The AI engine behind everyrelationship.
Outer Insights isn't a CRM with AI layered on top. Intelligence runs through the foundation of the platform itself, shaping how every part of it works.
Every match, message, and insight across the platform is generated, grounded, and governed by ARMS Core. Governance itself runs through BASE, the layer that checks every response before it is delivered, working quietly across your data so your team can move faster with more confidence.
Confidence isn't the same as correct.
Most AI tools generate an answer and leave you to verify it. BASE, the governance layer inside ARMS, verifies first, checking every response against your actual data before it reaches you and catching ungrounded claims before they're mistaken for fact.
Oversight also scales to the decision. A routine summary and a time-sensitive stakeholder communication aren't governed the same way, because they don't carry the same risk.
"Meridian typically invests at a 4.2x revenue multiple with a standard 15% option pool requirement."
Governed LLM Response Layer
Every output checked against your data before it reaches you
Real-Time Prompt Refinement
Suggested improvements as you type, not after you submit
Hallucination & Fabrication Detection
Flags ungrounded claims, figures, and sources
Context-Adaptive Oversight
Scrutiny calibrated to the stakes of each request
Chat With Your Portal Data
Ask questions of any uploaded document, record, or conversation
Explainable, Source-Grounded Answers
Every response traceable to the data behind it
Traditional databases return a list. ARMS returns a ranking, and shows you why.
Underneath every match is a Graph Attention Network that treats your relationships as a connected graph rather than a flat spreadsheet, reading mandate, geography, sector, stage, and round history as interdependent signals rather than static filters.
That graph is paired with high-dimensional vector search across our investor index, and closed end to end by agents that draft outreach grounded in the same matched data.
Led 4 comparable Series B rounds · 2 warm paths · mandate and cheque size both matched
Report-Grounded Matching
Verified mandate, sector, geography, round and history feed every match
Graph Neural Network (GAT) Matching
Models investor-founder fit as a relationship graph, not a filtered list
Vector-Based Investor Search
Semantic matching across millions of investor and company profiles
Dynamic Relationship Scoring
Every match ranked and re-scored as new data comes in
Continuous Model Retraining
Recommendations improve with every interaction
Retrieval-Augmented Generation (RAG) Outreach
Auto-drafted messages grounded in real matched investor data
Deep context on every investor, company, and individual.
Before a first conversation, know exactly who you're talking to. ARMS Core pulls mandate, portfolio history, deal activity, and relationship signals into a single readable profile, built automatically and refreshed as new data comes in.
No more piecing a picture together across ten browser tabs and outdated PDFs. One report, fully sourced and ready before your next meeting.
Your always-on analyst, inside every workflow.
The assistant lives inside the platform itself, not in a separate window. It acts with full awareness of the module you're already working in. No re-explaining your context, no copying data back and forth.
Every conversation, captured and understood.
Calls are transcribed in real time, speakers identified, and a structured summary of decisions and action items attached automatically to the relationship record.
D. ELLISCan you walk us through the option pool assumption?
S. WALTERSWe've modelled 10% post-close and can share the schedule.
Ask a question. See the answer.
A multi-agent system interprets your question, selects the appropriate visualisation, and builds it live from your data, not a static dashboard you have to interpret yourself.
The same intent-first approach powers search: ask in plain language and ARMS searches semantically across CRM records, documents, communications, and reports.
Explainable by design. Auditable by default.
Every capability on this page runs through the same governance layer. ARMS Core is built on patented innovations in intelligent system governance, with logging, feedback loops, and human-in-the-loop review designed in from the start, not bolted on after something went wrong.
Every output can be traced back to the specific data behind it, and every model is continuously retrained on real feedback rather than left to go stale.
Full Audit Logging
A reviewable trail for every AI action taken
Human-in-the-Loop Review
Critical decisions stay reviewable, not autonomous by default
Transparency-by-Design
Governance built into the system, not layered on top
Continuous Safety Monitoring
Ongoing oversight as models learn and evolve
See ARMS Core in action.
Every match, every message, every insight, powered by governed AI built for real relationships.
Request a Demo