Decision Referee
Decision support in Slack for teams making real company tradeoffs. Initial setup takes about 2–3 minutes, each decision takes about 30 seconds to 5 minutes depending on depth, and the analysis can use workspace-private firm context enriched from company website content and Brønnøysund Register Centre data where available. Built to be more grounded than a one-off generic AI answer, without pretending to replace judgement.
Not every decision needs the same level of scrutiny. All three commands run inside Slack and sit on top of the same workspace-private firm profile. The difference is how much structure, explanation, and challenge the team wants before making the call.
Every analysis ends with a signal. The point is not false certainty. It is a clear directional recommendation based on the decision input, the stored firm context, and the risks highlighted in the review.
GO
The decision appears reasonably aligned with current goals and manageable risk. Proceed, but with normal caution.
PAUSE
Moderate concerns or missing information. Consider addressing the flagged issues before committing.
STOP
Meaningful red flags were found. Review the analysis carefully and reconsider or restructure the decision.
Every analysis follows the same core structure. The deeper commands ask for more input, but the output format stays consistent so teams can scan it quickly in Slack.
Decision signal
Go, Pause, or Stop. A clear directional recommendation based on the analysis.
Key consideration(s)
The main factors that should shape the decision before the team commits.
Main risk(s)
The most important risks to watch, even if the overall signal is positive.
Hidden assumptions
The assumptions that could change the answer if they turn out to be wrong.
Deep onlyRecommendation
The clearest recommended direction based on the available information.
Next step
A concrete action the team should take next so the decision leads to clear follow-through.
Confidence
A stated confidence level in the recommendation based on the information available.
Decision Referee can use a workspace-private firm profile so the analysis is less generic than a one-off prompt. Setup usually takes about 2–3 minutes, starts in Slack with /firm_profile "Org. number" "domain/website", and can be expanded over time as the team adds more context.
Strategic fields
Team-maintained priorities, goals, constraints, and risk posture
Slack-managed profile flow
Context is created and edited in Slack rather than hidden in a separate admin tool
Company website enrichment
Company website content can be used to add business context automatically
Brønnøysund enrichment
For Norwegian companies, Brønnøysund Register Centre data can add basic company context automatically
Reusable context
Future analyses can reuse the firm context instead of starting from zero, improving relevance over time
Current, not permanent
The team can revise the context as strategy, constraints, or focus change
Less generic analysis
The aim is a better-grounded review, not a claim of complete company understanding
Decision support only
The context layer helps framing, but final responsibility stays with the team
Professional boundaries
It is not legal, financial, compliance, or other specialist advice
The point of the firm profile is simple: give the analysis more business reality than a generic prompt can provide. The profile is private to each workspace. Decision Referee can enrich it automatically with Brønnøysund Register Centre data and company website content, and that context helps improve decision relevance over time.
The core profile can include goals, constraints, strategy, time horizon, and risk tolerance. These fields are optional but useful, and they are maintained by the team in Slack so they can change as the business changes.
The first firm profile can be created in Slack with /firm_profile "Org. number" "domain/website". Setup usually takes about 2–3 minutes. That gives the product a starting point before any optional strategic fields are added.
Decision Referee can enrich the workspace profile with company website content and, for Norwegian companies, Brønnøysund Register Centre data. That helps fill in basic context and improve output quality, but it does not replace internal knowledge from the team.
Every decision can be followed up. Use /outcome to log what actually happened so the team builds a real record instead of relying on memory or hindsight.
01
Run analysis
Use any command to get a structured assessment of your decision.
02
Make the call
Decide whether to proceed, adjust, or abandon based on the analysis.
03
Log outcome
When you know the result, use /outcome to record what happened.
04
Learn from patterns
Over time, review which signals the team tends to follow, override, or ignore.
Every week, the team gets a summary of decision activity: how many decisions were logged, what signals were given, which outcomes were recorded, and where patterns may be worth reviewing.
8 / 12
Outcomes logged
12
Decisions made
4
Stop signals
2
Bias flags
Every analysis is stored the moment it runs. The input, the output, the signal, and the outcome if you logged one. It does not vanish into a Slack thread from three weeks ago.
Past decisions remain available as context for the team. That makes it easier to review similar tradeoffs over time instead of rediscovering the same history in scattered threads.
You start to see which kinds of decisions the team gets right and where it tends to override the signal. That track record is useful for calibration, especially as teams grow.
One place for what was decided, why, and what happened. New hires can read the history. Quarterly reviews have a foundation to work from.
The analysis is only as sharp as what you feed it. Vague input produces vague recommendations. Ten extra seconds of specificity in the description field changes the quality of everything that comes back.
Vague
Title: New market
Description: Should we expand?
Goal: Growth
Specific
Title: Enter Swedish B2B SaaS market Q2 2026
Description: Three inbound leads from Stockholm in the last month. Product already supports English. Swedish localization is roughly two weeks of work. No local presence yet, considering remote-first GTM with a partner.
Goal: Validate Nordic expansion thesis before committing to a local hire
Vague deep dive
Key risks: Might not work
Assumptions: That it will work
Stakeholders: Team
Specific deep dive
Key risks: No brand recognition in Sweden, unclear data processing regulations, 50K NOK validation budget
Assumptions: Swedish B2B procurement cycles are similar to Norwegian ones. English-first product is acceptable.
Stakeholders: CTO (localization scope), Head of Sales (pipeline capacity), Board (expansion budget sign-off)