Signal Brief.

The problem

The communications lead at a Canadian fintech starts each morning with low-grade dread, piecing together a media brief across four or five open tabs — Google News, Reddit, Twitter/X — in roughly 45 minutes. There is no single starting point, no severity framework beyond gut feel, and no way to tell an isolated customer complaint from an emerging journalist narrative.

Worse, the brief is descriptive, not actionable: it explains *what happened* but never *what to do about it*. By 10am it is already stale, and leadership follow-ups expose the gaps. As the persona puts it, confidence never peaks — residual doubt is the default even after the brief is sent.

The solution

SignalBrief is a daily AI media intelligence tool for Canadian fintech comms teams. Its core differentiator is the interpretation layer: every story is delivered as *what happened / why it matters / suggested action*, tagged by severity — Noise / Watch / Act / Crisis — on a consistent rubric rather than gut feel.

One trigger produces two outputs simultaneously: a full email brief grouped by severity, and a separate executive "so what" story short enough to forward to a CEO. It replaces the manual 45-minute ritual with a repeatable process that runs in under two minutes, and understands Canadian fintech regulatory context — OSFI, PIPEDA, Bank of Canada — and the local media hierarchy.

How it works

After a one-time company-profile setup, the app pulls live signals server-side from Google News RSS and Reddit public JSON, deduplicates and filters for relevance, then passes them to Claude Sonnet 4.6 via a single API call. The model filters, classifies severity with one-line reasoning, writes the "so what" layer per story, and assembles both outputs. All relevant context fits in Claude's 200K-token window, so no vector database is needed for the MVP.

Claude was chosen for strict template adherence, low hallucination (every claim must trace to a provided source), and long context. Prompt iteration was disciplined: a Canadian-context block lifted "why it matters" usefulness from 3.8 to 4.3, headline-only safeguards eliminated hallucination on thin inputs, and tier-specific tone guidance raised tone ratings to 4.6. Testing hit 87% severity-calibration agreement and a 0% hallucination rate.

Who it's for

SignalBrief is B2B, sold to communications and growth teams at digital-first Canadian fintechs of 50–500 employees — neobanks and financial apps. The persona is "Jane," a Director or Manager of Communications at a ~250-person fintech who can expense a $35/mo tool without CMO approval.

Jane isn't buying media-monitoring software; she's buying confidence, speed, executive trust, and protection from being blindsided. Winning looks like a brief ready in under five minutes, fewer leadership follow-up questions, and "so what" framing in every story.

Why it matters

The global AI media monitoring market is projected to nearly triple from ~$4B in 2025 to $9–11B by 2030, with the mid-market segment growing fastest as enterprise tools like Meltwater and Brandwatch ($15–30K/yr) stay out of reach. Post-SVB and post-crypto-contagion, brand trust is a board-level concern.

SignalBrief's wedge is a price-to-value gap: $35/mo replacing ~15 hours of manual work a month, at ~97% gross margin, with no vertical-specific AI brief tool for Canadian fintech today. It replaces the DIY ritual — Google Alerts plus manual search — not Meltwater, and in a tight community where word of mouth travels fast, that focus is the moat.

At a glance

Project

Signal Brief

Built by

Khushboo Sharma

One-liner

Signal Brief is a daily media intelligence tool for Canadian fintech communications teams.