Now accepting early access applications

AI-native survey
programming in minutes.

Upload your survey spec or write it with prompts. Review and refine with built-in AI assist. Ship in hours, high accuracy, without the grind.

Request early access Watch Demo
thefieldbase.com/builder/pid-00031
Fieldbase
← Go back
SURVEY STRUCTURE
68 q · 1196 lines
40
41- qid: Intro
42 type: info
43 title: Welcome
44 text: |
45 Thank you for participating in this research
46 about streaming video services.
47 The survey takes about 12 minutes.
48
49- qid: S1
50 type: single
51 text: In which country do you currently live?
52 instruction: <i>(Single response)</i>
53 required: True
54 order: fixed
55 skin: radio
56 options:
57 1: United States
58 2: United Kingdom
59 3: Canada
60 4: Australia
61
62# Punch hMarket immediately after S1 — carries the
63# country label for piping and quota.
64- script: |
65 if S1.value is not None:
66 hMarket.set(S1.value)
67
68- qid: hMarket
69 type: single
70 visibility: hide
71 title: Market (from S1) — used for piping
S1 · single · active · Step 2/37
Preview is current
S1 single
In which country do you currently live?
(Single response)
United States ▾
68 questions Hover a gap for + Ask AI here · click a question to scope it · Ctrl+K to edit
60%
Workload reduction on average per study
5x
Faster from spec to first preview
0
Syntax errors — validated at generation
Start your survey

Upload a spec.
Get a working survey.

Drop the Word doc or paste the questionnaire. Fieldbase reads the question text, options, programming notes, and routing instructions — and writes the survey code for you to review.

Start your survey
Brand Tracker Q3 2026 · pid-00031 — choose how to begin before you code.
Your data is not used for AI training. Uploads are processed securely; temporary processing data is deleted within 7 days.
Program with AI
Upload the questionnaire — AI writes the survey code, then lints it.
Drop a .docx, .pdf, or .txt file
Or click to browse
Your data is not used for AI training. Uploads are processed securely; temporary processing data is deleted within 7 days.
Brand Tracker Q3 2026 is ready
45 questions · 3 lists · opened in Builder
SURVEY STRUCTURE
  • S1single
  • {}script
  • hMarketsingle
  • S2single
  • {}script
  • S3single
  • Q1multi
  • Q2multi
  • Q3gridSingle
  • Q4cbc
45 questions
survey.yaml 
- qid: S1
  type: single
  text: In which country do you currently live?
  required: True
  options:
    1: United States
    2: United Kingdom
    3: Canada
    4: Australia
Open the Builder to review, run, or edit any question with inline AI.
AI assist in the editor

Describe the change. The AI writes it.
You accept, refine, or skip.

Scope with @, describe what you want. The AI proposes a diff — lint-checked and in your DSL — with Accept, Refine, or Reject buttons. Every change is logged in a thread the whole team can trace.

thefieldbase.com/builder/pid-00031
Fieldbase
survey.yaml 68 q · 1196 lines
72- qid: S1
73 type: single
74 text: In which country do you currently live?
75 required: True
76 options:
77 1: United States
78 2: United Kingdom
79 3: Canada
80 4: Australia
81
82# Market quota check — fires on S1 selection
83- script: |
84 if hMarket.value is not None:
85 cell = hMarket.value
86 if not quota("qtMarket", cell).isOpen:
87 endSurvey("quotafull")
88
89- qid: S2
90 type: number
91 text: What is your age?
92 min: 13
93 max: 99
94 required: True
95
96# Age gate — out-of-scope respondents terminated
97- script: |
98 if S2.value >= 65:
- if S2.value >= 65:
+ if S2.value < 18 or S2.value >= 65:
99 endSurvey("terminate")
100
101- qid: Q1
102 type: multi
103 text: Which of these brands have you used in the past 3 months?
104 min: 2
105 options: brandList
106
107- qid: Q2
108 type: gridSingle
109 text: How satisfied are you with each of these features?
110 rows: brandList
68 questions Everyone on this project sees replies · every change is logged
Results & export

See your results instantly.
Export clean data.

Auto-generated topline reports with cross-tabs and filters. Deterministic column naming across every export — no cleanup, no surprises between waves.

Fieldbase
Auto Topline
QAgenumber
QGendersingle
QBrandsUsedmulti
QMainBrandsingle
QSatisfactiongridSingle
QNPSnps
QRankrank
QSpendnumber
QFeedbackopen
Cuts Total · n=10,847 Female · n=4,831 18–34 · n=3,210 ⬡ Share cut
Filter QGender = Female✕ + Add filter Save as cut…
Brand awareness — past 3 months
QBrandsUsed · multi · n=10,847
Dell
64%
6,942
Apple
58%
6,291
HP
38%
4,122
Lenovo
31%
3,363
Asus
21%
2,278
Satisfaction — mean scores
QSatisfaction · gridSingle · scale 1–5
TotalDellAppleHPLenovo
Quality3.83.44.53.23.6
Value3.23.62.13.83.7
Ease of use3.93.54.63.33.4
Availability3.74.23.83.92.8
Likelihood to recommend — NPS
QNPS · nps · n=10,847
+42 NPS
18%
22%
60%
Detractors Passives Promoters
Fieldbase
Data Export
Brand Tracker Q1 2026
10,847 respondents · Export data
SPSS
.sav
CSV
.csv
Excel
.xlsx
ASCII
.dat
Export shape
Standard
One row per respondent. Flat data.
Long / Modeling
One row per task. For conjoint.
Column preview
ColumnSourceType
QBrandsUsed_1Dell0/1
QBrandsUsed_2HP0/1
QBrandsUsed_3Lenovo0/1
QMainBrandMost used brandint
QSatisfr1Quality1–5
QSatisfr2Value1–5
QSatisfr3Ease of use1–5
QSatisfr4Availability1–5
QNPSNPS score0–10
QFeedbackOpen textstr
↓ Export .sav
42 columns · 10,847 rows
What you get

Built for how modern
survey work actually happens.

Each capability is designed around the way survey programmers actually work today — fast, code-first, AI-augmented, with the research methods and output quality serious projects need.

Code-first authoring

Clean, readable syntax for every question type and every routing rule. Autocomplete and live validation from a single contract — every rule defined once, enforced everywhere.

single · multi · grid · cbc · maxdiff · cardSort

Conjoint and MaxDiff included

CBC with three response modes — choice, rating, ranking. Upload a design file, reference it in a single line. Utility-ready exports drop straight into your analysis tool.

design: project://laptop_cbc_v2

Quotas that hold

Cell locking on every completion. No over-quota, no soft counts, no manual reconciliation — even under heavy concurrent traffic. The numbers you set are the numbers you get.

quota("AgeGender", 1).isFull()

Live respondent preview

Every change updates the preview as you type. Branching, loops, piped values, hidden logic — see exactly what a respondent sees, without deploying. Test routing in seconds.

Analysis-ready exports

Deterministic column naming — every export column is mechanically derived from the question structure. No surprises, no manual mapping, no cleanup before analysis.

QBudgetr1c1 · QBudgetr1c2 · QBudgetr2c1

Theme framework

22 professional themes with seven style variants. Every visual property is a token — change how a survey looks without touching what it does. Respondent-facing UI that holds up to client review.

Explore the platform

Every capability, in depth.

The platform's real surface — question types, scripting, design assets, simulation — each with its own page you can read, share, or send to a stakeholder.

Question types & skins

15+ types — single, multi, grids, cardSort, CBC, MaxDiff, lookup, rank. 60 skin variants for different respondent contexts.

Explore question types →
Scripting & routing

Python-like - script blocks. Conditional routing, quota gates, computed hidden values, screenouts — real logic.

Explore scripting →
Choice modeling

Upload CBC / MaxDiff design files, reference in one line, and verify precision before fielding with Design Health + Precision scorecards.

Explore choice modeling →
Survey simulator

Fake respondents against the real quota engine. Catch routing gaps, load-test quotas, verify CBC and MaxDiff coverage.

Explore the simulator →
Author in any language

Prompt. Author.
In your language.

Prompt the AI in the language your team thinks in — French, Japanese, Spanish, German, Portuguese. The AI responds in-language; the generated content stays in that language too. Twelve languages supported today, more as teams ask.

Code stays portable. Structural keywords (qid, type, options) and identifiers stay English, so any reviewer on your team can read the YAML regardless of the content language.
Your data, your terms

Built for the way research data has to be handled.

Your data isn't used to train AI

Survey content, respondent answers, and the code you author are never added to AI training data. Your work — and your clients' data — stays your own.

ISO 27001 certified, GDPR aligned

Certified under ISO 27001 and aligned with GDPR through our parent company, Inginit. Data is encrypted in transit and at rest, with tenant isolation by project. Full details on the Inginit trust center.

You own and control your data

Export the full project — code, design files, responses — at any time. Delete on request, including from backups within standard retention windows. Any operational metrics we keep for product improvement are anonymised aggregates that can't be traced back to you.

Fieldbase is built by Inginit. For certifications, sub-processors, retention windows, and to request a DPA, see the Inginit trust center.

Now accepting early access

Write prompts to code.

The AI-native survey platform. Describe what to program. Fieldbase writes production-ready survey code.

Request early access →
⬡ Inline assist ⌘K

Program Q1 to Q5 with skip logic and piping. Read [PN] notes for routing, randomisation, exclusive codes, and terminates.

↵ Send Esc Cancel
→ program Q3 as gridSingle, read the PN for scale and rows
→ add screenout script if code 99 at Q1, terminate
→ pipe Q2 selected label into Q3 text, add probe if low rated
Any language

Your team's language.
Your team's surveys.

Author content and prompt the AI in any language. Keywords and identifiers stay English so exports drop into SPSS, Stata, or Excel without renaming.

EN English 
- qid: QBrandsUsed
  type: multi
  text: |
    Which brands have you
    used in the past 3 months?
  options:
    1: Dell
    2: HP
    99: None of these
FR Français 
- qid: QBrandsUsed
  type: multi
  text: |
    Quelles marques avez-vous
    utilisées au cours des
    3 derniers mois ?
  options:
    1: Dell
    2: HP
    99: Aucune de celles-ci
The AI prompt works in your language too:
FR Lis la section screener, structure et programme selon les specs
ES Lee la sección del screener, estructura y programa según los specs
DE Lies den Screener-Abschnitt, strukturiere und programmiere gemäß den Specs
The platform

A real IDE for surveys.

Question tree, code editor, live preview — three panels, always in sync.

survey.yaml
Screener
SQBrand
MQFeatures
ƒsQuota
Main
GQSatisf
◈QCbc
⊞QSort
- qid: QSatisfaction
  type: gridSingle
  text: |
    How satisfied with
    {{QMainBrand
    .selectedLabel}}?
  rows:
    1: Overall quality
    2: Value for money
    3: Ease of use
  scale:
    1: V. dissatisfied
    5: V. satisfied
How satisfied are you with Apple?
✓ Overall quality 4
Value for money —
Ease of use —
Availability —
← Back Next →
Start your survey

Upload a spec.
Get a working survey.

Drop a Word or PDF. The AI reads every question, code, and routing note.

Spec_v2.1.docx
Q1. Which brands have you used in the past 3 months?
[PN: MULTI. RANDOMIZE. EXCLUSIVE 99.]
1. Dell
2. HP
3. Apple
99. None of these
[IF 99, TERMINATE]
↓
survey.yaml 
- qid: QBrandsUsed  type: multi  order: random  options:    1: Dell    2: HP    3: Apple    99:      exclusive: True- script: |    if QBrandsUsed.selected.any(99):        endSurvey()
60%
workload
reduction
5×
spec to first
preview
0
syntax errors
at generation
Results & export

See results.
Export clean data.

Auto topline with filters. Deterministic columns across every wave.

Auto Topline
Brand awareness
Dell
64%
6,942
Apple
58%
6,291
HP
38%
4,122
Lenovo
31%
3,363
Data Export
SPSS CSV Excel ASCII
QBrandsUsed_1Dell0/1
QMainBrandMost usedint
QSatisfr1Quality1–5
QSatisfr2Value1–5
QFeedbackOpen textstr
What you get

Built for survey programmers.

✎
Code-first authoring
Clean syntax with autocomplete and live validation.
◈
Conjoint included
CBC and MaxDiff built in. One platform.
⬡
Quotas that hold
Cell locking. No over-quota under load.
▤
Clean exports
Deterministic columns. SPSS, CSV, Excel.
◐
22 themes
Professional respondent UI. Token-based.
ƒ
Script blocks
Python-like routing, piping, computed values.
Your data stays yours.
Not used for AI training. ISO 27001 + GDPR aligned via Inginit. Export anytime.
Early access

Get in early.

Fieldbase is in private beta. Email us with a few words about your team and the kind of research you run — we'll set up a call and get you onto the platform.

Email [email protected]

Replies within 48 hours · No mailing list, just a real conversation.