Ravensight

Huginn, the AI analyst

Huginn, the AI analyst, lives under Understand > Huginn. It picks its own read-only tools, runs them against the game you have selected, and answers with the numbers it used. A completed question is one shared AI unit, including every tool call it made on the way.

The tools it reaches for

get_overview, query_players, query_levels, get_retention, get_history, get_metrics, get_monetization, get_ad_revenue, get_campaigns, get_ltv_forecast, get_progression, get_economy, get_health, get_distribution, get_level_map, get_retention_split, get_data_quality, explain_metric_change, list_builds, compare_builds, get_funnel, get_property_breakdown and search_feedback cover the numbers, what players spent and what players wrote. get_code_brief reads the source scan's notes, list_playtest_jobs, list_playtest_findings, get_playtest_report and get_validation_report read playtests, list_feature_specs, get_feature_spec and get_feature_job read Build a feature, list_game_plans and get_game_plan read Create a game, and list_touchpoints and get_sales read marketing memory and the sales ledger, and the list_pulse_*, get_pulse_post and search_pulse_comments tools read Pulse. Huginn can call describe_studio_data to check what data the studio holds (see Studio data) before it chooses a tool. On deployments with journey graphs, get_journeys, find_bottlenecks and get_top_paths read those. When nothing else answers, it writes SQL with run_sql. Every tool is read-only, and the game it reads is the one you selected, never an argument the model can change.

Analyst context you write yourself

Findings are extracted by a model. Analyst context is the opposite: it is what you write, and the analyst is told to trust it over its own reading of the data. There are two levels, both inserted into the prompt verbatim and ahead of the findings. Game context lives in Settings > Huginn and holds up to 2,000 characters of facts about one game. Studio notes hold up to 1,000 characters of facts that are true across every game in the studio. A note like "levels 5 through 7 are an intentional difficulty spike" or "we ship weekly builds on Fridays" heads off a wrong conclusion before the analyst reaches for a tool. Writing either needs a member role or above; reading them needs nothing extra.

The same question across every game

Four questions can be asked of the whole studio rather than one game: the headline metrics, what players spent, a funnel, and an event property's breakdown. The analyst has them as compare_metrics, compare_monetization, compare_funnel and compare_property, and the API exposes them under /api/v1/studio/analytics/.... This is how a problem in one game is told apart from a habit of the studio: if the tutorial loses the same share of players in all three of your games, the tutorial design is the finding. Studio totals are sums and conversions are weighted by volume. There is deliberately no studio median, because the median of medians is not the median. Studio revenue is summed per currency and never across two, and there is no studio revenue per paying player.

A studio answer covers at most 12 games. When more were eligible the response says so rather than presenting twelve games as the studio: studio.truncated is true and studio.omittedGames counts the games left out. A property breakdown also carries studio.other, the count of events whose value is not in the merged top list.