001 / how it works >>>
From 90 minutes
to institutional memory.
One workshop captures how you think. One chat interface lets your team ask it anything. One weekly habit keeps it accurate. Here is exactly what happens at each stage, what you and your team see, and which part of the product does the work.
Capture the brain
A structured workshop extracts SOPs, decision principles, do-not-say rules, and tone-of-voice examples into queryable knowledge.
Ask at mission speed
Staff ask in the Decisionlore chat. Answers cite the source and escalate to you when the brain is uncertain.
Refine every week
Review what the bot told your team. Mark good, bad, or correct it. The brain compounds while competitors stay static.
Scale across the firm
Department-scoped brains, training modules, audit log, and billing, built for the realities of a 10-50 pax SME.
002 / capture the brain >>>
One workshop, structured so nothing gets missed.
90 minutes + about a week to upload documents
A structured workshop walks you through how you actually decide things: pricing exceptions, client-fit calls, escalation rules, the phrases you'd never let a staff member say to a client. Starter plans run this as a self-serve, in-app wizard. Growth includes one facilitated workshop a year; Pro includes two, where someone runs the session with you instead of you filling in a form alone.
Alongside the workshop, you upload what you already have: SOPs, manuals, policy docs, past client emails that show how a decision actually went. Each document is parsed, chunked, and embedded so it becomes searchable knowledge, not just a file sitting in a folder.
Boss principles: the rules the bot uses to answer in your voice, ordered by priority. The highest-priority rule wins when two principles conflict.
Tone examples: 10 to 15 real question-and-answer pairs showing how you actually respond, used as few-shot examples so answers sound like you, not a generic assistant.
Knowledge base: every SOP, manual, and policy document you upload, parsed, chunked, and embedded so it's queryable by anyone on the team.
003 / ask at mission speed >>>
Answers that cite their source and know when to stop.
seconds per answer, on demand
Staff ask questions the same way they'd ask a colleague, in the Decisionlore chat. The bot answers in your voice, using the principles and documents captured in the workshop.
Before an answer ever reaches a staff member, it runs the same five-point check a compliance officer would: cited to a source document, checked against your principles, screened by your do-not-say rules, escalated to you when confidence is low, and logged to an immutable audit trail. Nothing ships unchecked.
Ask the brain: the core chat surface where staff ask anything about how things work here, and the bot answers in the boss's voice.
Pinned answers: for high-stakes, recurring questions (refund policy, escalation chain, after-hours protocol), you can pin the exact canonical answer so there's zero variance.
Escalation: when the brain isn't confident, it says so and routes the question to you instead of guessing.
- cited to a source document0.99
- checked against boss principles0.97
- screened by do-not-say rules0.96
- escalated when confidence is low0.94
- logged to the immutable audit trail0.93
004 / refine every week >>>
A short weekly habit, not a redo of the workshop.
a few minutes a week
Every answer the bot gave your team that week lands in the review queue. You read through them and mark each one good, bad, or correct it outright. This is the whole mechanism that keeps the brain from going stale: a document dump answers the same way forever, this compounds.
Corrections don't just fix one answer. They sharpen the underlying principles and tone examples the bot draws from, so the same mistake doesn't keep recurring. It's the difference between a static FAQ and a system that's actually still learning from you.
Review queue: recent assistant answers awaiting your review. Marking them teaches the bot to sound more like you.
Compounding: corrections feed back into principles and tone examples, so the fix generalizes instead of patching one question.
Cadence: a short weekly pass, not a rebuild. Most of the heavy lifting happened in the capture workshop.
005 / scale across the firm >>>
Built for the realities of a 10-50 pax SME.
The core brain (chat, uploads, principles, and the review queue) is available on every plan. As the team grows past a handful of people, Growth and Pro plans add the surfaces below to keep the brain organized and the owner out of the weeds.
Departments
growth+Group your team so Sales asks the Sales brain and Ops asks the Ops brain, without noise between them. Useful once you're past about 15 team members.
Courses
growth+Paste an SOP or policy and the AI turns it into bite-size lessons with a quiz, grounded in exactly what you wrote. Assign it to staff in under a minute.
Training board
growth+See who's competent and who needs help, at a glance. Click through to any learner's graded answers.
Team
Invite your team to the workspace so everyone who needs it can query the brain directly.
Audit log
growth+Every meaningful action (logins, uploads, edits, chat) logged append-only and retained for 12 months, for PDPA and compliance visibility.
My training
growth+The staff-side view: bite-size courses assigned by a manager. Learn it, prove it, move on.
006 / faq >>>
Common questions about how it works.
How long does onboarding (setup) take?
The capture workshop itself is 90 minutes, self-serve on Starter or facilitated on Growth and Pro, plus about a week to upload and process your documents. Once that's done, staff can start asking questions immediately.
How does this reduce new hire time to productivity?
New hires ask the brain directly instead of waiting for a manager to be free, so the questions that used to sit in a Slack thread for a day get answered in seconds, cited to the actual SOP. How much that speeds things up depends on how completely you've captured your SOPs and decision principles in the workshop, not on the software alone.
Can I use this for better generative AI decision making across my team, not just onboarding?
Yes, that's the core use case beyond onboarding: any staff member facing a judgment call (a pricing exception, a client-fit decision) can ask the brain instead of guessing or asking a generic AI tool that doesn't know your rules. Every answer runs a five-point check, cited to a source document, checked against your principles, screened by your do-not-say rules, escalated when confidence is low, and logged, before it reaches anyone.
What happens if the brain doesn't know the answer?
It says so and escalates the question to you instead of guessing. That's a deliberate design choice: a wrong answer stated confidently is worse than no answer.
Run the workshop.
Keep the result.
14-day pilot. 90 minutes of your time. We run the workshop, and you keep the result, even if you don't subscribe.
to capture the brain
per answer
review cadence
built for