# How our research works | Envisioning

Source: https://www.envisioning.com/services/how

---

# How our research works.

The same five research steps support four ways of working with us.

We have run the method since 2010, first by hand and now with 10+ frontier models in parallel. You bring the question. We scan, compare what the models agree on, check the sources, and hand back findings your team can filter, share and rerun.

[Back to services](/services)

## What we do

You bring the question. We hand back a source-checked signal set and a radar built around it.

We scan across 10+ frontier models, compare where they overlap and where they diverge, check every signal against its source, and structure the findings so you can filter them. A Signals Session delivers that as one radar. Signals lets your team rerun it whenever the question changes.

## The method

The same five steps sit behind a session, a scan you run yourself, and a custom build.

Step 1

Frame

We define the strategic question, decision context, categories, and timing so the research is relevant from the start.

Step 2

Scan

We run the same question through 10+ frontier models at once, so the first pass covers more ground than one analyst or one model can.

Step 3

Converge

Where several models agree, we look closer. A signal stays in only when its sources back it up.

Step 4

Verify

Signals are checked against their sources and organized so you can filter and compare them.

Step 5

Deliver

You get something your team can use the same week, in planning and in workshops.

## What you get back

Results your team can reuse without us.

Every signal carries its sources and its categories, so the radar can be filtered, rerun and shared without asking us. The case studies under client impact show what that looked like for each client.

A signal set with a source attached to every entry.

An interactive radar your team can filter.

Findings stored as data, so a question asked in March can be rerun in September.

One view the whole team works from.

We are so pleased with our radar! It will be vital tool for us to connect the deep tech ecosystem and train 1 million people in deep technology

Isabel Vincent

Strategic Education Manager

![EIT](/api/client-logos/5a5da620-f221-4d34-91ea-b9b748535c49/svg?v=1c6r5er)

## Four ways to work with us

Three services and the platform underneath them. What changes is who runs the work and how much of it you own afterwards.

Fastest entry point

Signals Session

Run it as a one-off engagement with our analysts when you need a good answer fast.

[/services/research-sessions](/services/research-sessions)

Recurring capability

Signals

The platform underneath all three services, run by your own team. Your first scan is free.

[/signals](/signals)

Custom system

Bespoke Projects

We build the method into a tool of your own, with its own data model and access controls, when Signals off the shelf is not enough.

[/services/bespoke-projects](/services/bespoke-projects)

Capability in the team

Build Sessions

The other half of the work. Your team builds with AI for a half day and keeps the habit, rather than reading about what is coming.

[/services/build-sessions](/services/build-sessions)

## Start with the question

If the method makes sense, the next step is choosing the right entry point for your team.

A Signals Session is the fastest way to begin: one question, €7,500 per session, and a year of Signals Pro included. If you already know you need the platform or a system of your own, go straight to Signals or Bespoke Projects.

[Find the right service](/pricing/request)[Book a session](/pricing/request?service=research-session)
