Most "AI assistants" bolted onto a SaaS product are a chat window wrapped around a generic model with no real access to your account — confident-sounding, and just as often wrong. SAGE is built the other way: every answer comes from an agentic loop that actually calls Tubetific's own live modules — your channel data, your rank-tracker status, your niche's real signals — and then cites, per answer, exactly which real module backed each fact. If the data isn't there, SAGE says so instead of filling the gap with something plausible-sounding.
Quick answer
SAGE is Tubetific's AI research assistant, powered by Claude (Anthropic). It runs an agentic loop of real tool calls against a creator's own channel and niche data, answers with evidence links back to the exact module behind each fact, and builds a daily overnight brain note — included on Plus (Q&A only) and Pro (Q&A plus the brain note).
SAGE isn't a chatbot with a system prompt telling it to "act like a YouTube expert." It's an agentic loop, powered by Claude (Anthropic), with 13 registered tool functions it can call against Tubetific's real backend — things like pulling your channel's current data, checking your rank-tracker status for a specific keyword, reading your niche's live signal deltas, or listing which of your modules have fresh data to read. When you ask a question, SAGE decides which of those real tools it needs, calls them, and only then writes an answer grounded in what came back — not from Claude's general training knowledge about YouTube. Every answer also carries evidence links: real deep-links back to the specific module or route that supplied each cited fact, computed server-side from which tools SAGE actually called for that turn, not written after the fact to look credible. And SAGE tells you plainly when an answer came from its answer cache versus a fresh analysis, rather than presenting a reused answer as if it just ran the check again.
Agentic Q&A: a real tool-calling loop against 13 registered functions covering your channel, niche signals, rank tracker, keyword opportunities, and recent module activity — not a static script
Evidence links on every answer: computed server-side from which tools were actually called, linking back to the exact module or route behind each cited fact
Honest cache disclosure: SAGE marks each answer as freshly analyzed or reused from its answer cache, rather than blurring the difference
Overnight brain note: a proactive daily summary built from your real channel, niche, and RPM data, cached 24 hours in Supabase so SAGE doesn't rebuild your context from scratch on every page load
The same honesty rule as every other Tubetific signal: if real data isn't available to answer confidently, SAGE says so instead of returning a fabricated or generic-sounding answer
Not available on Free at all — Q&A starts on Plus, the brain note is Pro-only
No fixed menu of pre-written prompts — SAGE's chat panel takes a normal question about your channel, a niche, or a specific decision you're weighing.
Rather than answering from general knowledge, the agentic loop picks from its 13 registered tool functions — channel data, niche signals, rank-tracker status, keyword opportunities, recent module activity, and more — based on what your specific question actually requires.
Each tool call hits Tubetific's real backend for your account — the same underlying data the dashboards themselves show, not a separate, simplified copy built just for SAGE.
Every cited fact links to the real module or route that supplied it, computed server-side from the tools actually called for that turn — so you can go check the underlying signal yourself rather than taking the answer on faith.
Once a day, SAGE proactively summarizes what changed across your real channel, niche, and RPM data while you were away, and caches that summary for 24 hours — available to Pro plan holders as a standing briefing rather than something you have to remember to ask for.
A creator asks SAGE whether a specific video idea is worth pursuing
The distinguishing part isn't that SAGE gives an opinion — it's that the opinion is traceable back to specific, real tool calls a creator can independently verify on the linked module pages, rather than a confident-sounding paragraph with nothing real behind it.
| Capability | SAGE | A generic AI chat widget bolted onto a dashboard |
|---|---|---|
| Answers grounded in your real account data | Yes — 13 real tool calls against your channel/niche/rank data | Usually no — answers from the model's general training knowledge alone |
| Shows which real data backed each answer | Yes — evidence links computed from the tools actually called that turn | Rarely — answers presented with no traceable source |
| Discloses cached vs. freshly-run answers | Yes — every turn marked from_cache or fresh, honestly | Not applicable — most don't cache or disclose reuse at all |
| Proactive daily summary, not just reactive Q&A | Yes — the overnight brain note, Pro plan | Rarely — most are Q&A-only, nothing runs without being asked |
| Says "I don't have that data" instead of guessing | Yes — the same honesty rule applied to every other live signal | Inconsistent — many chat assistants fill gaps with plausible-sounding text |
Channel data, niche signals, rank-tracker status, problem context, workflow state, accountability context, SAGE alerts, user signals, signal deltas, keyword opportunities, recent module activity, and module listing/reading — a real, registered set the agentic loop chooses from per question, not a single generic lookup.
Computed server-side from which tools were actually called for that specific turn, linking back to the real module or route behind each cited fact — never authored by Claude after the fact to sound credible.
Every turn is marked whether it came from SAGE's answer cache or a fresh analysis, so a creator always knows whether the underlying numbers behind an answer were just recalculated.
A proactive daily summary of what changed across your real channel, niche, and RPM data, built without being asked and cached 24 hours in Supabase.
If real data isn't available to answer a question confidently, SAGE says so directly instead of returning a fabricated or generic-sounding answer — consistent with how Outlier Detection, Content Gap Analysis, and every other live module handle missing data.
Wrong
Assuming SAGE is a generic YouTube-advice chatbot like the AI features bundled into other tools.
Right
SAGE's answers come from an agentic loop that calls 13 real tool functions against your own channel/niche data first — it isn't answering from general training knowledge about YouTube the way a bolted-on chat widget typically does.
Wrong
Treating every SAGE answer as a brand-new, freshly-run analysis.
Right
SAGE marks each answer as fresh or reused from its answer cache — check that flag if it matters whether the specific numbers behind an answer were just recalculated or came from an earlier turn.
Wrong
Expecting SAGE to be available on the Free plan.
Right
SAGE has no Free-tier access at all. Q&A starts on Plus (15 questions/day); the proactive overnight brain note and a higher 50/day Q&A cap are Pro-only.
Wrong
Assuming the overnight brain note updates continuously through the day.
Right
It's built once daily and cached for 24 hours in Supabase — a standing daily briefing, not a live-refreshing feed. Ask SAGE a direct question in Q&A mode for anything more current than that day's note.
Is SAGE included on the Free plan?
No — Free has no SAGE access at all. SAGE Q&A becomes available starting on Plus ($19/mo, 15 questions/day); the proactive overnight brain note and a higher 50/day Q&A cap are Pro-specific ($49/mo).
Is SAGE just Claude with a YouTube-themed prompt?
No — the distinguishing part is the agentic tool-calling loop underneath. SAGE has 13 registered tool functions it can call against Tubetific's real backend for your account before writing an answer, and every answer carries evidence links back to whichever of those tools it actually used. A generic chat wrapper with no real account access wouldn't have either of those things.
How do I know if a SAGE answer is fresh or reused?
Every answer is marked whether it came from SAGE's answer cache (from_cache) or was freshly analyzed on that turn — this is disclosed honestly rather than presenting a reused answer as if it just ran the check again.
What are the evidence links SAGE shows under an answer?
Real deep-links back to the specific Tubetific module or route that supplied each cited fact — for example, a claim about a niche's RPM tier links to RPM Intelligence, and a claim about a keyword's gap score links to Content Gap Analysis. They're computed server-side from which tools SAGE actually called for that turn, not written afterward to look convincing.
Does SAGE ever make up an answer when it lacks real data?
No — if real data isn't available to answer confidently, SAGE says so rather than returning a fabricated or generic-sounding answer, the same honesty rule applied across every other Tubetific signal.
What's the difference between the brain note and Q&A mode?
The brain note is proactive — built once a day automatically, without being asked, and cached 24 hours — and is Pro-only. Q&A mode is reactive: you ask a specific question and SAGE runs its real tool-calling loop to answer it, available on both Plus and Pro.
Can SAGE see data outside my own account?
SAGE's tool functions are scoped to the real modules a creator has access to — channel data, niche signals, rank tracking, and similar — the same data already visible in the rest of the dashboard, not a separate broader dataset.
Is SAGE the same thing as the MCP connection in the app?
No — those are different, separately-built pieces. SAGE's tool-calling loop runs against Tubetific's own internal modules directly. The MCP integration (in the app under Connect) is a separate mechanism for connecting outside AI tools to Tubetific data; it isn't part of how SAGE itself answers a question.
Free to start. Real YouTube data from day one.
Launch Creator Hub