You bring one strategic question. We run our multi-model research workflow on it and come back with an interactive radar and a structured set of findings, usually within 48 hours. The work behind each radar (sourcing signals, vetting them, summarizing, categorizing) is what takes an analyst weeks. A year of the Signals Starter plan comes with it, so the research does not stop when the session ends.
Step 1
Brief
You tell us the question, the sector context, and when the decision has to be made.
Step 2
Scan
We run prompts across several frontier models in parallel to map what is out there.
Step 3
Converge
Where models agree, a signal is strong. Where they disagree, something new may be happening. Both are useful.
Step 4
Verify
Every signal is source-checked and scored for reliability before it reaches synthesis.
Step 5
Deliver
You get a radar you can share and findings your team can act on the same week.
Scanning, classification, and synthesis are done by people, at a research institute that has been doing foresight work since 2010.
A standard bundle of scans, aimed at your priority question rather than spread thin across a whole sector.
Reading weak signals is judgment work. We have done it across a lot of sectors, and the connections between them are usually where the value sits.
Typical turnaround is same day to 48 hours from brief to delivered radar.
The engagement bundles a full year of the Signals Starter plan, so your team keeps the account and can keep working with the research.
We handle setup, prompting, and curation. You show up for the brief and the readout.
Running AI research well takes practice. If you need an answer before you need a capability, this is the version where we absorb that.
A shared radar puts the whole room on one baseline. Most strategy disagreements turn out to be disagreements about facts.
Multi-model convergence with verification, rather than one model's opinion. The same approach behind the published case studies on this site.
The findings arrive as a product, not a transcript. It survives being forwarded to a team that was not in the room.
Multi-Model Convergence Method
See how we use model agreement for signal strength and model disagreement for novelty detection.
Read the method
How services work together
Compare the Signals Session, the self-serve Signals platform, and Bespoke delivery models before choosing your next step.
Compare service models