
AI codebook generation
Let AI suggest an initial framework from your data. You review every suggestion — nothing is applied automatically.
Evidence search
Semantic search over PDF corpora. Accepted passages are highlighted and tracked per row.
Calibration rounds
All coders label the same rows before the main run to align on the rubric.
Adjudication
Designated reviewer sees all coders' answers side by side and selects the authoritative code.
Wherever structured judgment meets unstructured evidence.
Any team that applies a framework to a body of material — repeatedly, verifiably, with others.
Academic Research
Systematic reviews, grounded theory, thematic analysis. Produce methodology sections that hold up to peer review.
UX Research
Interview transcripts, usability sessions, survey verbatims — structured into defensible findings teams can act on.
Policy Analysis
Government documents, public consultations, legislative records. Track how frameworks evolve across policy cycles.
Investigative Journalism
Collaborative coding of source materials, leaked documents, and structured datasets — with a full audit trail.
Clinical Research
Patient records, adverse event categorisation, trial documentation. Rigorous coding with IRR built in.
Market Insights
Focus groups, brand monitoring, competitor materials. Build a reusable framework that compounds across studies.
Simple pricing.
Free during Open Beta. Paid plans introduced after beta — early users notified in advance.
Free
during Open Beta
Everything included:
- Unlimited projects
- Dataset, document & multimedia
- Reusable codebooks
- Multi-coder workspaces
- IRR & calibration
- Export as Excel / CSV
- AI code suggestions
Paid plans will be introduced after the Open Beta. Early users will be notified in advance.
Frequently Asked Questions
Can't find an answer? Contact us for support.