A strategist
that reads your notes.
The Analyst is the chat surface over your Strategic Context. Ask any positioning question; it queries your audits, signals, decisions, battle cards, and win/loss patterns, and answers with quoted evidence and tool traces.
Not a generic LLM with a system prompt. An agent grounded in your workspace, forbidden from hallucinating, required to cite the evidence.
Pro workspaces. The Analyst reads what's in Strategic Context; it never browses the internet on your behalf.
“We can never find our own answers fast enough.”
The answer is somewhere. An audit you ran six months ago noted it. A rep wrote it up in a win/loss note. A decision doc covered it at an offsite. But nobody remembers where — and by the time someone finds it, the meeting has ended.
Most SaaS ‘AI assistants’ are thin wrappers over a general model with a system prompt. They don’t know your audits, your pillars, your losses, your competitor set. They’re confidently wrong.
The Analyst is built the other way. It’s an agent with structured tools into your Strategic Context. When it doesn’t know, it says so. When it does, it shows you where.
Agent, not autocomplete.
Six properties that separate the Analyst from a ChatGPT tab with your pillar doc pasted in.
Grounded in your context
Every answer is traced to memory entries, audit snapshots, or signals in your workspace. Anonymous data never enters the prompt; other tenants are never reachable.
Tool use, visible
The Analyst has nine structured tools (search_context, list_audits, list_signals, list_battle_cards, get_positioning_brief, and more). Every call is logged under the answer so you can see what it actually did.
No-hallucination stance
The system prompt forbids invented answers. When evidence is absent, the Analyst says ‘I don’t have that in your context yet’ rather than guessing.
Specificity over politeness
No ‘great question, let me walk you through an approach.’ Direct answers with citations, or a clear no. Calibrated for strategic work, not content generation.
Writes back when asked
Ask it to draft a brief update, a battle-card patch, or a new memory entry, and it’ll stage the write with a diff — you approve before it commits.
Reasoning-quality model
Backed by Claude Sonnet 4.6 with tool-use mode; falls back to DeepSeek for cost-sensitive work. Quality comes first; cost is managed, not dictated.
Strategic question, cited answer.
Why are we losing mid-market deals to Competitor A in the last quarter?
Based on the last 30 days of Strategic Context, three signals line up:
(1) Win/loss pattern (3 reviews tagged competitor=a, segment=mid-market): governance trust is the tipped factor in all three. Buyers said your row-level-security story was ‘aspirational’ while theirs felt concrete.
(2) Competitor Signals (14 days ago): Competitor A shipped a governance whitepaper and updated their /vs/acme page with a governance-focused comparison. Severity: high.
(3) Positioning Audit (last run): governance lens scored 54/100 — the second-lowest of your eight lenses. Primary finding was that your /security page doesn’t exist; governance lives only as a sub-point on /pricing.
Recommended action: publish /vs/competitor-a/governance (scoped in Win/Loss Review #12) and re-run the audit. The pattern should break within 60 days.
- → list_reviews(competitor=a, segment=mid-market)
- → list_signals(competitor=a, severity=high)
- → get_audit(latest=true)
- → search_context(query='governance')
Sample. Every claim in the response maps to a real memory entry, signal, or audit finding in your workspace.
Short answers.
No. Your workspace data is never used to train the underlying model. Queries go to Anthropic/DeepSeek under our accounts; results stay inside your tenant.
No, deliberately. The Analyst is scoped to your Strategic Context. For ‘what did the competitor just ship,’ use Competitor Signals; for ‘what did Perplexity say about us,’ use AI Visibility.
Large enough that we rarely hit it. The Analyst uses tool calls to search your context rather than stuffing everything into one prompt. The search is fast and the response is grounded.
Claude Sonnet 4.6 for reasoning; DeepSeek V3 as a cost-sensitive fallback. Both are configurable per workspace.
Related capabilities.
Strategic Context
The substrate. Everything the Analyst knows lives there. Without it, the Analyst has nothing to say.
See Strategic Context →Positioning Brief
The Analyst writes the Brief. Ask a strategic question in chat, regenerate the Brief, forward it to the team — same evidence, same voice.
See Positioning Brief →Positioning Audit
Every audit enriches what the Analyst can reason over. The more audits, signals, and reviews in context, the more specific the answers.
Run a free audit →Your context. Your Analyst. Your evidence trail.
The free Positioning Audit writes the first memory entries into Strategic Context. Once a workspace accumulates audits, signals, and reviews, the Analyst becomes the fastest way to answer ‘why are we seeing this?’