Sigblaze
Sign inJoin the waitlist

Research plans that inherit your company, ICP and competitors

State a goal and get a bounded plan of tasks and queries, run against the context you stored once. The sources and claims it produces feed insights, and approved insights become actions.

Join the waitlist See the workflow

In the product

workspace / fieldnote / context + planSynthetic data
Workspace context · Context v4
Company
Fieldnote (fictional)
Product
Client project hub for small agencies
ICP
Agencies with 10–50 staff running 8+ retainers
Competitors
Meridian, Tallyhouse
Research goal

How does Meridian position onboarding for agencies, and is the claim holding up?

Bounded plan · 4 tasks
T1Collect Meridian's own onboarding and pricing claims3 queries
T2Find agency-side accounts of Meridian onboarding4 queries
T3Check changelog for onboarding changes2 queries
T4Compare against Tallyhouse onboarding language3 queries
Context v4 is attached to this run, so the run can be re-read later.

The problem

Each new chat session starts with no context

A general AI chat forgets your company, your ICP and the competitors you already track. You re-explain them every time, and two runs of the same question can end up answering different things.

How it works

  1. Hold context in the workspace. Company, product, ICP, market and competitors are stored once and versioned. Each research run records which context version it used.
  2. State a goal. A goal is a question you need answered, such as how Meridian positions onboarding for agencies.
  3. Review a bounded plan. The goal becomes a set of tasks and queries that you can read before any retrieval runs.
  4. Retrieve into stored sources. Results are normalised and de-duplicated into sources that stay in the workspace after the run.
  5. Hand over to evidence. Stored sources move on to claim extraction, which produces the claims you can inspect in Evidence.

Evidence and provenance

  • Each run points to the context version it used.
  • Each task keeps the queries that produced its sources.
  • Each source keeps its retrieval date and origin.
  • The plan is stored with the run, so you can see what was and was not researched.

Capability status

  • Workspace contextCompany, product, ICP, market and competitor context that every research run inherits.
    Private alpha
  • Research goals and plansA goal becomes a bounded set of tasks and queries instead of one giant prompt.
    Private alpha
  • Live retrieval and sourcesRetrieved pages are normalised and de-duplicated into stored sources with provenance metadata.
    Private alpha

Research is the deliberate step where you ask a question. Its sources feed Evidence, and Monitor keeps the answer current afterwards.

FAQ

Research: common questions

Can I edit the plan before it runs?

The plan is shown as bounded tasks and queries so it can be reviewed. Exact editing controls follow the current private-alpha build.

What sources does it retrieve from?

It uses live retrieval through a provider adapter. The specific providers are not listed publicly yet.

How is Research different from Monitor?

Research answers one question when you ask it. Monitor re-checks the pages and queries you choose on a schedule and raises a signal when they change.

Next: Signals

Join the waitlist for the private alpha

Sigblaze is built for agencies, fractional GTM teams and B2B SaaS teams that research the same markets on a schedule. Pricing is not set.

Join the waitlist See how it works