01Pilot method
Give your writer a promise they can support.
Find claims that need evidence before they reach a campaign. Give the copywriter a specific promise to verify or revise before you buy traffic.
AI research for marketing and product teams
Find the gaps.Before you commit.
Give your next launch a better starting point.
Check the promises in your copy and explore questions a buyer might ask. Decide what to fix, show or back up before you brief the writer, book the shoot or publish the page.
Research / Simulation / Creation / Validation / LearningPilot methods + suite roadmap
Research with a job to do
Give your team
a brief they can act on.
A question about fit can become a shot to film. A mismatched citation can become a correction before publication. Here are three ways to put our pilot studies to work.
Give your crew a product demonstration to film.
Illustrative action / simulation pilotThe starting point
An adventure-led ad for a riding accessory. The product guide asks buyers to check mounting measurements before ordering.
Proposed action
Open with the documented measurement check. Ask the supplier to confirm compatibility beyond those measurements before making a fit claim.
Why it helps
Put a specific shot in the production brief. Give the crew a fit demonstration to film and the writer a buying question to answer.
Example brief for a fictional product. AI comments informed this proposed revision; customer preference is untested.
What the pilot found
In a saved OASIS run, two AI participants saw a fit-first opening and an adventure opening. Both commented on the fit-first opening: one asked for compatibility details; the other wanted performance evidence. These comments suggest questions to investigate with customers. They do not establish which ad would perform better.
The Counterfield suite
One API.
Ten research areas.
Start with a claim check or an ad brief. Our roadmap extends the same study API to content, product decisions and strategy. Each area shows its current availability.
02Small simulation pilot
Give the next shoot a clearer brief.
Explore possible objections with AI participants. Turn useful questions into product demonstrations, explanations or prompts for customer interviews.
03Connection planned
Give the reader an answer worth finding.
Turn reviewed product facts and site evidence into a brief the writer can use. Develop a useful article with an editor responsible for approval and publication.
04Planned workflow
Find what the concept still needs to prove.
Explore questions about a proposed feature, package or price. Use them to choose what to ask customers before committing to a build.
05Planned workflow
Find a signal worth investigating.
Bring dated sources to a market question. Examine what changed and what remains an assumption before committing more research to a trend.
06Planned workflow
See what would change your decision.
Explore a price change, new competitor or other stated condition. Identify the assumptions your strategy depends on and which ones to test first.
07Pilot methods
Keep the wrong facts out of your copy.
Spot citations for another product and claims a source does not establish. Give the editor specific corrections to make before publication.
08Outcome records available
Give the next test a starting point.
Keep reported results beside the study so the next team can see what was tried and observed.Automated learning and prediction are planned.
09Planned workflow
Find questions your onboarding needs to answer.
Rehearse a journey with AI participants to explore where explanations may be missing. Take those questions into observed customer testing before changing the flow.
10Planned workflow
Make the announcement easier to understand.
Explore how a proposed message could be interpreted by a defined audience. Identify terms to explain and context to add before release.
We run pilot methods in an operator-managed test environment. We need to build and validate the planned workflows before offering them to customers.
Planned extensions
Make the next decision
with what you already know.
The next phase connects studies to company knowledge and the tools where your team works. These extensions are planned.
Keep useful research in use.
Put past findings and product facts within reach of the next brief. Study history is available now; search across company sources is planned.
Give customers a reason to stay in your product.
Let customers check an idea inside your SaaS or client service. Build the workflow around the shared study API and your own interface.
Revisit the brief when the facts change.
A new product specification or customer question could trigger a fresh review. The connectors and trigger controls are planned.
The shared study system
Set up a study.
Review it with your team.
Keep your question and sources in the same record as the findings. Attach results from a customer test to compare them with your starting assumptions.
- 1 / 3
Define the question.
Supply the sources and alternatives. Set the call and cost limits before the run.
- 2 / 3
Choose a method.
Run a pilot method and inspect its findings beside the sources. Review model disagreements before making a decision.
- 3 / 3
Record the outcome.
Add measured results after the test. Include the version and time window, and check attribution before drawing a conclusion.
Built to be embedded
Put research where
your team already works.
Add a claim check or a simulation to your existing software. Your team keeps its workflow; your developers use one study format across research methods.
AdScope and Orrery are our first planned product connections. Agree on access and permitted data with an operator before starting a trial.
One study contractAPI v0.1 / staging
- POST
- /v1/studiesSave the question and sources.
- POST
- /v1/studies/:id/runsStart a method within the study's limits.
- GET
- /v1/runs/:idRead the result, run status and cost.
- POST
- /v1/studies/:id/outcomesStore the results you report.
Tenant isolation · Source hashes · Idempotent requests · Run limits · Cancellation
Before you plan a study
Check the scope
before you start.
Choose a method you can use and the result you need. Agree sources and cost limits with your team.
Pilot access
You can run hook reviews, citation checks and a two-participant simulation in our test environment. An operator must arrange access.
Research areas
We are building a research engine for teams and software products. AdScope and Orrery are our first integration targets; the suite covers ten research areas.
Prediction limits
Treat simulated reactions as hypotheses. Measure demand, clicks or conversions with customer research and observed tests. We need to test forecasts against real outcomes before offering them.
Content production
Orrery supports briefs and a writing-review flow. We need to connect Counterfield to that flow before offering article production through this API. A person approves the draft before the site owner publishes it.
Conflicting reviews
Inspect the cited source and each reviewer's judgment. Our pilot checks withhold a recommendation after a required source or order check fails. Check the source's accuracy before using the claim.
Reported outcomes
Attach measured results to the study and content or product version. Use a controlled test before claiming your change improved results. We plan learning methods that use observed outcomes.
Counterfield
Bring the brief
you need to improve.
Choose the decision you need to make and prepare a study brief.
This preview does not submit an enquiry or start a study.