How the product has improved, in real numbers.

July to October 2026

• Boards built

  105×

  Businesses with a live growth board that month.

  Jul

  2

  Aug

  118

  Sep

  141

  Oct

  209

• Plays written

  507×

  Ranked moves written for those boards.

  Jul

  21

  Aug

  4,603

  Sep

  8,855

  Oct

  10,638

• Live market lookups

  2×

  Searches, rankings and site reads used as evidence. Started in September.

  Jul

  —

  Aug

  —

  Sep

  21,216

  Oct

  40,248

• Quality checks run

  8×

  Each play and found post graded before it reaches a board. Started in September.

  Jul

  —

  Aug

  —

  Sep

  5,807

  Oct

  45,270

• Verified market facts

  13×

  Dated facts about competitors, prices and buyers, saved and reused across plans.

  Jul

  —

  Aug

  —

  Sep

  184

  Oct

  2,346

• Relevant finds kept

  +19.8 pts

  Share of found posts, podcasts and newsletters that passed the relevance check.

  Jul

  —

  Aug

  —

  Sep

  40.2%

  Oct

  60.0%

A play then and now

• EVIDENCE

  August: Written from the business description alone.

  Now: Backed by live searches, rankings and dated market facts.

• YOUR SITE

  August: Could suggest pages you already had.

  Now: Reads your live site and sitemap first and skips existing pages.

• CHECKS

  August: Every generated play went straight to the board.

  Now: Each play is graded; generic or off-market ones are dropped.

• DUPLICATES

  August: Refills could repeat earlier plays.

  Now: Near-copies of live or skipped plays are removed automatically.

What every play now carries

• EVIDENCE

  The live finding that triggered the play, such as a ranking gap or a competitor change, with the date it was checked.

• BUYER INTENT

  Who the move reaches and why they are ready to act, in the words buyers use.

• SITE CHECK

  Confirmation that your site does not already cover it, read from your live pages and sitemap.

• ACTION

  Concrete steps plus a ready-to-review draft, worked in Sidekick or by your own AI assistant.

Where the live lookups go

• Search

  Google rankings, search gaps and competitor positions.

• Your site

  Live pages and sitemap, so plays never repeat what exists.

• Communities

  Buyer conversations on Reddit, LinkedIn and X.

• Podcasts and newsletters

  Shows and lists that reach the same buyers.

• AI search

  How the business appears in answers from tools like ChatGPT and Perplexity.

What gets thrown away

In October, 40% of found posts, podcasts and newsletters were rejected before reaching a board: low buying intent, wrong market or unrelated to the business. In September, 60% were rejected. Rejecting noise is part of the quality, not a loss of it.

What changed

1. July 2026

   First prototype boards built for a handful of businesses.

2. August 2026

   Launch. Plans shaped to the type of business — agency, local service, software or trade — so plays fit how each one sells.

3. September 2026

   Live market lookups added, and every play and found post graded before saving. Weak, generic or off-market items are dropped instead of shown.

4. October 2026

   Plans read your live site and sitemap first, reuse dated market facts, skip pages you already have, and remove near-duplicate plays automatically.

5. October 2026

   Autopilot launched: your own AI assistant (ChatGPT, Claude, Cursor and others) can read the board and work plays through MCP, instead of you copying each step by hand.

How the engine changed

• JUL

  Engine: A reasoning model writes plans; a fast model answers questions in Sidekick.

  Why it mattered: Structured, ranked moves instead of open-ended chat advice.

• AUG

  Engine: Plans shaped by business type, plus a separate fast model for drafts and a safety check.

  Why it mattered: Advice fits how each kind of business actually wins customers.

• SEP

  Engine: Live search, ranking and community lookups feed the planner; a separate judge grades every play and find.

  Why it mattered: Plays cite real evidence, and weak or unrelated items are dropped before you see them.

• OCT

  Engine: Newer, faster drafting model; site and sitemap read first; dated market facts reused; Autopilot opens the board to your own AI.

  Why it mattered: No repeated pages or near-copies, better drafts, and plays your assistant can work for you.

• NEXT

  Engine: Results from finished plays, such as search ranking changes, feed back into new plans.

  Why it mattered: Boards learn from what actually moved the numbers for that business.

How we measure

Counts are totals across all customer boards for each calendar month. Relevance is the share of found items that pass our check for real buyer intent and a match with the business. A dash means the check did not exist yet. A lower keep rate early on meant more noise was being found, not that more was being shown. No customer names or details are shared.